![Robotic Systems – Applications, Control and Programming
18
World Health Organisation, “Fact sheet n.297,” Available from: http://www.who.int/
mediacen-ter/factsheets/ fs297, 2006.
Yim, M.; Shen, W., Salemi, B., Rus, D., Moll, M., Lipson, H., Klavins, E. & Chirikjian, G.
(2007). Modular self-reconfigurable robot systems [Grand Challenges of Robotics].
In: IEEE Rob. Autom Mag., 14(1), pp. 865–872.](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-28-320.jpg)
![2
Target Point Manipulation
Inside a Deformable Object
Jadav Das and Nilanjan Sarkar
Vanderbilt University, Nashville, TN
USA
1. Introduction
Target point manipulation inside a deformable object by a robotic system is necessary in
many medical and industrial applications such as breast biopsy, drug injection, suturing,
precise machining of deformable objects etc. However, this is a challenging problem because
of the difficulty of imposing the motion of the internal target point by a finite number of
actuation points located at the boundary of the deformable object. In addition, there exist
several other important manipulative operations that deal with deformable objects such as
whole body manipulation [1], shape changing [2], biomanipulation [3] and tumor
manipulation [4] that have practical applications. The main focus of this chapter is the target
point manipulation inside a deformable object. For instance, a positioning operation called
linking in the manufacturing of seamless garments [5] requires manipulation of internal
points of deformable objects. Mating of a flexible part in electric industry also results in the
positioning of mated points on the object. In many cases these points cannot be manipulated
directly since the points of interest in a mating part is inaccessible because of contact with a
mated part. Additionally, in medical field, many diagnostic and therapeutic procedures
require accurate needle targeting. In case of needle breast biopsy [4] and prostate cancer
brachytherapy [6], needles are used to access a designated area to remove a small amount of
tissue or to implant radio-active seed at the targeted area. The deformation causes the target
to move away from its original location. To clarify the situation we present a schematic of
needle insertion for breast biopsy procedure as shown in Figure 1. When tip of the needle
reaches the interface between two different types of tissue, its further insertion will push the
tissue, instead of piercing it, causing unwanted deformations. These deformations move the
target away from its original location as shown in Figure 1(b). In this case, we cannot
manipulate the targeted area directly because it is internal to the organ. It must be
manipulated by controlling some other points where forces can be applied as shown in
Figure 1(c). Therefore, in some cases one would need to move the positioned points to the
desired locations of these deformable objects (e.g., mating two deformable parts for sewing
seamlessly) while in other cases one may need to preserve the original target location (e.g.,
guiding the tumor to fall into the path of needle insertion). In either of these situations, the
ability of a robotic system to control the target of the deformable object becomes important,
which is the focus of this chapter.
To control the position of the internal target point inside a deformable object requires
appropriate contact locations on the surface of the object. Therefore, we address the issue of](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-29-320.jpg)
![Robotic Systems – Applications, Control and Programming
20
determining the optimal contact locations for manipulating a deformable object such that
the internal target point can be positioned to the desired location by three robotic fingers
using minimum applied forces. A position-based PI controller is developed to control the
motion of the robotic fingers such that the robotic fingers apply minimum force on the
surface of the object to position the internal target point to the desired location. However,
the controller for target position control is non-collocated since the internal target point is
not directly actuated by the robotic fingers. It is known in the literature that non-collocated
control of a deformable object is not passive, which may lead to instability [7]. In order to
protect the object and the robotic fingers from physical damage and also in order to
diminish the deterioration of performance caused by unwanted oscillation, it is
indispensable to build stable interaction between the robotic fingers and the object. Here we
consider that the plant (i.e., the deformable object) is passive and does not generate any
energy. So, in order to have stable interaction, it is essential that the controller for the robotic
fingers must be stable. Thus, we present a new passivity-based non-collocated controller for
the robotic fingers to ensure safe and accurate position control of the internal target point.
The passivity theory states that a system is passive if the energy flowing in exceeds the
energy flowing out. Creating a passive interface adds the required damping force to make
the output energy lower than the input energy. To this end we develop a passivity observer
(PO) and a passivity controller (PC) based on [8] for individual robotic finger where PO
monitors the net energy flow out of the system and PC will supply the necessary damping
force to make PO positive. Our approach extends the concept of PO and PC in [8] to multi-
point contacts with the deformable object.
Fig. 1. Schematics of needle breast biopsy procedure: (a) needle insertion, (b) target
movement, and (c) target manipulation
The remainder of this chapter is organized as follows: we discuss various issues and prior
research in Section 2. The problem description is stated in Section 3. Section 4 outlines the
mathematical modelling of the deformable object. A framework for optimal contact
locations is presented in Section 5. The control methods are discussed in Section 6. The
effectiveness of the derived control law is demonstrated by simulation in Section 7. Finally,
the contributions of this work and the future directions are discussed in Section 8.
2. Issues and prior research
A considerable amount of work on multiple robotic systems has been performed during the
last few decades [9-11, 12-15]. Mostly, the position and/or force control of multiple
manipulators handling a rigid object were studied in [9-11]. However, there were some
works on handling deformable object by multiple manipulators presented in [12-15]. Saha
(a) (b) (c)](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-30-320.jpg)
![Target Point Manipulation Inside a Deformable Object 21
and Isto [12] presented a motion planner for manipulating deformable linear objects using
two cooperating robotic arms to tie self-knots and knots around simple static objects. Zhang
et al. [13] presented a microrobotic system that is capable of picking up and releasing
operation of microobjects. Sun et al. [14] presented a cooperation task of controlling the
reference motion and the deformation when handling a deformable object by two
manipulators. In [15], Tavasoli et al. presented two-time scale control design for trajectory
tracking of two cooperating planar rigid robots moving a flexible beam. However, to the
best of our knowledge the works on manipulating an internal target point inside a
deformable object are rare [4, 5]. Mallapragada et al. [4] developed an external robotic
system to position the tumor in image-guided breast biopsy procedures. In their work, three
linear actuators manipulate the tissue phantom externally to position an embedded target
inline with the needle during insertion. In [5] Hirai et al. developed a robust control law for
manipulation of 2D deformable parts using tactile and vision feedback to control the motion
of the deformable object with respect to the position of selected reference points. These
works are very important to ours present application, but they did not address the optimal
locations of the contact points for effecting the desired motion.
A wide variety of modeling approaches have been presented in the literature dealing with
computer simulation of deformable objects [16]. These are mainly derived from physically-
based models to produce physically valid behaviors. Mass-spring models are one of the
most common forms of deformable objects. A general mass-spring model consists of a set of
point masses connected to its neighbors by massless springs. Mass-spring models have been
used extensively in facial animation [17], cloth motion [18] and surgical simulation [19].
Howard and Bekey [20] developed a generalized method to model an elastic object with the
connections of springs and dampers. Finite element models have been used in the computer
simulation to model facial tissue and predict surgical outcomes [21, 22]. However, the works
on controlling an internal point in a deformable object are not attempted.
In order to manipulate the target point to the desired location, we must know the
appropriate contact locations for effecting the desired motion. There can be an infinite
number of possible ways of choosing the contact location based on the object shapes and
task to be performed. Appropriate selection of the contact points is an important issue for
performing certain tasks. The determination of optimal contact points for rigid object was
extensively studied by many researchers with various stability criteria. Salisbury [23] and
Kerr [24] discussed that a stable grasp was achieved if and only if the grasp matrix is full
row rank. Abel et al. [25] modelled the contact interaction by point contact with Coulomb
friction and they stated that optimal grasp has minimum dependency on frictional forces.
Cutkosky [26] discussed that the size and shape of the object has less effect on the choice of
grasp than by the tasks to be performed after examining a variety of human grasps. Ferrari
et al. [27] defined grasp quality to minimize either the maximum value or sum of the finger
forces as optimality criteria. Garg and Dutta [28] shown that the internal forces required for
grasping deformable objects vary with size of object and finger contact angle. In [29],
Watanabe and Yoshikawa investigated optimal contact points on an arbitrary shaped object
in 3D using the concept of required external force set. Ding et al. proposed an algorithm for
computing form closure grasp on a 3D polyhedral object using local search strategy in [30].
In [31, 32], various concepts and methodologies of robot grasping of rigid objects were
reviewed. Cornella et al. [33] presented a mathematical approach to obtain optimal solution
of contact points using the dual theorem of nonlinear programming. Saut et al. [34]](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-31-320.jpg)
![Robotic Systems – Applications, Control and Programming
22
presented a method for solving the grasping force optimization problem of multi-fingered
dexterous hand by minimizing a cost function. All these works are based on grasp of rigid
objects.
There are also a few works based on deformable object grasping. Like Gopalakrishnan and
Goldberg [35] proposed a framework for grasping deformable parts in assembly lines based
on form closure properties for grasping rigid parts. Foresti and Pellegrino [36] described an
automatic way of handling deformable objects using vision technique. The vision system
worked along with a hierarchical self-organizing neural network to select proper grasping
points in 2D. Wakamatsu et al. [37] analyzed grasping of deformable objects and introduced
bounded force closure. However, position control of an internal target point in a deformable
object by multi-fingered gripper has not been attempted. In our work, we address the issue
of determining the optimal contact locations for manipulating a deformable object such that
the internal target point can be positioned to the desired location by three robotic fingers
using minimum applied forces.
The idea of passivity can be used to guarantee the stable interaction without exact
knowledge of model information. Anderson and Spong [38] published the first solid result
by passivation of the system using scattering theory. A passivity based impedance control
strategy for robotic grasping and manipulation was presented by Stramigioli et al. [39].
Recently, Hannaford and Ryu [40] proposed a time-domain passivity control based on the
energy consumption principle. The proposed algorithm did not require any knowledge
about the dynamics of the system. They presented a PO and a PC to ensure stability under a
wide variety of operating conditions. The PO can measure energy flow in and out of one or
more subsystems in real-time by confining their analysis to system with very fast sampling
rate. Meanwhile the PC, which is an adaptive dissipation element, absorbs exactly net
energy output measured by the PO at each time sample. In [41], a model independent
passivity-based approach to guarantee stability of a flexible manipulator with a non-
collocated sensor-actuator pair is presented. This technique uses an active damping element
to dissipate energy when the system becomes active. In our work we use the similar concept
of PO and PC to ensure stable interaction between the robotic fingers and the deformable
object. Our work also extends the concept of PO and PC for multi-point contact with the
deformable object.
3. Problem description
Consider a case in which multiple robotic fingers are manipulating a deformable object in a
2D plane to move an internal target point to a desired location. Before we discuss the design
of the control law, we present a result from [42] to determine the number of actuation points
required to position the target at an arbitrary location in a 2D plane. The following
definitions are given according to the convention in [42].
Manipulation points: are defined as the points that can be manipulated directly by robotic
fingers. In our case, the manipulation points are the points where the external robotic fingers
apply forces on the deformable object.
Positioned points: are defined as the points that should be positioned indirectly by
controlling manipulation points appropriately. In our case, the target is the position point.
The control law to be designed is non-collocated since the internal target point is not directly
actuated by the robotic fingers. The following result is useful in determining the number of
actuation points required to accurately position the target at the desired location.](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-32-320.jpg)
![Target Point Manipulation Inside a Deformable Object 23
Result [42]: The number of manipulated points must be greater than or equal to that of the
positioned points in order to realize any arbitrary displacement.
In our present case, we assume that the number of positioned points is one, since we are
trying to control the position of the target. Hence, ideally the number of contact points
would also be one. But practically, we assume that there are two constraints: (1) we do not
want to apply shear force on the deformable object to avoid the damage to the surface, and
(2) we can only apply control force directed into the deformable object. We cannot pull the
surface since the robotic fingers are not attached to the surface. Thus we need to control the
position of the target by applying only unidirectional compressive force.
In this context, there exists a theorem on the force direction closure in mechanics that helps
us determining the equivalent number of compressive forces that can replace one
unconstrained force in a 2D plane.
Theorem [43]: A set of wrenches w can generate force in any direction if and only if there
exists a three-tuple of wrenches 1 2 3
{ , , }
w w w whose respective force directions 1
f , 2
f , 3
f
satisfy:
i. Two of the three directions 1
f , 2
f , 3
f are independent
ii. A strictly positive combination of the three directions is zero.
1 2 3 0
+ + =
f f f
α β γ (1)
where α , β , and γ are constants. The ramification of this theorem for our problem is that
we need three control forces distributed around the object such that the end points of their
direction vectors draw a non-zero triangle that includes their common origin point. With
such an arrangement we can realize any arbitrary displacement of the target point. Thus the
problem can be stated as:
Problem statement: Given the number of actuation points, the initial target and its desired
locations, find appropriate contact locations and control action such that the target point is
positioned to its desired location by controlling the boundary points of the object with
minimum force.
4. Deformable object modelling
Consider a schematic in Figure 2 where three robotic fingers are positioning an internal
target (point A) in a deformable object to the desired location (point B). We assume that all
the end-effectors of the robotic fingers are in contact with the deformable object such that
they can apply only push on the object as needed.
The coordinate systems are defined as follows: w is the task coordinate system, o is the
object coordinate system, fixed on the object and i is the i-th robotic finger coordinate
system, fixed on the i-th end-effectors located at the grasping point. In order to formulate
the optimal contact locations, we model the deformable object using mass-spring-damper
systems. The point masses are located at the nodal points and a Voigt element [20] is
inserted between them. Figure 3 shows a single layer of the deformable object. Each element
is labeled as j
E for 1,2, ,
=
j NE where NE is total number of elements in a single layer.
Position vector of the i-th mesh point is defined as [ ]
=
p T
i i i
x y , 1,2,3,...,
=
i N where, N
is total number of point masses. k and c are the spring stiffness and the damping
coefficient, respectively. Assume that no moment exerts on each mesh point. Then, the
resultant force exerted on the mesh point, pi , can be calculated as](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-33-320.jpg)
![Target Point Manipulation Inside a Deformable Object 25
where, ( )
n r
i i is the unit inner normal of i-th contact and i
f denotes the i-th finger’s force.
We assume that the contact forces should exist in the friction cone to manipulate objects
without slip of the fingertip. Now we need to find three distinct points, 1 1
( )
r θ , 2 2
( )
r θ , and
3 3
( )
r θ , on the boundary of the object such that Equation (3) is satisfied. Here, 1
θ , 2
θ , and 3
θ
are the three contact point locations measured anti-clockwise with respect to the x axis as
shown in Figure 4. In addition, we assume that the normal forces have to be non-negative to
avoid separation and slippage at the contact points, i.e.,
0
≥
i
f , 1,2,3
=
i (4)
Fig. 4. Three fingers grasp of a planar object
A physically realizable grasping configuration can be achieved if the surface normals at
three contact points positively span the plane so that they do not all lie in the same half-
plane [44]. Therefore, a realizable grasp can be achieved if the pair-wise angles satisfy the
following constraints
min max
| |
≤ − ≤
i j
θ θ θ θ , low high
≤ ≤
i
θ θ θ , , 1,2,3
=
i j , ≠
i j (5)
A unique solution to realizable grasping may not always exist. Therefore, we develop an
optimization technique that minimizes the total force applied on to the object to obtain a
particular solution. The optimal locations of the contact points would be the solution of the
following optimization problem.
min f f
T
sunject to
3
1
( )
=
=
w n r
i i i
i
f
min max
≤ − ≤
i j
θ θ θ θ , , 1,2,3
=
i j , ≠
i j
0
≥
i
f , 1,2,3
=
i
0 360
≤ ≤
i
θ , 1,2,3
=
i
(6)](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-35-320.jpg)
![Robotic Systems – Applications, Control and Programming
26
Once we get the optimal contact locations, all three robotic fingers can be located at their
respective positions to effect the desired motion at those contact points.
6. Design of the controller
In this section, a control law for the robotic fingers is developed to guide a target from any
point A to an arbitrary point B within the deformable object as shown in Figure 2.
6.1 Target position control
At any given time-step, point A is the actual location of the target and point B is the desired
location of the target. n1, n2 and n3 are unit vectors which determine the direction of force
application of the actuation points with respect to the global reference frame w . Let
assume, pd is the position vector of point B and p is the position vector of point A.
Referring to Figure 2, the position vector of point A is given by
[ ]
=
p T
x y (7)
where, x and y are the position coordinates of point A in the global reference frame w . The
desired target position is represented by point B whose position vector is given by
[ ]
=
p T
d d d
x y (8)
where, d
x and d
y are the desired target position coordinates. The target position error, e, is
given by
= −
e p p
d (9)
Once the optimal contact locations are determined from Equation (6), the planner generates
the desired reference locations for these contact points by projecting the error vector
between the desired and the actual target locations in the direction of the applied forces,
which is given by
*
= ⋅ i
e e n
i (10)
where,
[ ]
=
i
n T
xi yi
n n (11)
All robotic fingers are controlled by their individual controllers using the following
proportional-integral (PI) control law
* *
= +
e e
i Pi i Ii i
f K K dt , 1,2,3
=
i (12)
where, Pi
K , and Ii
K are the proportional and integral gains. Note that in the control law
(12), mechanical properties of the deformable object are not required. Forces applied by the
fingers on the surface of the deformable object are calculated by projecting the error vector
in the direction of the applied forces. But the Equation (12) does not guarantee that the
system will be stable. Thus a passivity-based control approach based on energy monitoring
is developed to guarantee the stability of the system.](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-36-320.jpg)
![Target Point Manipulation Inside a Deformable Object 27
6.2 Passivity-based control
A passivity-based control approach based on energy monitoring is developed for
deformable object manipulation to guarantee passivity (and consequently stability) of the
system. The main reason to use passivity-based control is to ensure stability without the
need of having an accurate model of the deformable object. It is not possible to develop a
precise dynamic model of a deformable object due to complex nonlinear internal dynamics
as well as variation in geometry and mechanical properties. Thus passivity based control is
an ideal candidate to ensure stability since it is a model independent technique. The basic
idea is to use a PO to monitor the energy generated by the controller and to dissipate the
excess energy using a PC when the controller becomes active [41], without the need for
modeling the dynamics of the plant (deformable object).
Passivity Observer (PO)
We develop a network model with PO and PC similar to [41] as shown in Figure 5. The PO
monitors the net energy flow of the individual finger’s controller. When the energy becomes
negative, PC dissipates excess energy from the individual controller. Similar to [41] energy
is defined as the integral of the inner product between conjugate input and output, which
may or may not correspond to physical energy. Definition of passivity [41] states that the
energy applied to a passive network must be positive for all time. Figure 5 shows a network
representation of the energetic behavior of this control system. The block diagram in Figure
5 is partitioned into three elements: the trajectory generator, the controller and the plant.
Each controller corresponds to one finger. Since three robotic fingers are used for planar
manipulation, three individual controller transfer energy to the plant.
The connection between the controller and the plant is a physical interface at which
conjugate variables ( i
f , i
v ; where i
f is the force applied by i-th finger and i
v is the velocity
of i-th finger) define physical energy flow between controller and plant. The forces and
velocities are given by
T
f f f
1 2 3
[ ]
=
f (13)
T
v v v
1 2 3
[ ]
=
v (14)
The desired target velocity is obtained by differentiating (8) with respect to time and is
given by
T
d d d
x y
[ ]
=
p
(15)
where, d
x
and d
y
are the desired target velocities, respectively. The desired target velocity
along the direction of actuation of the i-th robotic finger is given by
di d i
v = ⋅
p n
(16)
The trajectory generator essentially computes the desired target velocity along the direction
of actuation of the robotic fingers. If the direction of actuation of the robotic fingers, i
n , and
desired target velocity, d
p
, are known with respect to a global reference frame then the
trajectory generator computes the desired target velocity along the direction of actuation of
the fingers using Equation (16).](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-37-320.jpg)
![Robotic Systems – Applications, Control and Programming
28
The connections between the trajectory generator and the controller, which traditionally
consist of a one-way command information flow, are modified by the addition of a virtual
feedback of the conjugate variable [41]. For the system shown in Figure 5, output of the
trajectory generator is the desired target velocity, di
v , along direction of i-th finger and
output of the controller is calculated from Equation (12).
Fig. 5. Network representation of the control system. 1i
α and 2i
α are the adjustable damping
elements at each port, i=1,2,3
For both connections, virtual feedback is the force applied by the robotic fingers. Integral of
the inner product between trajectory generator output ( di
v ) and its conjugate variable ( i
f )
defines “virtual input energy.” The virtual input energy is generated to give a command to
the controller, which transmits the input energy to the plant through the controller in the
form of “real output energy.” Real output energy is the physical energy that enters to the
plant (deformable object) at the point where the robotic finger is in contact with the object.
Therefore, the plant is a three-port system since three fingers manipulate the object. The
conjugate pair that represents the power flow is i
f , i
v (the force and the velocity of i-th
finger, respectively). The reason for defining virtual input energy is to transfer the source of
energy from the controllers to the trajectory generator. Thus the controllers can be
represented as two-ports which characterize energy exchange between the trajectory
generator and the plant. Note that the conjugate variables that define power flow are
discrete time values and so the analysis is confined to systems having a sampling rate
substantially faster than the system dynamics.
For regulating the target position during manipulation, 0
=
di
v . Hence the trajectory
generator is passive since it does not generate energy. However, for target tracking, 0
≠
di
v](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-38-320.jpg)
![Target Point Manipulation Inside a Deformable Object 29
and 0
≠
i
f . Therefore the trajectory generator is not passive because it has a velocity source
as a power source. It is shown that even if the system has an active term, stability is
guaranteed as long as the active term is not dependent on the system states [45]. Therefore,
passivity of the plant and controllers is sufficient to ensure system stability.
Here, we consider that the plant is passive. Now we design a PO for sufficiently small time-
step ΔT as:
0
( ) ( ( ) ( ) ( ) ( ))
=
= Δ −
k
i k i j di j i j i j
j
E t T f t v t f t v t (17)
where, ΔT is the sampling period and = ×Δ
j
t j T . In normal passive operation, ( )
i j
E t
should always be positive. In case when ( ) 0
<
i j
E t , the PO indicates that the i-th controller is
generating energy and going to be active. The sufficient condition to make the whole system
passive can be written as
0 0
( ) ( ) ( ) ( )
= =
Δ ≥ Δ
k k
i j di j i j i j
j j
T f t v t T f t v t , 0
∀ ≥
k
t , 1,2,3
=
i (18)
where k means the k-th step sampling time.
The active and passive port can be recognized by monitoring the conjugate signal pair of
each port in real time. A port is active if 0
<
fv that means energy is flowing out of the
network system and it is passive if 0
≥
fv , that means energy is flowing into the network
system. The input and output energy can be computed as [46]
1
1
1
( 1) ( ) ( ) if ( ) ( ) 0
( )
( 1) if ( ) ( ) 0
− + >
=
− ≤
T
T i i di i di
i T
i i di
E k f k v k f k v k
E k
E k f k v k
(19)
2
2
2
( 1) ( ) ( ) if ( ) ( ) 0
( )
( 1) if ( ) ( ) 0
− − <
=
− ≥
T
T i i di i di
i T
i i di
E k f k v k f k v k
E k
E k f k v k
(20)
1
1
1
( 1) ( ) ( ) if ( ) ( ) 0
( )
( 1) if ( ) ( ) 0
− − <
=
− ≥
P
P i i i i i
i P
i i i
E k f k v k f k v k
E k
E k f k v k
(21)
2
2
2
( 1) ( ) ( ) if ( ) ( ) 0
( )
( 1) if ( ) ( ) 0
− + >
=
− ≤
P
P i i i i i
i P
i i i
E k f k v k f k v k
E k
E k f k v k
(22)
where, 1 ( )
T
i
E k and 2 ( )
T
i
E k are the energy flowing in and out at the trajectory side of the
controller port, respectively, whereas 1 ( )
P
i
E k and 2 ( )
P
i
E k are the energy flowing in and out at
the plant side of the controller port, respectively. So the time domain passivity condition is
given by
1 1 2 2
( ) ( ) ( ) ( )
+ ≥ +
T P T P
i i i i
E k E k E k E k , 0
∀ ≥
k (23)
Net energy output of an individual controller is given by](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-39-320.jpg)
![Robotic Systems – Applications, Control and Programming
40
8. Conclusion and future work
In this chapter, an optimal contact formulation and a control action are presented in which a
deformable object is manipulated externally by three robotic fingers such that an internal
target point is positioned to the desired location. First, we formulated an optimization
technique to determine the contact locations around the periphery of the object so that the
target can be manipulated with minimum force applied on the object. The optimization
technique considers a model of the deformable object. However, it is difficult to build an
exact model of the deformable object in general due to nonlinear elasticity, friction,
parameter variations, and other uncertainties. Therefore, we considered a coarse model of
the deformable object to determine the optimal contact locations which is more realizable. A
time-domain passivity control scheme with adjustable dissipative elements has been
developed to guarantee the stability of the whole system. Extensive simulation results
validate the optimal contact formulation and stable interaction between the robotic fingers
and the object.
9. References
[1] D. Sun and Y. H. Liu, Modeling and impedance control of a two-manipulator system
handling a flexible beam, ASME Journal of Dynamic Systems, Measurement, and
Control, vol. 119, pp. 736-742, 1997.
[2] P. Dang, F. L. Lewis, K. Subbarao and H. Stephanou, Shape control of flexible structure
using potential field method, 17th IEEE International Conference on Control
Applications, Texas, USA, pp. 540-546, 2008.
[3] X. Liu, K. Kim, Y. Zhang and Y. Sun, Nanonewton force sensing and control in
microrobotic cell manipulation, The International Journal of Robotics Research, vol.
28, issue 8, pp. 1065-1076, 2009.
[4] V. G. Mallapragada, N. Sarkar and T. K. Podder, Robot-assisted real-time tumor
manipulation for breast biopsy, IEEE Transactions on Robotics, vol. 25, issue 2, pp.
316-324, 2009.
[5] S. Hirai and T. Wada, Indirect simultaneous positioning of deformable objects with multi
pinching fingers based on uncertain model, Robotica, Millennium Issue on
Grasping and Manipulation, vol. 18, pp. 3-11, 2000.
[6] S. Nath, Z. Chen, N. Yue, S. Trumpore and R. Peschel, Dosimetric effects of needle
divergence in prostate seed implant using 125I and 103Pd radioactive seeds,
Medical Physics, vol. 27, pp. 1058-1066, 2000.
[7] A. Albu-Schaffer, C. Ott and G. Hirzinger, Constructive energy shaping based
impedance control for a class of underactuated Euler-Lagrange systems, IEEE
International Conference on Robotics and Automation, pp. 1387-1393, 2005.
[8] B. Hannaford and R. Jee-Hwan, Time-domain passivity control of haptic interfaces, IEEE
Transactions on Robotics and Automation, vol. 18, pp. 1-10, 2002.
[9] S. Ali A. Moosavian and R. Rastegari, Multiple-arm space free-flying robots for
manipulating objects with force tracking restrictions, Robotics and Autonomous
System, vol. 54, issue 10, pp. 779-788, 2006.
[10]Z. Li, S. S. Ge and Z. Wang, Robust adaptive control of coordinated multiple mobile
manipulators, Mechatronics, vol. 18, issues 5-6, pp. 239-250, 2008.](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-50-320.jpg)
![Target Point Manipulation Inside a Deformable Object 41
[11]M. Namvar and F. Aghili, Adaptive force-motion control of coordinated robots
interacting with geometrically unknown environments, IEEE Transactions on
Robotics, vol. 21, issue 4, pp. 678-694, 2005.
[12]M. Saha and P. Isto, Manipulation planning for deformable linear objects, IEEE
Transactions on Robotics, vol. 23, issue 6, pp. 1141-1150, 2007.
[13]Y. Zhang, B. K. Chen, X. Liu and Y. Sun, Autonomous robotic pick-and-place of
microobjects, IEEE Transactions on Robotics, vol. 26, issue 1, pp. 200-207, 2010.
[14]D. Sun and Y. H. Liu, Modeling and impedance control of a two-manipulator system
handling a flexible beam, ASME Journal of Dynamic Systems, Measurement, and
Control, vol. 119, no. 4, pp. 736-742 ,1997.
[15]A. Tavasoli, M. Eghtesad and H. Jafarian, Two-time scale control and observer design
for trajectory tracking of two cooperating robot manipulators moving a flexible
beam, Robotics and Autonomous Systems, vol. 57, issue 2, pp. 212-221, 2009.
[16]S. F. F. Gibson and B. Mirtich, A survey of deformable modeling in computer graphics,
MERL Technical Report, TR97-19, 1997.
[17]Y. Zhang, E. C. Prakash and E. Sung, A new physical model with multilayer architecture
for facial expression animation using dynamics adaptive mesh, IEEE Transactions
on Visualization and Computer Graphics, vol. 10, issue 3, pp. 339-352, 2004.
[18]M. Meyer, G. Debunne, M. Desbrun and A. H. Barr, Interactive animation of cloth-like
objects in virtual reality, The Journal of Visualization and Computer Animation,
vol. 12, no. 1, pp. 1-12(12), 2001.
[19]A. Joukhadar and C. Laugier, Dynamic simulation: model, basic algorithms, and
optimization, In J.-P. Laumond and M. Overmars, editors, Algorithms for Robotic
Motion and Manipulation, pp. 419-434, A. K. Peters Ltd., 1997.
[20]A. M. Howard, and G. A. Bekey, Recursive learning for deformable object manipulation,
Proc. of International Conference on Advanced Robotics, pp. 939-944, 1997.
[21]R. M. Koch, M. H. Gross, F. R. Carls, D. F. Von Buren, G. Fankhauser, Y. I. H. Parish,
Simulating facial surgery using finite element models, In ACM SIGGRAPH 96
Conf. Proc., pp. 421-428, 1996.
[22]S. D. Pieper, D. R. Laub, Jr., and J. M. Rosen, A finite element facial model for simulating
plastic surgery, Plastic and Reconstructive Surgery, 96(5), pp. 1100-1105, Oct. 1995.
[23]J. K. Salisbury, Kinematic and force analysis of articulated hands, PhD Thesis,
Department of Mechanical Engineering, Standford University, Standford, CA, 1982.
[24]J. Kerr, Analysis of multifingered hand, PhD Thesis, Department of Mechanical
Engineering, Standford University, Standford, CA, 1984.
[25]J. M. Abel, W. Holzmann and J. M. McCarthy, On grasping planar objects with two
articulated fingers, IEEE Journal of Robotics and Automation, vol. 1, pp. 211-214,
1985.
[26]M. R. Cutkosky, Grasping and fine manipulation for automated manufacturing, PhD
Thesis, Department of Mechanical Engineering, Carnegie Mellon University,
Pittsburgh, PA, 1985.
[27]C. Ferrari and J. Canny, Planning optimal grasp, IEEE International Conference on
Robotics and Automation, pp. 2290-2295, 1992.
[28]S. Garg and A. Dutta, Grasping and manipulation of deformable objects based on
internal force requirements, International Journal of Advanced Robotic Systems,
Vol. 3, No. 2, pp. 107-114, 2006.](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-51-320.jpg)
![Robotic Systems – Applications, Control and Programming
42
[29]T. Watanabe and T. Yoshikawa, Grasping optimization using a required external force
set, IEEE Transactions on Automation Science and Engineering, Vol. 4, pp. 52-66,
2007.
[30]D. Ding, Y. H. Liu and S. Wang, Compuattion of 3D form-closure grasps, IEEE
Transactions on Robotics and Autonamtion, Vol. 17, pp. 515-522, 2001.
[31]A. Bicchi and V. Kumar, Robotic grasping and contact: a review, Proc. IEEE
International Conference on Robotics and Automation, pp. 348-353, 2000.
[32]M. T. Mason, Mechanics of robotic manipulation, The MIT Press.
[33]J. Cornella, R. Suarez, R. Carloni and C. Melchiorri, Dual programming based approach
for optimal grasping force distribution, Journal of Mechatronics, Vol. 18, pp. 348-
356, 2008.
[34]P. Saut, C. Remond, V. Perdereau and M. Drouin, Online computation of grasping force
in multi-fingered hands, Proc. IEEE/RSJ International Conference on Intelligent
Robots and Systems, pp. 1223-1228, 2005.
[35]K. Gopalakrishanan and K. Goldberg, D-space and deform closure grasps of deformable
parts, International Journal of Robotics Research, vol. 24, pp. 889-910, 2005.
[36]G. L. Foresti and F. A. Pellegrino, Automatic visual recognition of deformable objects for
grasping and manipulation, IEEE Transaction on Systems, Man, and Cybernetics:
Applications and Reviews, vol. 34, pp. 325-333, 2004.
[37]H. Wakamatsu, S. Hirai and K. Iwata, Static analysis of deformable object grasping
based on bounded force closure, IEEE International Conference on Robotics and
Automation, vol.4, pp. 3324-3329, 1996.
[38]R. J. Anderson and M. W. Spong, Asymptotic stability for force reflecting teleoperators
with time delays, in, 1989. IEEE International Conference on Robotics and
Automation, vol.3, pp. 1618-1625, 1989.
[39]S. Stramigioli, C. Melchiorri and S. Andreotti, A passivity-based control scheme for
robotic grasping and manipulation, The 38th Conference on Decision and Control,
Phoenix, Arizona, USA, 1999.
[40]B. Hannaford and J.-H. Ryu, Time domain passivity control of haptic interfaces, IEEE
International Conference on Robotics and Automation, vol.2, pp. 1863-1869, 2001.
[41]J.-H. Ryu, D.-S. Kwon and B. Hannaford, Control of a flexible manipulator with
noncollocated feedback: time-domain passivity approach, IEEE Transactions on
Robotics, vol. 20, pp. 776-780, 2004.
[42]T. Wada, S. Hirai, S. Kawamura and N. Kamiji, Robust manipulation of deformable
objects by a simple PID feedback, IEEE International Conference on Robotics and
Automation, vol.1, pp. 85-90, 2001.
[43]V. D. Nguyen, Constructing force-closure grasps, IEEE International Conference on
Robotics and Automation, pp. 1368-1373, 1986.
[44]J. Ponce and B. Faverjon, On computing three-finger force-closure grasps of polygonal
objects, Fifth International Conference on Advanced Robotics, vol.2, pp. 1018-1023,
1991.
[45]J. E. Colgate and N. Hogan, Robust control of dynamically interacting systems,
International Journal of Control, vol. 48, pp. 65 - 88, 1988.
[46]J. J.-H. Ryu and C. Preusche, Stable bilateral control of teleoperators under time-varying
communication delay: time domain passivity approach, IEEE International
Conference on Robotics and Automation, pp. 3508-3513, 2007.](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-52-320.jpg)
![Winglebury, and we know that Great Winglebury was a fictitious
name for Rochester. But which was the inn that received this
whimsical name at the hands of the novelist under whose roof
Horace Hunter penned his challenge to that base umbrella-maker
Alexander Trott, we are unable to state. On the other hand, the
Winglebury Arms where Alexander Trott was staying at the time was
the Bull Hotel, Rochester.[3] The Red House, Battersea, casually
mentioned in the chapter on “The River” as the “Red-us,” was a
popular tavern and tea-gardens in those days and notorious for its
pigeon-shooting; indeed, tradition has it that it took the lead in the
quality and quantity of the sport, and that the crack shots assembled
there to determine important matches. It was also famous as the
winning-post of many a boat race from Westminster Bridge, and was
the place “where all the prime of life lads assembled,” the joy and
fun of which is vividly described by Dickens in the chapter referred
to. It was a red-bricked building, and a prominent landmark of what
was then known as Battersea Fields, the one-time scene of many a
duel.
The Cross Keys mentioned in the chapter on “Omnibuses” we have
already referred to when dealing with Great Expectations; whilst for
particulars of the Golden Cross, the busy coaching-inn mentioned in
“Hackney Coach Stands,” and in “Early Coaches,” we must refer the
reader to “The Inns and Taverns of Pickwick.”
The Freemasons’ Tavern in the chapter on “Public Dinners” does not
receive much attention from Dickens. He is describing the public
dinner given in aid of the “Indigent Orphans Friends’ Benevolent
Institution,” and no reference beyond the use of the name is made
to the building itself. The tavern still stands to-day, and no doubt
more glorious in its splendour than it was on the occasion of the
public dinner Dickens refers to. It is used to-day for similar purposes,
the ceremony and atmosphere at which being little changed from
what it was then. It is interesting to note that in the same building a
farewell dinner was given Dickens on the eve of his departure for
America in 1867, with Lord Lytton in the chair.](https://image.slidesharecdn.com/2047119-250529122314-40a1b1b4/85/Robotic-Systems-Applns-Ctl-And-Pgmg-A-Dutta-60-320.jpg)
Robotic Systems Applns Ctl And Pgmg A Dutta
- 1. Robotic Systems Applns Ctl And Pgmg A Dutta download https://ebookbell.com/product/robotic-systems-applns-ctl-and- pgmg-a-dutta-4094238 Explore and download more ebooks at ebookbell.com
- 2. Here are some recommended products that we believe you will be interested in. You can click the link to download. Robotic Systems For Handling And Assembly 1st Edition Ulrike Thomas https://ebookbell.com/product/robotic-systems-for-handling-and- assembly-1st-edition-ulrike-thomas-4194590 Robotic Systems Applications Control And Programming Ashish Dutta https://ebookbell.com/product/robotic-systems-applications-control- and-programming-ashish-dutta-5903362 Robotic Systems Applications Control And Programming Ashish Dutta https://ebookbell.com/product/robotic-systems-applications-control- and-programming-ashish-dutta-5411184 Intelligent Robotic Systems Design Planning And Control Witold Jacak https://ebookbell.com/product/intelligent-robotic-systems-design- planning-and-control-witold-jacak-4103108
- 3. Multiagent Robotic Systems Jiming Liu https://ebookbell.com/product/multiagent-robotic-systems-jiming- liu-4111170 Autonomous Robotic Systems Soft Computing And Hard Computing Methodologies And Applications 1st Edition J Mira https://ebookbell.com/product/autonomous-robotic-systems-soft- computing-and-hard-computing-methodologies-and-applications-1st- edition-j-mira-4189056 Intelligent Robotic Systems First Edition Tzafestas Spyros G https://ebookbell.com/product/intelligent-robotic-systems-first- edition-tzafestas-spyros-g-12060092 Towards Autonomous Robotic Systems 23rd Annual Conference Taros 2022 Culham Uk September 79 2022 Proceedings Salvador Pachecogutierrez https://ebookbell.com/product/towards-autonomous-robotic-systems-23rd- annual-conference-taros-2022-culham-uk-september-79-2022-proceedings- salvador-pachecogutierrez-46098666 Optoelectronic Devices In Robotic Systems Oleg Sergiyenko Editor https://ebookbell.com/product/optoelectronic-devices-in-robotic- systems-oleg-sergiyenko-editor-46890924
- 5. Contents Preface IX Part 1 Applications 1 Chapter 1 Modular Robotic Approach in Surgical Applications – Wireless Robotic Modules and a Reconfigurable Master Device for Endoluminal Surgery – 3 Kanako Harada, Ekawahyu Susilo, Takao Watanabe, Kazuya Kawamura, Masakatsu G. Fujie, Arianna Menciassi and Paolo Dario Chapter 2 Target Point Manipulation Inside a Deformable Object 19 Jadav Das and Nilanjan Sarkar Chapter 3 Novel Assistive Robot for Self-Feeding 43 Won-Kyung Song and Jongbae Kim Chapter 4 Robot Handling Fabrics Towards Sewing Using Computational Intelligence Methods 61 Zacharia Paraskevi Chapter 5 Robotic Systems for Radiation Therapy 85 Ivan Buzurovic, Tarun K. Podder and Yan Yu Chapter 6 Robotic Urological Surgery: State of the Art and Future Perspectives 107 Rachid Yakoubi, Shahab Hillyer and Georges-Pascal Haber Chapter 7 Reconfigurable Automation of Carton Packaging with Robotic Technology 125 Wei Yao and Jian S. Dai Chapter 8 Autonomous Anthropomorphic Robotic System with Low-Cost Colour Sensors to Monitor Plant Growth in a Laboratory 139 Gourab Sen Gupta, Mark Seelye, John Seelye and Donald Bailey
- 6. VI Contents Part 2 Control and Modeling 159 Chapter 9 CPG Implementations for Robot Locomotion: Analysis and Design 161 Jose Hugo Barron-Zambrano and Cesar Torres-Huitzil Chapter 10 Tracking Control in an Upper Arm Exoskeleton with Differential Flatness 183 E. Y. Veslin, M. Dutra, J. Slama, O. Lengerke and M. J. M. Tavera Chapter 11 Real-Time Control in Robotic Systems 209 Alex Simpkins Chapter 12 Robot Learning from Demonstration Using Predictive Sequence Learning 235 Erik Billing, Thomas Hellström and Lars-Erik Janlert Chapter 13 Biarticular Actuation of Robotic Systems 251 Jan Babič Chapter 14 Optimization and Synthesis of a Robot Fish Motion 271 Janis Viba, Semjons Cifanskis and Vladimirs Jakushevich Chapter 15 Modeling of Elastic Robot Joints with Nonlinear Damping and Hysteresis 293 Michael Ruderman Chapter 16 Gravity-Independent Locomotion: Dynamics and Position-Based Control of Robots on Asteroid Surfaces 313 Marco Chacin and Edward Tunstel Chapter 17 Kinematic and Inverse Dynamic Analysis of a C5 Joint Parallel Robot 339 Georges Fried, Karim Djouani and Amir Fijany Chapter 18 Utilizing the Functional Work Space Evaluation Tool for Assessing a System Design and Reconfiguration Alternatives 361 A. Djuric and R. J. Urbanic Chapter 19 Requirement Oriented Reconfiguration of Parallel Robotic Systems 387 Jan Schmitt, David Inkermann, Carsten Stechert, Annika Raatz and Thomas Vietor Part 3 Vision and Sensors 411 Chapter 20 Real-Time Robotic Hand Control Using Hand Gestures 413 Jagdish Lal Raheja, Radhey Shyam, G. Arun Rajsekhar and P. Bhanu Prasad
- 7. Contents VII Chapter 21 Robotics Arm Visual Servo: Estimation of Arm-Space Kinematics Relations with Epipolar Geometry 429 Ebrahim Mattar Chapter 22 Design and Construction of an Ultrasonic Sensor for the Material Identification in Robotic Agents 455 Juan José González España, Jovani Alberto Jiménez Builes and Jaime Alberto Guzmán Luna Part 4 Programming and Algorithms 471 Chapter 23 Robotic Software Systems: From Code-Driven to Model-Driven Software Development 473 Christian Schlegel, Andreas Steck and Alex Lotz Chapter 24 Using Ontologies for Configuring Architectures of Industrial Robotics in Logistic Processes 503 Matthias Burwinkel and Bernd Scholz-Reiter Chapter 25 Programming of Intelligent Service Robots with the Process Model “FRIEND::Process” and Configurable Task-Knowledge 529 Oliver Prenzel, Uwe Lange, Henning Kampe, Christian Martens and Axel Gräser Chapter 26 Performance Evaluation of Fault-Tolerant Controllers in Robotic Manipulators 553 Claudio Urrea, John Kern and Holman Ortiz Chapter 27 An Approach to Distributed Component-Based Software for Robotics 571 A. C. Domínguez-Brito, J. Cabrera-Gámez, J. D. Hernández-Sosa, J. Isern-González and E. Fernández-Perdomo Chapter 28 Sequential and Simultaneous Algorithms to Solve the Collision-Free Trajectory Planning Problem for Industrial Robots – Impact of Interpolation Functions and the Characteristics of the Actuators on Robot Performance 591 Francisco J. Rubio, Francisco J. Valero, Antonio J. Besa and Ana M. Pedrosa Chapter 29 Methodology for System Adaptation Based on Characteristic Patterns 611 Eva Volná, Michal Janošek, Václav Kocian, Martin Kotyrba and Zuzana Oplatková
- 9. Preface Over the last few decades the focus of robotics research has moved beyond the traditional area of industrial applications to newer areas, including healthcare and domestic applications. These newer applications have given a strong impetus to the development of advanced sensors, control strategies and algorithms. The first section of this book contains advanced applications of robotics in surgery, rehabilitation, modular robotics among others. This is followed by sections on control and sensors, while the fourth section is devoted to robot algorithms. I would like to thank all the authors for entrusting us with their work and specially the editorial members of InTech publishing. Dr. Ashish Dutta Department of Mechanical Engineering Indian Institute of Technology, Kanpur India
- 13. 1 Modular Robotic Approach in Surgical Applications – Wireless Robotic Modules and a Reconfigurable Master Device for Endoluminal Surgery – Kanako Harada, Ekawahyu Susilo, Takao Watanabe, Kazuya Kawamura, Masakatsu G. Fujie, Arianna Menciassi and Paolo Dario 1The University of Tokyo 2Scuola Superiore Sant’Anna 3Italian Institute of Technology 4Waseda University 1,4Japan 2,3Italy 1. Introduction The trend in surgical robots is moving from traditional master-slave robots to miniaturized devices for screening and simple surgical operations (Cuschieri, A. 2005). For example, capsule endoscopy (Moglia, A. 2007) has been conducted worldwide over the last five years with successful outcomes. To enhance the dexterity of commercial endoscopic capsules, capsule locomotion has been investigated using legged capsules (Quirini, M. 2008) and capsules driven by external magnetic fields (Sendoh, M. 2003; Ciuti, G. 2010; Carpi, F. 2009). Endoscopic capsules with miniaturized arms have also been studied to determine their potential for use in biopsy (Park, S.-K. 2008). Furthermore, new surgical procedures known as natural orifice transluminal endoscopic surgery (NOTES) and Single Port Access surgery are accelerating the development of innovative endoscopic devices (Giday, S. 2006; Bardaro, S.J. 2006). These advanced surgical devices show potential for the future development of minimally invasive and endoluminal surgery. However, the implementable functions in such devices are generally limited owing to space constraints. Moreover, advanced capsules or endoscopes with miniaturized arms have rather poor dexterity because the diameter of such arms must be small (i.e. a few millimeters), which results in a small force being generated at the tip. A modular surgical robotic system known as the ARES (Assembling Reconfigurable Endoluminal Surgical system) system has been proposed based on the aforementioned motivations (Harada, K. 2009; Harada, K. 2010; Menciassi, A. 2010). The ARES system is designed for screening and interventions in the gastrointestinal (GI) tracts to overcome the intrinsic limitations of single-capsules or endoscopic devices. In the proposed system,
- 14. Robotic Systems – Applications, Control and Programming 4 miniaturized robotic modules are ingested and assembled in the stomach cavity. The assembled robot can then change its configuration according to the target location and task. Modular surgical robots are interesting owing to their potential for application as self- reconfigurable modular robots and innovative surgical robots. Many self-reconfigurable modular robots have been investigated worldwide (Yim, M. 2007; Murata, S. 2007) with the goal of developing systems that are robust and adaptive to the working environment. Most of these robots have been designed for autonomous exploration or surveillance tasks in unstructured environments; therefore, there are no strict constraints regarding the number of modules, modular size or working space. Because the ARES has specific applications and is used in the GI tract environment, it raises many issues that have not been discussed in depth in the modular robotic field. Modular miniaturization down to the ingestible size is one of the most challenging goals. In addition, a new interface must be developed so that surgeons can intuitively maneuver the modular surgical robot. The purpose of this paper is to clarify the advantages of the modular approach in surgical applications, as well as to present proof of concept of the modular robotic surgical system. The current paper is organized as follows: Section 2 describes the design of the ARES system. Section 3 details the design and prototyping of robotic modules, including the experimental results. Section 4 describes a reconfigurable master device designed for the robotic modules, and its preliminary evaluation is reported. 2. Design of the modular surgical system 2.1 Clinical indications and proposed procedures The clinical target of the ARES system is the entire GI tract, i.e., the esophagus, stomach, small intestine, and colon. Among GI tract pathologies that can benefit from modular robotic features, biopsy for detection of early cancer in the upper side of the stomach (the fundus and the cardia) was selected as the surgical task to be focused on as a first step. Stomach cancer is the second leading cause of cancer-related deaths worldwide (World Health Organization 2006), and stomach cancer occurring in the upper side of the stomach has the worst outcome in terms of the 5-year survival ratio (Pesic, M. 2004). Thus, early diagnosis of cancer utilizing an advanced endoluminal device may lead to better prognosis. The stomach has a large volume (about 1400 ml) when distended, which provides working space to assemble the ingested robotic modules and change the topology of the assembled robot inside (i.e. reconfiguration). Each robotic module should be small enough to be swallowed and pass through the whole GI tract. Because the size of the commercial endoscopic capsules (11 mm in diameter and 26 mm in length (Moglia, A. 2007)) has already been shown to be acceptable for the majority of patients as an ingestible device, each module needs to be miniaturized to this size before being applied to clinical cases. The surgical procedures proposed for the ARES system (Harada, K. 2010) are shown in Fig. 1. Prior to the surgical procedure, the patient drinks a liquid to distend the stomach to a volume of about 1400 ml. Next, the patient ingests 10-15 robotic modules that complete the assembly process before the liquid naturally drains away from the stomach in 10-20 minutes. The number of the modules swallowed depends on the target tasks and is determined in advance based on the pre-diagnosis. Magnetic self-assembly in the liquid using permanent magnets was selected for this study since its feasibility has already been demonstrated (Nagy, Z. 2007). Soon after the assembly, the robot configures its topology
- 15. Modular Robotic Approach in Surgical Applications – Wireless Robotic Modules and a Reconfigurable Master Device for Endoluminal Surgery – 5 according to preoperative planning by repeated docking and undocking of the modules (the undocking mechanism and electrical contacts between modules are necessary for reconfiguration, but they have not been implemented in the presented design). The robotic modules are controlled via wireless bidirectional communication with a master device operated by the surgeon, while the progress in procedure is observed using intraoperative imaging devices such as fluoroscopy and cameras mounted on the modules. After the surgical tasks are completed, the robot reconfigures itself to a snake-like shape to pass through the pyloric sphincter and travel to examine the small intestine and the colon, or it completely disassembles itself into individual modules so that it can be brought out without external aid. One of the modules can bring a biopsy tissue sample out of the body for detailed examination after the procedure is complete. Fig. 1. Proposed procedures for the ARES system 2.2 Advantages of the modular approach in surgical applications The modular approach has great potential to provide many advantages to surgical applications. These advantages are summarized below using the ARES system as shown in Fig.2. The numbering of the items in Fig.2 is correlated with the following numbering. i. The topology of the modular surgical robot can be customized for each patient according to the location of the disease and the size of the body cavity in which the modular robot is deployed. A set of functional modules such as cameras, needles and forceps can be selected for each patient based on the necessary diagnosis and surgical operation. ii. The modular approach facilitates delivery of more components inside a body cavity that has small entrance/exit hole(s). As there are many cavities in the human body, the modular approach would benefit treatment in such difficult-to-reach places. Because
- 16. Robotic Systems – Applications, Control and Programming 6 several functional modules can be used simultaneously, the modular robot may perform rather complicated tasks that a single endoscopic capsule or an endoscopic device is not capable of conducting. For example, if more than two camera modules are employed, the surgeon can conduct tasks while observing the site from different directions. iii. Surgical tools of relatively large diameter can be brought into the body cavity. Conventionally, small surgical forceps that can pass through an endoscopic channel of a few millimeters have been used for endoluminal surgery. Conversely, surgical devices that have the same diameter as an endoscope can be used in the modular surgical system. Consequently, the force generated at the tip of the devices would be rather large, and the performance of the functional devices would be high. iv. The surgical system is more adaptive to the given environment and robust to failures. Accordingly, it is not necessary for the surgical robot to equip all modules that might be necessary in the body because the surgeons can decide whether to add modules with different functionalities, even during the surgical operation. After use, the modules can be detached and discarded if they are not necessary in the following procedures. Similarly, a module can be easily replaced with a new one in case of malfunction. As these advantages suggest, a modular surgical robot would be capable of achieving rather complicated tasks that have not been performed using existing endoluminal surgical devices. These advantages are valid for modular robots that work in any body cavity with a small entrance and exit. Moreover, this approach may be introduced to NOTES or Single Port Access surgery, in which surgical devices must reach the abdominal cavity through a small incision. In Section 3, several robotic modules are proposed, and the performance of these modules is reported to show the feasibility of the proposed surgical system. (i) (ii) (iii) (iv) Fig. 2. Advantages of the modular approach in surgical applications
- 17. Modular Robotic Approach in Surgical Applications – Wireless Robotic Modules and a Reconfigurable Master Device for Endoluminal Surgery – 7 3. Robotic modules 3.1 Design and prototyping of the robotic modules Figure 3 shows the design and prototypes of the Structural Module and the Biopsy Module (Harada, K. 2009, Harada, K. 2010). The Structural Module has two degrees of freedom (±90° of bending and 360° of rotation). The Structural Module contains a Li-Po battery (20 mAh, LP2-FR, Plantraco Ltd., Canada), two brushless DC geared motors that are 4 mm in diameter and 17.4 mm in length (SBL04-0829PG337, Namiki Precision Jewel Co. Ltd., Japan) and a custom-made motor control board capable of wireless control (Susilo, E. 2009). The stall torque of the selected geared motor is 10.6 mNm and the speed is 112 rpm when controlled by the developed controller. The bending mechanism is composed of a worm and a spur gear (9:1 gear reduction), whereas the rotation mechanism is composed of two spur gears (no gear reduction). All gears (DIDEL SA, Switzerland) were made of nylon, and they were machined to be implemented in the small space of the capsule. Two permanent magnets (Q-05-1.5-01-N, Webcraft GMbH, Switzerland) were attached at each end of the module to help with self-alignment and modular docking. The module is 15.4 mm in diameter and 36.5 mm in length; it requires further miniaturization before clinical application. The casing of the prototype was made of acrylic plastic and fabricated by 3D rapid prototyping (Invison XT 3-D Modeler, 3D systems, Inc., USA). The total weight is 5.6 g. Assuming that the module would weigh 10 g with the metal chassis and gears, the maximum torque required for lifting two connected modules is 5.4 mNm for both the bending DOF and rotation DOF. Assuming that the gear transmission efficiency for the bending mechanism is 30%, the stall torque for the bending DOF is 28.6 mNm. On the other hand, the stall torque for the rotation DOF is 8.5 mNm when the transmission efficiency for the rotation mechanism is 80%. The torque was designed to have sufficient force for surgical operation, but the transmission efficiency of the miniaturized plastic gears was much smaller than the theoretical value as explained in the next subsection. • Controller The aforementioned brushless DC motor came with a dedicated motor driving board (SSD04, Namiki Precision Jewel Co., Ltd., 19.6 mm × 34.4 mm × 3 mm). This board only allows driving of one motor; hence, two boards are required for a robotic module with 2 DOFs. Because there was not sufficient space for the boards in the robotic module, a custom made high density control board was designed and developed in-house. This control board consisted of one CC2430 microcontroller (Texas Instrument, USA) as the main wireless controller and three sets of A3901 dual bridge motor drivers (Allegro MicroSystem, Inc., USA). The fabricated board is 9.6 mm in diameter, 2.5 mm in thickness and 0.37 g in weight, which is compatible with swallowing. The A3901 motor driver chip was originally intended for a brushed DC motor, but a software commutation algorithm was implemented to control a brushless DC motor as well. An IEEE 802.15.4 wireless personal area network (WPAN) was introduced as an embedded feature (radio peripheral) of the microcontroller. The implemented algorithm enables control of the selected brushless DC motor in Back Electro- Motive Force (BEMF) feedback mode or slow speed stepping mode. When the stepping mode is selected, the motor can be driven with a resolution of 0.178º. For the modular approach, each control board shall be equipped with a wired locating system for intra-modular communication in addition to the wireless communication. Aside from wireless networking, the wired locating system, which is not implemented in the presented design, would be useful for identification of the sequence of the docked modules
- 18. Robotic Systems – Applications, Control and Programming 8 in real time. The wired locating system is composed of three lines, one for serial multidrop communication, one for a peripheral locator and one as a ground reference. When the modules are firmly connected, the intra-modular communication can be switched from wireless to wired to save power while maintaining the predefined network addresses. When one module is detached intentionally or by mistake, it will switch back to wireless mode. • Battery The battery capacity carried by each module may differ from one to another (e.g. from 10 mAh to 50 mAh) depending on the available space inside the module. For the current design, a 20 mAh Li-Po battery was selected. Continuous driving of the selected motor on its maximum speed using a 20 mAh Li-Po battery was found to last up to 17 minutes. A module does not withdraw power continuously because the actuation mechanisms can maintain their position when there is no current to the motor owing to its high gear reduction (337:1). A module consumes power during actuation, but its power use is very low in stand-by mode. • Biopsy Module The Biopsy Module is a Functional Module that can be used to conduct diagnosis. The grasping mechanism has a worm and two spur gears, which allows wide opening of the grasping parts. The grasping parts can be hidden in the casing at the maximum opening to prevent tissue damage during ingestion. The motor and other components used for the Biopsy Module are the same as for the Structural Module. The brushless DC geared motors (SBL04-0829PG337, Namiki Precision Jewel Co. Ltd., Japan), the control board, a worm gear and two spur gears (9:1 gear reduction) were implemented in the Biopsy Module. A permanent magnet (Q-05-1.5-01-N, Webcraft GMbH, Switzerland) was placed at one side to be connected to another Structural Module. Fig. 3. Design and prototypes of the structural module (left) and the biopsy module (right)
- 19. Modular Robotic Approach in Surgical Applications – Wireless Robotic Modules and a Reconfigurable Master Device for Endoluminal Surgery – 9 The Biopsy Module can generate a force of 7.1 N at its tip, and can also open the grasping parts to a width of 19 mm with an opening angle of 90 degrees. These values are much larger than those of conventional endoscopic forceps, which are 2-4 mm in diameter. As a demonstration, Figure 3 shows the Biopsy Module holding a coin weighing 7.5 g. In conventional endoscopy, forceps are inserted through endoscopic channels that are parallel to the direction of the endoscopic view, which often results in the forceps hiding the target. Conversely, the Biopsy Module can be positioned at any angle relative to the endoscopic view owing to the modular approach, thereby allowing adequate approach to the target. 3.2 Performance of the Structural Module The mechanical performance of the bending and rotation DOFs of the Structural Module was measured in preliminary tests (Menciassi, A. 2010), and the results are summarized in Fig.4. The bending angle was varied by up to ± 90° in steps of 10° three times in succession. The measured range of the bending angle was -86.0° to +76.3°, and the maximum error was 15.8°. The rotation angle was increased from 0° to 180° in steps of 45° three times in succession, and the measured range of the rotational angle was between 0° and 166.7° with a maximum error of 13.3°. The difference between the driven angle and the measured angle was due to backlash of the gears and the lack of precision and stiffness of the casing made by 3D rapid prototyping. Regardless of the errors and the hysteresis, the repeatability was sufficient for the intended application for both DOFs. These results indicate that the precision of each motion can be improved by changing the materials of the gears and the casing. Since the motor can be controlled with a resolution of 0.178°, very precise surgical tasks can be achieved using different manufacturing processes. Measured Angle (deg.) CommandedAngle(deg.) 0 45 90 135 180 180 135 90 45 0 45 90 -45 -90 Measured Angle (deg.) 45 90 -45 -90 CommandedAngle(deg.) 0 Fig. 4. Bending angle measurement (left), rotation angle measurement (middle), and torque measurement (right) (Menciassi, A. 2010) In addition to the angle measurements, both bending and rotation torque were measured. The torque was measured by connecting cylindrical parts with permanent magnets at both ends until the bending/rotational motion stopped. The length and weight of each cylinder was designed in advance, and several types of cylinders were prepared. The measured bending torque was 6.5 mNm and the rotation torque was 2.2 mNm. The figure also shows one module lifting up two modules attached to its bending mechanism as a demonstration. The performance in terms of precision and generated torque, which are very important for reconfiguration and surgical tasks, was sufficient; however, the precision was limited owing
- 20. Robotic Systems – Applications, Control and Programming 10 to the aforementioned fabrication problems. The thin walls of the casing made of acrylic plastic were easily deformed, which caused friction between the parts. The casing made of metal or PEEK and tailor-made metal gears with high precision will improve the mechanism rigidity and performance, thus producing the optimal stability. 3.3 Possible designs of robotic modules Figure 5 shows various designs of robotic modules that can be implemented in the modular surgical robot. The modules can be categorized into three types: structural modules, functional modules, and other modules. Structural modules are used to configure a robotic topology. Functional modules are used for diagnosis or intervention, while other modules can be added to enhance the performance and robustness of the robotic system. Obviously, an assembled robot made of different types of modules (i.e. a robot with high heterogeneity) may provide high dexterity, but the self-assembly in the stomach and control of the modules would become more difficult. To optimize the level of heterogeneity, self-assembly of the robotic modules must be developed so that the reconfiguration of the robotic topology following the self-assembly can be planned in advance. Employing pre-assembled modular arms or tethered modules can be another option to facilitate assembly in a body cavity; however, this would require further anesthesia, and it would hinder the promotion of massive screening. Fig. 5. Various designs of the robotic modules
- 21. Modular Robotic Approach in Surgical Applications – Wireless Robotic Modules and a Reconfigurable Master Device for Endoluminal Surgery – 11 4. Reconfigurable master device 4.1 Design and prototyping of the reconfigurable master device One main advantage of using a modular approach in surgical applications is the adaptivity to the given environment as mentioned in Section 2.2. Wherever the robotic platform is deployed in the GI tract, the robotic topology can be changed based on preoperative plans or the in-situ situation to fit in any particular environment. This dynamic changing and reshaping of the robotic topology should be reflected on the user interface. Since it is possible for a robotic topology to have redundant DOFs, the master device for the modular surgical system needs to be able to handle the redundancy that is inherent to modular robots. Based on these considerations, we propose a reconfigurable master device that resembles the robotic platform (Fig.6). When the assembled robot changes its topology, the master device follows the same configuration. The robotic module shown in Fig. 6 has a diameter of 15.4 mm, while a module of the reconfigurable master device has a diameter of 30 mm. The master modules can be easily assembled or disassembled using set screws, and it takes only a few seconds to connect one module to another. Each robotic module is equipped with two motors as described in the previous section; thus, each master module is equipped with two potentiometers (TW1103KA, Tyco Electronics) that are used as angular position sensors. Calculating the angular position of each joint of the reconfigurable master device is quite straightforward. A common reference voltage is sent from a data acquisition card to all potentiometers, after which the angular position can be calculated from the feedback readings. Owing to the identical configuration, the angle of each joint of the robotic modules can be easily determined, even if the topology has redundancy. Fig. 6. Robotic modules (top line) and the reconfigurable master device (bottom line): one module (left), assembled modules (middle) and prototypes (right)
- 22. Robotic Systems – Applications, Control and Programming 12 The advantages of the proposed master device include intuitive manipulation. For example, the rotational movement of a structural module used to twist the arm is limited to ± 180°, and the master module also has this limitation. This helps surgeons intuitively understand the range of the motion and the reachable working space of the modules. Using a conventional master manipulator or an external console, it is possible that the slave manipulator cannot move owing to its mechanical constraints, while the master manipulator can still move. However, using the proposed master device, the surgeon can intuitively understand the mechanical constraints by manipulating the master device during practice/training. Furthermore, the position of the master arm can indicate where the robotic modules are, even if they are outside of the camera module's view. These characteristics increase the safety of the operation. This feature is important because the entire robotic system is placed inside the body. In other surgical robotic systems, the position or shape of the robotic arms is not important as they are placed outside of the body and can be seen during operation. Unlike other master devices, it is also possible for two or more surgeons to move the reconfigurable master device together at the same time using multi arms with redundant DOFs. 4.2 Evaluation A simulation-based evaluation setup was selected to simplify the preliminary evaluation of the feasibility of the reconfigurable master device. The authors previously developed the Slave Simulator to evaluate workloads for a master-slave surgical robot (Kawamura, K. 2006). The Slave Simulator can show the motion of the slave robot in CG (Computer Graphics), while the master input device is controlled by an operator. Because the simulator can virtually change the parameters of the slave robot or its control, it is easy to evaluate the parameters as well as the operability of the master device. This Slave Simulator was appropriately modified for the ARES system. The modified Slave Simulator presents the CG models of the robotic modules to the operator. The dimension and DOFs of each module in CG were determined based on the design of the robotic modules. The angle of each joint is given by the signal from the potentiometers of the reconfigurable master device, and the slave modules in CG move in real time to reproduce the configuration of the master device. This Slave Simulator is capable of altering joint positions and the number of joints of the slave arms in CG so that the workspace of the reconfigurable master device can be reproduced in a virtual environment for several types of topologies. The simulator is composed of a 3D viewer that uses OpenGL and a physical calculation function. This function was implemented to detect a collision between the CG modules and an object placed in the workspace. To simplify the experiments to evaluate the feasibility of the proposed master device and usefulness of the developed simulator, only one arm of the reconfigurable master device was used. Three topologies that consist of one Biopsy Module and one or two Structural Module(s) were selected as illustrated in Fig.7. Topology I consists of a Structural Module and a Biopsy Module, and the base is fixed so that the arm appears with an angle of 45 degrees. One Structural Module is added to Topology I to configure Topology II, and Topology III is identical to Topology II, but placed at 0 degrees. Both Topology II and Topology III have redundant DOFs. The projection of the workspace of each arm and the shared workspace are depicted in Fig.8. A target object on which the arm works in the experiments must be placed in this shared area, which makes it easy to compare topologies.
- 23. Modular Robotic Approach in Surgical Applications – Wireless Robotic Modules and a Reconfigurable Master Device for Endoluminal Surgery – 13 A bar was selected as the target object instead of a sphere because the height of the collision point is different for each topology when the object appears in the same position in the 2D plane. The experiment was designed so that a bar appears at random in the shared workspace. The bar represents a target area at which the Biopsy Module needs to collect tissue samples, and this experiment is a simple example to select one topology among three choices given that the arm can reach the target. We assumed that this choice may vary depending on the user, and this experiment was designed to determine if the reconfigurability of the master device, i.e. customization of the robot, provides advantages and improves performance. Topology I Topology II Topology III Combination Master device Simulator Fig. 7. Three topologies used in the experiments During the experiment, the operator of the reconfigurable master device could hear a beeping sound when the distal end of the arm (i.e. the grasping part of the biopsy module) touched the bar. The task designed for the experiments was to move the arm of the reconfigurable master device as quickly as possible, touch the bar in CG, and then maintain its position for three seconds. The plane in which the bar stands is shown in grids (Fig.9), and the operator obtains 3D perception by observing these grids. The plane with the grids is the same for all topologies. The angle of the view was set so that the view is similar to that from the camera module in Fig.6. Five subjects (a-e) participated in the experiments, none of whom were surgeons. Each subject was asked to freely move the master device to learn how to operate it; however, this practice was allowed for one minute before starting the experiments. Each subject started from Topology I, then tested Topology II and finally Topology III. The time needed to touch the bar and maintain it for three seconds was measured. This procedure was repeated ten times for each topology with a randomized position of the bar. During the procedure, the bar appeared at random; however, it always appeared in the shared workspace to ensure
- 24. Robotic Systems – Applications, Control and Programming 14 that the arm could reach it. After finishing the experiment, the subjects were asked to fill in a questionnaire (described below) for each topology. The subjects were also asked which topology they preferred. Fig. 8. Workspace of each topology and the shared workspace Fig. 9. Simulator and the master device during one test A NASA TLX questionnaire (NASA TLX (website)) was used to objectively and quantitatively evaluate the workload that the subjects felt during the experiments. This method has versatile uses, and we selected this method also because it was used to evaluate the workload in a tele-surgery environment (Kawamura, K. 2006). This method evaluates Metal Demand, Physical Demand, Temporal Demand, Performance, Effort and Frustration, and gives a score that represents the overall workload that the subject felt during the task. 4.3 Results The time spent conducting the given task, the workload score evaluated using the NASA TLX questionnaire and the preference of the topology determined by the subjects are
- 25. Modular Robotic Approach in Surgical Applications – Wireless Robotic Modules and a Reconfigurable Master Device for Endoluminal Surgery – 15 summarized in Table 1. For each item, a smaller value indicates a more favorable evaluation by the subject. Considering the results of the time and workload score, Topology II was most difficult. The difference between Topology I and III was interesting. Two of the subjects ((b) and (c)) preferred Topology I, which did not have a redundant DOF. Conversely, three of the subjects ((a), (d) and (e)) preferred Topology III because they could select the path to reach the target owing to the redundant DOF. The average scores of the NASA TLX parameters shown in Fig.10 suggest that the Physical Demand workload was high for Topology I, while the Effort workload was high for Topology III. The two subjects who preferred Topology I rather than Topology III claimed that it was not easy to determine where the bar was located when Topology III was used owing to the lack of 3D perception. In addition, they reported that the bar seemed to be placed far from the base. However, the bar appeared randomly, but in the same area; therefore, the bar that appeared in the experiment that employed Topology III was not placed farther from the base when compared to the experiments that used Topology I or Topology II. Accordingly, these two subjects may have had difficulty obtaining 3D perception from the gridded plane. In Topology III, the arm was partially out of view in the initial position; thus, the operator needed to obtain 3D perception by seeing the grids. It is often said that most surgeons can obtain 3D perception even if they use a 2D camera, and our preliminary experimental results imply that this ability might differ by individual. Some people appear to obtain 3D perception primarily by seeing the relative positions between the target and the tool they move. Redundant DOFs may also be preferred by operators with better 3D perception capability. Although the experiments were preliminary, there must be other factors that accounted for the preference of the user. Indeed, it is likely that the preferable topology varies depending on the user, and the developed simulator would be useful to evaluate these variations. The proposed reconfigurable master device will enable individual surgeons to customize the robot and interface as they prefer. Topology Subject Average a b c d e Time (s) I 5.7 4.1 4.0 5.8 5.0 4.9 II 7.6 6.2 4.8 5.5 6.7 6.1 III 4.9 4.3 5.6 4.4 4.7 4.8 Work Load: NASA-TLX Score I 30.7 11.3 28.3 32.0 73.3 35.1 II 47.6 26.7 28.0 53.0 68.3 44.7 III 37.0 5.0 24.3 14.3 61.3 28.4 Preference I 3 1 1 3 3 2.2 II 2 3 2 2 2 2.2 III 1 2 3 1 1 1.6 Table 1. Experimental results
- 26. Robotic Systems – Applications, Control and Programming 16 0 5 10 15 Mental Demand Physical Demand Temporal Demand Performance Effort Frustration Topology I Topology II Topology III Fig. 10. NASA TLX parameters for three topologies 5. Conclusion A modular robot was proposed for endoluminal surgery. The design, prototyping and evaluation of the modules were reported. Although there are some issues related to the fabrication problems, the results of the performance tests show the feasibility of the modular surgical system. A reconfigurable master device has also been proposed, and its feasibility was evaluated by simulation-based experiments. The preliminary results showed that the preferred topology may vary depending on the user. Moreover, the reconfigurable master device would enable each surgeon to customize the surgical system according to his/her own preferences. Development of the robotic modules and the reconfigurable master device provided proof of concept of the modular robotic system for endoluminal surgery, suggesting that the modular approach has great potential for surgical applications. 6. Acknowledgments This study was supported in part by the European Commission in the framework of the ARES Project (Assembling Reconfigurable Endoluminal Surgical System, NEST-2003-1- ADVENTURE/15653), by the European Union Institute in Japan at Waseda University (EUIJ Waseda, http://www.euij-waseda.jp/eng/) within the framework of its Research Scholarship Programme, and by the Global COE Program "Global Robot Academia" from the Ministry of Education, Culture, Sports, Science and Technology of Japan. The authors are grateful to Mr. Nicodemo Funaro for manufacturing the prototypes and Ms. Sara Condino for her invaluable technical support. 7. References Bardaro, S. J. & Swanström, L. (2006). Development of advanced endoscopes for Natural Orifice Transluminal Endoscopic Surgery (NOTES). In: Minim. Invasive Ther. Allied Technol., 15(6), pp. 378–383.
- 27. Modular Robotic Approach in Surgical Applications – Wireless Robotic Modules and a Reconfigurable Master Device for Endoluminal Surgery – 17 Carpi, F. & Pappone, C. (2009). Magnetic maneuvering of endoscopic capsules by means of a robotic navigation system. In: IEEE Trans Biomed Eng, 56(5), pp. 1482-90. Ciuti, G.; Valdastri, P., Menciassi, A. & Dario, P. (2010). Robotic magnetic steering and locomotion of capsule endoscope for diagnostic and surgical endoluminal procedures. In: Robotica, 28, pp. 199-207. Cuschieri, A (2005). Laparoscopic surgery: current status, issues and future developments. In: Surgeon, 3(3), pp. 125–138. Giday, S.; Kantsevoy, S. & Kalloo, A. (2006). Principle and history of natural orifice translumenal endoscopic surgery (notes). In: Minim. Invasive Ther. Allied Technol., 15(6), pp. 373–377. Harada, K.; Susilo, E., Menciassi, A. & Dario, P. (2009). Wireless reconfigurable modules for robotic endoluminal surgery. In: Proc. IEEE Int. Conf. on Robotics and Automation, pp. 2699-2704. Harada, K.; Oetomo, D., Susilo, E., Menciassi, A., Daney, D., Merlet, J.-P. & Dario, P. (2010). A Reconfigurable Modular Robotic Endoluminal Surgical System: Vision and preliminary results, In: Robotica, 28, pp. 171-183. Kawamura, K.; Kobayashi, Y & Fujie, M. G. (2006). Development of Real-Time Simulation for Workload Quantization in Robotic Tele-surgery. In: Proc. IEEE Int. Conf. on Robotics and Biomimetics, pp.1420-25. Menciassi, A.; Valdastri, P., Harada, K. & Dario, P. (2010). Single and Multiple Robotic Capsules for Endoluminal Diagnosis and Surgery. In: Surgical Robotics - System Applications and Visions, Rosen, J., Hannaford, B. & Satava, M., pp. 313-354., Springer-Verlag, 978-1441911254. Moglia, A.; Menciassi, A. Schurr, M. & Dario, P. (2007). Wireless capsule endoscopy: From diagnostic devices to multipurpose robotic systems. In: Biomed Microdevices, 9, pp. 235–243. Murata, S. & Kurokawa, H. (2007). Self-reconfigurable robots. In: IEEE Rob. Autom Mag., 14(1), pp. 71–78. Nagy, Z.; Abbott, J. J. & Nelson, B. J. (2007). The magnetic self-aligning hermaphroditic connector: A scalable approach for modular microrobotics. In: Proc. IEEE/ASME Int. Conf. Advanced Intelligent Mechatronics. pp. 1–6. NASA TLX. available from http://humansystems.arc.nasa.gov/groups/TLX/ Park, S. K.; Koo, K. I., Bang, S. M., Park J. Y., Song, S. Y., & Cho, D. G. (2008). A novel microactuator for microbiopsy in capsular endoscopes. In: J. Micromech. Microeng., 18 (2), 025032. Pesic, M.; Karanikolic, A., Djordjevic, N., Katic, V., Rancic, Z., Radojkovic, M., Ignjatovic N. & Pesic, I. (2004). The importance of primary gastric cancer location in 5-year survival rate. In: Arch Oncol, 12, pp. 51–53. Quirini, M.; Menciassi, A., Scapellato, S., Stefanini, C., and Dario, P. (2008). Design and fabrication of a motor legged capsule for the active exploration of the gastrointestinal tract. In: Proc. IEEE/ASME Trans. Mechatronics, 13, pp. 169–179. Sendoh, M.; Ishiyama, K. & Arai, K. (2003). Fabrication of magnetic actuator for use in a capsule endoscope. In: IEEE Trans. Magnetics, 39(5), pp. 3232–34. Susilo, E.; Valdastri, P., Menciassi, A. & Dario, P. (2009). A miniaturized wireless control platform for robotic capsular endoscopy using advanced pseudokernel approach. In: Sensors and Actuators A, 156(1), pp.49-58.
- 28. Robotic Systems – Applications, Control and Programming 18 World Health Organisation, “Fact sheet n.297,” Available from: http://www.who.int/ mediacen-ter/factsheets/ fs297, 2006. Yim, M.; Shen, W., Salemi, B., Rus, D., Moll, M., Lipson, H., Klavins, E. & Chirikjian, G. (2007). Modular self-reconfigurable robot systems [Grand Challenges of Robotics]. In: IEEE Rob. Autom Mag., 14(1), pp. 865–872.
- 29. 2 Target Point Manipulation Inside a Deformable Object Jadav Das and Nilanjan Sarkar Vanderbilt University, Nashville, TN USA 1. Introduction Target point manipulation inside a deformable object by a robotic system is necessary in many medical and industrial applications such as breast biopsy, drug injection, suturing, precise machining of deformable objects etc. However, this is a challenging problem because of the difficulty of imposing the motion of the internal target point by a finite number of actuation points located at the boundary of the deformable object. In addition, there exist several other important manipulative operations that deal with deformable objects such as whole body manipulation [1], shape changing [2], biomanipulation [3] and tumor manipulation [4] that have practical applications. The main focus of this chapter is the target point manipulation inside a deformable object. For instance, a positioning operation called linking in the manufacturing of seamless garments [5] requires manipulation of internal points of deformable objects. Mating of a flexible part in electric industry also results in the positioning of mated points on the object. In many cases these points cannot be manipulated directly since the points of interest in a mating part is inaccessible because of contact with a mated part. Additionally, in medical field, many diagnostic and therapeutic procedures require accurate needle targeting. In case of needle breast biopsy [4] and prostate cancer brachytherapy [6], needles are used to access a designated area to remove a small amount of tissue or to implant radio-active seed at the targeted area. The deformation causes the target to move away from its original location. To clarify the situation we present a schematic of needle insertion for breast biopsy procedure as shown in Figure 1. When tip of the needle reaches the interface between two different types of tissue, its further insertion will push the tissue, instead of piercing it, causing unwanted deformations. These deformations move the target away from its original location as shown in Figure 1(b). In this case, we cannot manipulate the targeted area directly because it is internal to the organ. It must be manipulated by controlling some other points where forces can be applied as shown in Figure 1(c). Therefore, in some cases one would need to move the positioned points to the desired locations of these deformable objects (e.g., mating two deformable parts for sewing seamlessly) while in other cases one may need to preserve the original target location (e.g., guiding the tumor to fall into the path of needle insertion). In either of these situations, the ability of a robotic system to control the target of the deformable object becomes important, which is the focus of this chapter. To control the position of the internal target point inside a deformable object requires appropriate contact locations on the surface of the object. Therefore, we address the issue of
- 30. Robotic Systems – Applications, Control and Programming 20 determining the optimal contact locations for manipulating a deformable object such that the internal target point can be positioned to the desired location by three robotic fingers using minimum applied forces. A position-based PI controller is developed to control the motion of the robotic fingers such that the robotic fingers apply minimum force on the surface of the object to position the internal target point to the desired location. However, the controller for target position control is non-collocated since the internal target point is not directly actuated by the robotic fingers. It is known in the literature that non-collocated control of a deformable object is not passive, which may lead to instability [7]. In order to protect the object and the robotic fingers from physical damage and also in order to diminish the deterioration of performance caused by unwanted oscillation, it is indispensable to build stable interaction between the robotic fingers and the object. Here we consider that the plant (i.e., the deformable object) is passive and does not generate any energy. So, in order to have stable interaction, it is essential that the controller for the robotic fingers must be stable. Thus, we present a new passivity-based non-collocated controller for the robotic fingers to ensure safe and accurate position control of the internal target point. The passivity theory states that a system is passive if the energy flowing in exceeds the energy flowing out. Creating a passive interface adds the required damping force to make the output energy lower than the input energy. To this end we develop a passivity observer (PO) and a passivity controller (PC) based on [8] for individual robotic finger where PO monitors the net energy flow out of the system and PC will supply the necessary damping force to make PO positive. Our approach extends the concept of PO and PC in [8] to multi- point contacts with the deformable object. Fig. 1. Schematics of needle breast biopsy procedure: (a) needle insertion, (b) target movement, and (c) target manipulation The remainder of this chapter is organized as follows: we discuss various issues and prior research in Section 2. The problem description is stated in Section 3. Section 4 outlines the mathematical modelling of the deformable object. A framework for optimal contact locations is presented in Section 5. The control methods are discussed in Section 6. The effectiveness of the derived control law is demonstrated by simulation in Section 7. Finally, the contributions of this work and the future directions are discussed in Section 8. 2. Issues and prior research A considerable amount of work on multiple robotic systems has been performed during the last few decades [9-11, 12-15]. Mostly, the position and/or force control of multiple manipulators handling a rigid object were studied in [9-11]. However, there were some works on handling deformable object by multiple manipulators presented in [12-15]. Saha (a) (b) (c)
- 31. Target Point Manipulation Inside a Deformable Object 21 and Isto [12] presented a motion planner for manipulating deformable linear objects using two cooperating robotic arms to tie self-knots and knots around simple static objects. Zhang et al. [13] presented a microrobotic system that is capable of picking up and releasing operation of microobjects. Sun et al. [14] presented a cooperation task of controlling the reference motion and the deformation when handling a deformable object by two manipulators. In [15], Tavasoli et al. presented two-time scale control design for trajectory tracking of two cooperating planar rigid robots moving a flexible beam. However, to the best of our knowledge the works on manipulating an internal target point inside a deformable object are rare [4, 5]. Mallapragada et al. [4] developed an external robotic system to position the tumor in image-guided breast biopsy procedures. In their work, three linear actuators manipulate the tissue phantom externally to position an embedded target inline with the needle during insertion. In [5] Hirai et al. developed a robust control law for manipulation of 2D deformable parts using tactile and vision feedback to control the motion of the deformable object with respect to the position of selected reference points. These works are very important to ours present application, but they did not address the optimal locations of the contact points for effecting the desired motion. A wide variety of modeling approaches have been presented in the literature dealing with computer simulation of deformable objects [16]. These are mainly derived from physically- based models to produce physically valid behaviors. Mass-spring models are one of the most common forms of deformable objects. A general mass-spring model consists of a set of point masses connected to its neighbors by massless springs. Mass-spring models have been used extensively in facial animation [17], cloth motion [18] and surgical simulation [19]. Howard and Bekey [20] developed a generalized method to model an elastic object with the connections of springs and dampers. Finite element models have been used in the computer simulation to model facial tissue and predict surgical outcomes [21, 22]. However, the works on controlling an internal point in a deformable object are not attempted. In order to manipulate the target point to the desired location, we must know the appropriate contact locations for effecting the desired motion. There can be an infinite number of possible ways of choosing the contact location based on the object shapes and task to be performed. Appropriate selection of the contact points is an important issue for performing certain tasks. The determination of optimal contact points for rigid object was extensively studied by many researchers with various stability criteria. Salisbury [23] and Kerr [24] discussed that a stable grasp was achieved if and only if the grasp matrix is full row rank. Abel et al. [25] modelled the contact interaction by point contact with Coulomb friction and they stated that optimal grasp has minimum dependency on frictional forces. Cutkosky [26] discussed that the size and shape of the object has less effect on the choice of grasp than by the tasks to be performed after examining a variety of human grasps. Ferrari et al. [27] defined grasp quality to minimize either the maximum value or sum of the finger forces as optimality criteria. Garg and Dutta [28] shown that the internal forces required for grasping deformable objects vary with size of object and finger contact angle. In [29], Watanabe and Yoshikawa investigated optimal contact points on an arbitrary shaped object in 3D using the concept of required external force set. Ding et al. proposed an algorithm for computing form closure grasp on a 3D polyhedral object using local search strategy in [30]. In [31, 32], various concepts and methodologies of robot grasping of rigid objects were reviewed. Cornella et al. [33] presented a mathematical approach to obtain optimal solution of contact points using the dual theorem of nonlinear programming. Saut et al. [34]
- 32. Robotic Systems – Applications, Control and Programming 22 presented a method for solving the grasping force optimization problem of multi-fingered dexterous hand by minimizing a cost function. All these works are based on grasp of rigid objects. There are also a few works based on deformable object grasping. Like Gopalakrishnan and Goldberg [35] proposed a framework for grasping deformable parts in assembly lines based on form closure properties for grasping rigid parts. Foresti and Pellegrino [36] described an automatic way of handling deformable objects using vision technique. The vision system worked along with a hierarchical self-organizing neural network to select proper grasping points in 2D. Wakamatsu et al. [37] analyzed grasping of deformable objects and introduced bounded force closure. However, position control of an internal target point in a deformable object by multi-fingered gripper has not been attempted. In our work, we address the issue of determining the optimal contact locations for manipulating a deformable object such that the internal target point can be positioned to the desired location by three robotic fingers using minimum applied forces. The idea of passivity can be used to guarantee the stable interaction without exact knowledge of model information. Anderson and Spong [38] published the first solid result by passivation of the system using scattering theory. A passivity based impedance control strategy for robotic grasping and manipulation was presented by Stramigioli et al. [39]. Recently, Hannaford and Ryu [40] proposed a time-domain passivity control based on the energy consumption principle. The proposed algorithm did not require any knowledge about the dynamics of the system. They presented a PO and a PC to ensure stability under a wide variety of operating conditions. The PO can measure energy flow in and out of one or more subsystems in real-time by confining their analysis to system with very fast sampling rate. Meanwhile the PC, which is an adaptive dissipation element, absorbs exactly net energy output measured by the PO at each time sample. In [41], a model independent passivity-based approach to guarantee stability of a flexible manipulator with a non- collocated sensor-actuator pair is presented. This technique uses an active damping element to dissipate energy when the system becomes active. In our work we use the similar concept of PO and PC to ensure stable interaction between the robotic fingers and the deformable object. Our work also extends the concept of PO and PC for multi-point contact with the deformable object. 3. Problem description Consider a case in which multiple robotic fingers are manipulating a deformable object in a 2D plane to move an internal target point to a desired location. Before we discuss the design of the control law, we present a result from [42] to determine the number of actuation points required to position the target at an arbitrary location in a 2D plane. The following definitions are given according to the convention in [42]. Manipulation points: are defined as the points that can be manipulated directly by robotic fingers. In our case, the manipulation points are the points where the external robotic fingers apply forces on the deformable object. Positioned points: are defined as the points that should be positioned indirectly by controlling manipulation points appropriately. In our case, the target is the position point. The control law to be designed is non-collocated since the internal target point is not directly actuated by the robotic fingers. The following result is useful in determining the number of actuation points required to accurately position the target at the desired location.
- 33. Target Point Manipulation Inside a Deformable Object 23 Result [42]: The number of manipulated points must be greater than or equal to that of the positioned points in order to realize any arbitrary displacement. In our present case, we assume that the number of positioned points is one, since we are trying to control the position of the target. Hence, ideally the number of contact points would also be one. But practically, we assume that there are two constraints: (1) we do not want to apply shear force on the deformable object to avoid the damage to the surface, and (2) we can only apply control force directed into the deformable object. We cannot pull the surface since the robotic fingers are not attached to the surface. Thus we need to control the position of the target by applying only unidirectional compressive force. In this context, there exists a theorem on the force direction closure in mechanics that helps us determining the equivalent number of compressive forces that can replace one unconstrained force in a 2D plane. Theorem [43]: A set of wrenches w can generate force in any direction if and only if there exists a three-tuple of wrenches 1 2 3 { , , } w w w whose respective force directions 1 f , 2 f , 3 f satisfy: i. Two of the three directions 1 f , 2 f , 3 f are independent ii. A strictly positive combination of the three directions is zero. 1 2 3 0 + + = f f f α β γ (1) where α , β , and γ are constants. The ramification of this theorem for our problem is that we need three control forces distributed around the object such that the end points of their direction vectors draw a non-zero triangle that includes their common origin point. With such an arrangement we can realize any arbitrary displacement of the target point. Thus the problem can be stated as: Problem statement: Given the number of actuation points, the initial target and its desired locations, find appropriate contact locations and control action such that the target point is positioned to its desired location by controlling the boundary points of the object with minimum force. 4. Deformable object modelling Consider a schematic in Figure 2 where three robotic fingers are positioning an internal target (point A) in a deformable object to the desired location (point B). We assume that all the end-effectors of the robotic fingers are in contact with the deformable object such that they can apply only push on the object as needed. The coordinate systems are defined as follows: w is the task coordinate system, o is the object coordinate system, fixed on the object and i is the i-th robotic finger coordinate system, fixed on the i-th end-effectors located at the grasping point. In order to formulate the optimal contact locations, we model the deformable object using mass-spring-damper systems. The point masses are located at the nodal points and a Voigt element [20] is inserted between them. Figure 3 shows a single layer of the deformable object. Each element is labeled as j E for 1,2, , = j NE where NE is total number of elements in a single layer. Position vector of the i-th mesh point is defined as [ ] = p T i i i x y , 1,2,3,..., = i N where, N is total number of point masses. k and c are the spring stiffness and the damping coefficient, respectively. Assume that no moment exerts on each mesh point. Then, the resultant force exerted on the mesh point, pi , can be calculated as
- 34. Robotic Systems – Applications, Control and Programming 24 ∂ = − ∂ w p i i U (2) where, U denotes the total potential energy of the object Fig. 2. Schematic of the robotic fingers manipulating a deformable object Fig. 3. Model of a deformable object with interconnected mass-spring-damper 5. Framework for optimal contact locations We develop an optimization technique that satisfies the force closure condition for three fingers planar grasp. The resultant wrench for the contacts of three robotic fingers is given by 3 1 ( ) = = w n r i i i i f , 2 ( ) ( 0, 1 3) ∀ ∈ℜ ∃ ≥ ≤ ≤ w i f i (3)
- 35. Target Point Manipulation Inside a Deformable Object 25 where, ( ) n r i i is the unit inner normal of i-th contact and i f denotes the i-th finger’s force. We assume that the contact forces should exist in the friction cone to manipulate objects without slip of the fingertip. Now we need to find three distinct points, 1 1 ( ) r θ , 2 2 ( ) r θ , and 3 3 ( ) r θ , on the boundary of the object such that Equation (3) is satisfied. Here, 1 θ , 2 θ , and 3 θ are the three contact point locations measured anti-clockwise with respect to the x axis as shown in Figure 4. In addition, we assume that the normal forces have to be non-negative to avoid separation and slippage at the contact points, i.e., 0 ≥ i f , 1,2,3 = i (4) Fig. 4. Three fingers grasp of a planar object A physically realizable grasping configuration can be achieved if the surface normals at three contact points positively span the plane so that they do not all lie in the same half- plane [44]. Therefore, a realizable grasp can be achieved if the pair-wise angles satisfy the following constraints min max | | ≤ − ≤ i j θ θ θ θ , low high ≤ ≤ i θ θ θ , , 1,2,3 = i j , ≠ i j (5) A unique solution to realizable grasping may not always exist. Therefore, we develop an optimization technique that minimizes the total force applied on to the object to obtain a particular solution. The optimal locations of the contact points would be the solution of the following optimization problem. min f f T sunject to 3 1 ( ) = = w n r i i i i f min max ≤ − ≤ i j θ θ θ θ , , 1,2,3 = i j , ≠ i j 0 ≥ i f , 1,2,3 = i 0 360 ≤ ≤ i θ , 1,2,3 = i (6)
- 36. Robotic Systems – Applications, Control and Programming 26 Once we get the optimal contact locations, all three robotic fingers can be located at their respective positions to effect the desired motion at those contact points. 6. Design of the controller In this section, a control law for the robotic fingers is developed to guide a target from any point A to an arbitrary point B within the deformable object as shown in Figure 2. 6.1 Target position control At any given time-step, point A is the actual location of the target and point B is the desired location of the target. n1, n2 and n3 are unit vectors which determine the direction of force application of the actuation points with respect to the global reference frame w . Let assume, pd is the position vector of point B and p is the position vector of point A. Referring to Figure 2, the position vector of point A is given by [ ] = p T x y (7) where, x and y are the position coordinates of point A in the global reference frame w . The desired target position is represented by point B whose position vector is given by [ ] = p T d d d x y (8) where, d x and d y are the desired target position coordinates. The target position error, e, is given by = − e p p d (9) Once the optimal contact locations are determined from Equation (6), the planner generates the desired reference locations for these contact points by projecting the error vector between the desired and the actual target locations in the direction of the applied forces, which is given by * = ⋅ i e e n i (10) where, [ ] = i n T xi yi n n (11) All robotic fingers are controlled by their individual controllers using the following proportional-integral (PI) control law * * = + e e i Pi i Ii i f K K dt , 1,2,3 = i (12) where, Pi K , and Ii K are the proportional and integral gains. Note that in the control law (12), mechanical properties of the deformable object are not required. Forces applied by the fingers on the surface of the deformable object are calculated by projecting the error vector in the direction of the applied forces. But the Equation (12) does not guarantee that the system will be stable. Thus a passivity-based control approach based on energy monitoring is developed to guarantee the stability of the system.
- 37. Target Point Manipulation Inside a Deformable Object 27 6.2 Passivity-based control A passivity-based control approach based on energy monitoring is developed for deformable object manipulation to guarantee passivity (and consequently stability) of the system. The main reason to use passivity-based control is to ensure stability without the need of having an accurate model of the deformable object. It is not possible to develop a precise dynamic model of a deformable object due to complex nonlinear internal dynamics as well as variation in geometry and mechanical properties. Thus passivity based control is an ideal candidate to ensure stability since it is a model independent technique. The basic idea is to use a PO to monitor the energy generated by the controller and to dissipate the excess energy using a PC when the controller becomes active [41], without the need for modeling the dynamics of the plant (deformable object). Passivity Observer (PO) We develop a network model with PO and PC similar to [41] as shown in Figure 5. The PO monitors the net energy flow of the individual finger’s controller. When the energy becomes negative, PC dissipates excess energy from the individual controller. Similar to [41] energy is defined as the integral of the inner product between conjugate input and output, which may or may not correspond to physical energy. Definition of passivity [41] states that the energy applied to a passive network must be positive for all time. Figure 5 shows a network representation of the energetic behavior of this control system. The block diagram in Figure 5 is partitioned into three elements: the trajectory generator, the controller and the plant. Each controller corresponds to one finger. Since three robotic fingers are used for planar manipulation, three individual controller transfer energy to the plant. The connection between the controller and the plant is a physical interface at which conjugate variables ( i f , i v ; where i f is the force applied by i-th finger and i v is the velocity of i-th finger) define physical energy flow between controller and plant. The forces and velocities are given by T f f f 1 2 3 [ ] = f (13) T v v v 1 2 3 [ ] = v (14) The desired target velocity is obtained by differentiating (8) with respect to time and is given by T d d d x y [ ] = p (15) where, d x and d y are the desired target velocities, respectively. The desired target velocity along the direction of actuation of the i-th robotic finger is given by di d i v = ⋅ p n (16) The trajectory generator essentially computes the desired target velocity along the direction of actuation of the robotic fingers. If the direction of actuation of the robotic fingers, i n , and desired target velocity, d p , are known with respect to a global reference frame then the trajectory generator computes the desired target velocity along the direction of actuation of the fingers using Equation (16).
- 38. Robotic Systems – Applications, Control and Programming 28 The connections between the trajectory generator and the controller, which traditionally consist of a one-way command information flow, are modified by the addition of a virtual feedback of the conjugate variable [41]. For the system shown in Figure 5, output of the trajectory generator is the desired target velocity, di v , along direction of i-th finger and output of the controller is calculated from Equation (12). Fig. 5. Network representation of the control system. 1i α and 2i α are the adjustable damping elements at each port, i=1,2,3 For both connections, virtual feedback is the force applied by the robotic fingers. Integral of the inner product between trajectory generator output ( di v ) and its conjugate variable ( i f ) defines “virtual input energy.” The virtual input energy is generated to give a command to the controller, which transmits the input energy to the plant through the controller in the form of “real output energy.” Real output energy is the physical energy that enters to the plant (deformable object) at the point where the robotic finger is in contact with the object. Therefore, the plant is a three-port system since three fingers manipulate the object. The conjugate pair that represents the power flow is i f , i v (the force and the velocity of i-th finger, respectively). The reason for defining virtual input energy is to transfer the source of energy from the controllers to the trajectory generator. Thus the controllers can be represented as two-ports which characterize energy exchange between the trajectory generator and the plant. Note that the conjugate variables that define power flow are discrete time values and so the analysis is confined to systems having a sampling rate substantially faster than the system dynamics. For regulating the target position during manipulation, 0 = di v . Hence the trajectory generator is passive since it does not generate energy. However, for target tracking, 0 ≠ di v
- 39. Target Point Manipulation Inside a Deformable Object 29 and 0 ≠ i f . Therefore the trajectory generator is not passive because it has a velocity source as a power source. It is shown that even if the system has an active term, stability is guaranteed as long as the active term is not dependent on the system states [45]. Therefore, passivity of the plant and controllers is sufficient to ensure system stability. Here, we consider that the plant is passive. Now we design a PO for sufficiently small time- step ΔT as: 0 ( ) ( ( ) ( ) ( ) ( )) = = Δ − k i k i j di j i j i j j E t T f t v t f t v t (17) where, ΔT is the sampling period and = ×Δ j t j T . In normal passive operation, ( ) i j E t should always be positive. In case when ( ) 0 < i j E t , the PO indicates that the i-th controller is generating energy and going to be active. The sufficient condition to make the whole system passive can be written as 0 0 ( ) ( ) ( ) ( ) = = Δ ≥ Δ k k i j di j i j i j j j T f t v t T f t v t , 0 ∀ ≥ k t , 1,2,3 = i (18) where k means the k-th step sampling time. The active and passive port can be recognized by monitoring the conjugate signal pair of each port in real time. A port is active if 0 < fv that means energy is flowing out of the network system and it is passive if 0 ≥ fv , that means energy is flowing into the network system. The input and output energy can be computed as [46] 1 1 1 ( 1) ( ) ( ) if ( ) ( ) 0 ( ) ( 1) if ( ) ( ) 0 − + > = − ≤ T T i i di i di i T i i di E k f k v k f k v k E k E k f k v k (19) 2 2 2 ( 1) ( ) ( ) if ( ) ( ) 0 ( ) ( 1) if ( ) ( ) 0 − − < = − ≥ T T i i di i di i T i i di E k f k v k f k v k E k E k f k v k (20) 1 1 1 ( 1) ( ) ( ) if ( ) ( ) 0 ( ) ( 1) if ( ) ( ) 0 − − < = − ≥ P P i i i i i i P i i i E k f k v k f k v k E k E k f k v k (21) 2 2 2 ( 1) ( ) ( ) if ( ) ( ) 0 ( ) ( 1) if ( ) ( ) 0 − + > = − ≤ P P i i i i i i P i i i E k f k v k f k v k E k E k f k v k (22) where, 1 ( ) T i E k and 2 ( ) T i E k are the energy flowing in and out at the trajectory side of the controller port, respectively, whereas 1 ( ) P i E k and 2 ( ) P i E k are the energy flowing in and out at the plant side of the controller port, respectively. So the time domain passivity condition is given by 1 1 2 2 ( ) ( ) ( ) ( ) + ≥ + T P T P i i i i E k E k E k E k , 0 ∀ ≥ k (23) Net energy output of an individual controller is given by
- 40. Robotic Systems – Applications, Control and Programming 30 1 2 1 2 2 2 1 2 ( ) ( ) ( ) ( ) ( ) ( 1) ( 1) ( 1) ( 1) = − + − + − − + − − T P P T i i i i i i di i i E k E k E k E k E k k v k k v k α α (24) where, the last two terms are the energy dissipated at the previous time step. 1 ( 1) − i k α and 2 ( 1) − i k α are the damping coefficient calculated based on PO discussed below. Passivity Controller (PC) In order to dissipate excess energy of the controlled system, a damping force should be applied to its moving parts depending on the causality of the port. As it is well known, such a force is a function of the system's velocities giving the physical damping action on the system. Mathematically, the damping force is given by = d f v α (25) where α is the adjustable damping factor and v is the velocity. From this simple observation, it seems necessary to measure and use the velocities of the robotic fingers in the control algorithm in order to enhance the performance by means of controlling the damping forces acting on the systems. On the other hand, velocities measurements are not always available and in these cases position measurements can be used to estimate velocities and therefore to inject damping. When the observed energy becomes negative, the damping coefficient is computed using the following relation (which obeys the constitutive Equation (25)). Therefore, the algorithm used for a 2-port network with impedance causality (i.e., velocity input, force output) at each port is given by the following steps: 1. Two series PCs are designed for several cases as given below: Case 1: If ( ) 0 ≥ i E k , i.e., if the output energy is less than the input energy, there is no need to activate any PCs. Case 2: If ( ) 0 < i E k , i.e., if the output energy is more than the input energy, i.e., 2 1 ( ) ( ) > P T i i E k E k , then we need to activate only the plant side PC. 1 2 2 ( ) 0 ( ) ( )/ ( ) = = − i i i i k k E k v k α α (26) Case 3: Similarly, if ( ) 0 < i E k , 2 1 ( ) ( ) > T P i i E k E k , then we need to activate only the trajectory side PC. 2 1 2 ( ) ( )/ ( ) ( ) 0 = − = i i di i k E k v k k α α (27) 2. The contributions of PCs are converted into power variables as 1 2 ( ) ( ) ( ) ( ) ( ) ( ) = = t i i di p i i i f k k v k f k k v k α α (28) 3. Modified outputs are ( ) ( ) ( ) ( ) ( ) ( ) = + = + T t i i i p P i i i f k f k f k f k f k f k (29)
- 41. Target Point Manipulation Inside a Deformable Object 31 where, ( ) t i f k and ( ) p i f k are the PCs’ outputs at trajectory and plant sides of the controller ports, respectively. ( ) T i f k and ( ) P i f k are the modified outputs at trajectory and plant sides of the controller ports, respectively. 7. Simulation and discussion We perform extensive simulations of positioning an internal target point to a desired location in a deformable object by external robotic fingers to demonstrate the feasibility of the concept. We discretize the deformable object with elements of mass-spring-damper. We choose m=0.006 kg for each point mass, k=10 N/m for spring constant and c=5 Ns/m for damping coefficient. With this parameter set up, we present four different simulation tasks. Task 1: In Task 1, we present the optimal contact locations of various objects using three robotic fingers such that an internal target point is positioned to the desired location with minimum force. The optimal contact locations are computed using Equation (6) as shown in Figures 6 to 8. In these figures, the base of the arrow represents the initial target location and the arrow head denotes the desired location of the target point. The contact locations are depicted by the bold red dots on the periphery of the deformable object. Note that in determining the optimal contact locations, we introduced minimum angle constraints between any two robotic fingers to achieve a physically realizable grasping configuration. 1 2 3 1 2 3 2 3 1 1 2 3 (a) (b) (c) (d) Fig. 6. Optimal contact locations ( 1 θ , 2 θ , 3 θ ): (a) 59.98o, 204.9o, 244.9o, (b) 14.96o, 159.9o, 199.9o, (c) 7.54o, 182.54o, 327.54o, and (d) 48.59o, 88.59o, 234.39o
- 42. Robotic Systems – Applications, Control and Programming 32 1 2 3 1 2 3 1 2 3 1 2 3 (a) (c) (b) (d) Fig. 7. Optimal contact locations ( 1 θ , 2 θ , 3 θ ): (a) 0o, 170o, 253.8o, (b) 29.07o, 116.93o, 233.86o, (c) 0o, 175o, 320o, and (d) 76.93o, 116.93o, 261.94o 2 3 1 3 2 1 1 2 3 2 3 1 (a) (c) (b) (d) Fig. 8. Optimal contact locations ( 1 θ , 2 θ , 3 θ ): (a) 25.18o, 199.48o, 262.22o, (b) 0o, 175o, 262.62o, (c) 141.05o, 303.66o, 343.66o and (d) 96.37o, 169.35o, 288.29o
- 43. Target Point Manipulation Inside a Deformable Object 33 Task 2: In Task 2, we present a target positioning operation when the robotic fingers are not located at their optimal contact locations. For instance, we choose that the robotic fingers are located at 0, 120, and 240 degrees with respect to the x-axis as shown in Figure 9. We assume that the initial position of the target is at the center of the section of the deformable object, i.e., (0, 0) mm. The goal is to position the target at the desired location (5, 5) mm with a smooth Fig. 9. Deformable object with contact points located at 0, 120 and 240 degrees with respect to x-axis Fig. 10. The desired (red dashed) and the actual (blue solid) straight lines when robotic fingers are located at 0, 120, and 240 degrees with respect to x-axis 0 1 2 3 4 5 6 0 1 2 3 4 5 6 x (mm) y (mm) desired actual -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 x (m) y (m) y x 1 2 3
- 44. Robotic Systems – Applications, Control and Programming 34 straight line trajectory. In this simulation, we choose KPi =1000 and KIi =1000, i=1,2,3. Figure 10 shows the actual and desired position trajectories of the target point. It is noticed that there is some error present in the tracking of the desired trajectory. Robotic fingers forces generated by the PI controller are presented in Figure 11 and the POs are falling to negative as shown in Figure 12. Negative values of POs signify that the interaction between the robotic fingers and the deformable object is not stable. Fig. 11. Controller forces when robotic fingers are located at 0, 120, and 240 degrees with respect to x-axis Fig. 12. POs when robotic fingers are located at 0, 120, and 240 degrees with respect to x-axis 0 1 2 3 4 5 6 7 8 9 10 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 time (sec.) Controller forces (N) f1 f2 f3 0 1 2 3 4 5 6 7 8 9 10 -0.01 0 0.01 E 1 (Nm) 0 1 2 3 4 5 6 7 8 9 10 -0.02 -0.01 0 0.01 E 2 (Nm) 0 1 2 3 4 5 6 7 8 9 10 -0.05 0 0.05 E 3 (Nm) time (sec.)
- 45. Target Point Manipulation Inside a Deformable Object 35 Task 3: In Task 3, we consider the same task as discussed above under Task 2 but the robotic fingers are positioned at their optimal contact locations (Figure 8(a)) and the target is following the desired straight line trajectory. In this case, PCs are not turned on while performing the task. A simple position based PI controller generates the control command based on the error between the desired and the actual location of the target. Figure 13 shows that the target Fig. 13. The desired (red dashed) and the actual (blue solid) straight lines when PCs are not turned on Fig. 14. Controller forces when PCs are not turned on 0 1 2 3 4 5 6 7 8 9 10 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 time (sec.) Controller forces (N) f1 f2 f3 0 1 2 3 4 5 6 0 1 2 3 4 5 6 x (mm) y (mm) desired actual
- 46. Robotic Systems – Applications, Control and Programming 36 Fig. 15. (a) POs for three robotic fingers when PCs are not turned on, (b) a magnified version of (a) for first few seconds (a) (b) 0 1 2 3 4 5 6 7 8 9 10 -0.01 0 0.01 E 1 (Nm) 0 1 2 3 4 5 6 7 8 9 10 -5 0 5 x 10 -3 E 2 (Nm) 0 1 2 3 4 5 6 7 8 9 10 -5 0 5 x 10 -3 E 3 (Nm) time (sec.) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -0.01 0 0.01 E 1 (Nm) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -1 0 1 x 10 -6 E 2 (Nm) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -5 0 5 x 10 -7 E 3 (Nm) time (sec.)
- 47. Target Point Manipulation Inside a Deformable Object 37 tracked the desired position trajectory. Robotic fingers forces generated by the PI controller are presented in Figure 14. Force values in Figure 14 are quite less than those in Figure 11 because of the optimal contact location of the robotic fingers. However, the POs for robotic fingers 2 and 3 are become negative as shown in Figure 15. Negative values of the POs signify that the output energy of the 2-port network is greater than the input energy. Since the plant is considered to be passive, the only source of generating extra energy is the controller that makes the whole system unstable. So we must engage passivity controller to modify the controller output by dissipating the extra amount of energy. Task 4: In Task 4, the PCs are turned on and the robotic fingers are commanded to effect the desired motion of the target. The PCs are activated when the POs cross zero from a positive value. The required damping forces are generated to dissipate only the excess amount of energy generated by the controller. In this case, the target tracks the desired straight line trajectory well with the POs remaining positive. Figure 16 represents the actual and the desired trajectories of the target when PCs are turned on. For this case, the PCs on the plant side are only activated whereas the PCs on the trajectory side remain idle. Figure 17 shows the PCs forces generated at the plant side when the POs cross zero. The POs become positive during interaction between the robotic fingers and the object as shown in Figure 18. Hence, the stability of the overall system is guaranteed. The PCs on the trajectory side are shown in Figure 19, which are all zeros. The modified controller outputs to move the target point are shown in Figure 20. Fig. 16. The desired (red dashed) and the actual (blue solid) straight lines when PCs are turned on 0 1 2 3 4 5 6 0 1 2 3 4 5 6 x (mm) y (mm) desired actual
- 48. Robotic Systems – Applications, Control and Programming 38 Fig. 17. Required forces supplied by PCs at the plant side when PCs are turned on Fig. 18. POs for three robotic fingers when PCs are turned on 0 1 2 3 4 5 6 7 8 9 10 -0.01 0 0.01 E 1 (Nm) 0 1 2 3 4 5 6 7 8 9 10 -0.1 0 0.1 0.2 E 2 (Nm) 0 1 2 3 4 5 6 7 8 9 10 -0.01 0 0.01 0.02 E 3 (Nm) time (sec.) 0 1 2 3 4 5 6 7 8 9 10 -0.1 0 0.1 f 1 p (N) 0 1 2 3 4 5 6 7 8 9 10 0 0.2 0.4 f 2 p (N) 0 1 2 3 4 5 6 7 8 9 10 0 0.2 0.4 f 3 p (N) time (sec.)
- 49. Target Point Manipulation Inside a Deformable Object 39 Fig. 19. PCs forces at the trajectory side when PCs are turned on Fig. 20. Modified controller forces when PCs are turned on 0 1 2 3 4 5 6 7 8 9 10 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 time (sec.) Modified controller forces (N) f1 P f2 P f3 P 0 1 2 3 4 5 6 7 8 9 10 -1 0 1 f 1 t (N) 0 1 2 3 4 5 6 7 8 9 10 -1 0 1 f 2 t (N) 0 1 2 3 4 5 6 7 8 9 10 -1 0 1 f 3 t (N) time (sec.)
- 50. Robotic Systems – Applications, Control and Programming 40 8. Conclusion and future work In this chapter, an optimal contact formulation and a control action are presented in which a deformable object is manipulated externally by three robotic fingers such that an internal target point is positioned to the desired location. First, we formulated an optimization technique to determine the contact locations around the periphery of the object so that the target can be manipulated with minimum force applied on the object. The optimization technique considers a model of the deformable object. However, it is difficult to build an exact model of the deformable object in general due to nonlinear elasticity, friction, parameter variations, and other uncertainties. Therefore, we considered a coarse model of the deformable object to determine the optimal contact locations which is more realizable. A time-domain passivity control scheme with adjustable dissipative elements has been developed to guarantee the stability of the whole system. Extensive simulation results validate the optimal contact formulation and stable interaction between the robotic fingers and the object. 9. References [1] D. Sun and Y. H. Liu, Modeling and impedance control of a two-manipulator system handling a flexible beam, ASME Journal of Dynamic Systems, Measurement, and Control, vol. 119, pp. 736-742, 1997. [2] P. Dang, F. L. Lewis, K. Subbarao and H. Stephanou, Shape control of flexible structure using potential field method, 17th IEEE International Conference on Control Applications, Texas, USA, pp. 540-546, 2008. [3] X. Liu, K. Kim, Y. Zhang and Y. Sun, Nanonewton force sensing and control in microrobotic cell manipulation, The International Journal of Robotics Research, vol. 28, issue 8, pp. 1065-1076, 2009. [4] V. G. Mallapragada, N. Sarkar and T. K. Podder, Robot-assisted real-time tumor manipulation for breast biopsy, IEEE Transactions on Robotics, vol. 25, issue 2, pp. 316-324, 2009. [5] S. Hirai and T. Wada, Indirect simultaneous positioning of deformable objects with multi pinching fingers based on uncertain model, Robotica, Millennium Issue on Grasping and Manipulation, vol. 18, pp. 3-11, 2000. [6] S. Nath, Z. Chen, N. Yue, S. Trumpore and R. Peschel, Dosimetric effects of needle divergence in prostate seed implant using 125I and 103Pd radioactive seeds, Medical Physics, vol. 27, pp. 1058-1066, 2000. [7] A. Albu-Schaffer, C. Ott and G. Hirzinger, Constructive energy shaping based impedance control for a class of underactuated Euler-Lagrange systems, IEEE International Conference on Robotics and Automation, pp. 1387-1393, 2005. [8] B. Hannaford and R. Jee-Hwan, Time-domain passivity control of haptic interfaces, IEEE Transactions on Robotics and Automation, vol. 18, pp. 1-10, 2002. [9] S. Ali A. Moosavian and R. Rastegari, Multiple-arm space free-flying robots for manipulating objects with force tracking restrictions, Robotics and Autonomous System, vol. 54, issue 10, pp. 779-788, 2006. [10]Z. Li, S. S. Ge and Z. Wang, Robust adaptive control of coordinated multiple mobile manipulators, Mechatronics, vol. 18, issues 5-6, pp. 239-250, 2008.
- 51. Target Point Manipulation Inside a Deformable Object 41 [11]M. Namvar and F. Aghili, Adaptive force-motion control of coordinated robots interacting with geometrically unknown environments, IEEE Transactions on Robotics, vol. 21, issue 4, pp. 678-694, 2005. [12]M. Saha and P. Isto, Manipulation planning for deformable linear objects, IEEE Transactions on Robotics, vol. 23, issue 6, pp. 1141-1150, 2007. [13]Y. Zhang, B. K. Chen, X. Liu and Y. Sun, Autonomous robotic pick-and-place of microobjects, IEEE Transactions on Robotics, vol. 26, issue 1, pp. 200-207, 2010. [14]D. Sun and Y. H. Liu, Modeling and impedance control of a two-manipulator system handling a flexible beam, ASME Journal of Dynamic Systems, Measurement, and Control, vol. 119, no. 4, pp. 736-742 ,1997. [15]A. Tavasoli, M. Eghtesad and H. Jafarian, Two-time scale control and observer design for trajectory tracking of two cooperating robot manipulators moving a flexible beam, Robotics and Autonomous Systems, vol. 57, issue 2, pp. 212-221, 2009. [16]S. F. F. Gibson and B. Mirtich, A survey of deformable modeling in computer graphics, MERL Technical Report, TR97-19, 1997. [17]Y. Zhang, E. C. Prakash and E. Sung, A new physical model with multilayer architecture for facial expression animation using dynamics adaptive mesh, IEEE Transactions on Visualization and Computer Graphics, vol. 10, issue 3, pp. 339-352, 2004. [18]M. Meyer, G. Debunne, M. Desbrun and A. H. Barr, Interactive animation of cloth-like objects in virtual reality, The Journal of Visualization and Computer Animation, vol. 12, no. 1, pp. 1-12(12), 2001. [19]A. Joukhadar and C. Laugier, Dynamic simulation: model, basic algorithms, and optimization, In J.-P. Laumond and M. Overmars, editors, Algorithms for Robotic Motion and Manipulation, pp. 419-434, A. K. Peters Ltd., 1997. [20]A. M. Howard, and G. A. Bekey, Recursive learning for deformable object manipulation, Proc. of International Conference on Advanced Robotics, pp. 939-944, 1997. [21]R. M. Koch, M. H. Gross, F. R. Carls, D. F. Von Buren, G. Fankhauser, Y. I. H. Parish, Simulating facial surgery using finite element models, In ACM SIGGRAPH 96 Conf. Proc., pp. 421-428, 1996. [22]S. D. Pieper, D. R. Laub, Jr., and J. M. Rosen, A finite element facial model for simulating plastic surgery, Plastic and Reconstructive Surgery, 96(5), pp. 1100-1105, Oct. 1995. [23]J. K. Salisbury, Kinematic and force analysis of articulated hands, PhD Thesis, Department of Mechanical Engineering, Standford University, Standford, CA, 1982. [24]J. Kerr, Analysis of multifingered hand, PhD Thesis, Department of Mechanical Engineering, Standford University, Standford, CA, 1984. [25]J. M. Abel, W. Holzmann and J. M. McCarthy, On grasping planar objects with two articulated fingers, IEEE Journal of Robotics and Automation, vol. 1, pp. 211-214, 1985. [26]M. R. Cutkosky, Grasping and fine manipulation for automated manufacturing, PhD Thesis, Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA, 1985. [27]C. Ferrari and J. Canny, Planning optimal grasp, IEEE International Conference on Robotics and Automation, pp. 2290-2295, 1992. [28]S. Garg and A. Dutta, Grasping and manipulation of deformable objects based on internal force requirements, International Journal of Advanced Robotic Systems, Vol. 3, No. 2, pp. 107-114, 2006.
- 52. Robotic Systems – Applications, Control and Programming 42 [29]T. Watanabe and T. Yoshikawa, Grasping optimization using a required external force set, IEEE Transactions on Automation Science and Engineering, Vol. 4, pp. 52-66, 2007. [30]D. Ding, Y. H. Liu and S. Wang, Compuattion of 3D form-closure grasps, IEEE Transactions on Robotics and Autonamtion, Vol. 17, pp. 515-522, 2001. [31]A. Bicchi and V. Kumar, Robotic grasping and contact: a review, Proc. IEEE International Conference on Robotics and Automation, pp. 348-353, 2000. [32]M. T. Mason, Mechanics of robotic manipulation, The MIT Press. [33]J. Cornella, R. Suarez, R. Carloni and C. Melchiorri, Dual programming based approach for optimal grasping force distribution, Journal of Mechatronics, Vol. 18, pp. 348- 356, 2008. [34]P. Saut, C. Remond, V. Perdereau and M. Drouin, Online computation of grasping force in multi-fingered hands, Proc. IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 1223-1228, 2005. [35]K. Gopalakrishanan and K. Goldberg, D-space and deform closure grasps of deformable parts, International Journal of Robotics Research, vol. 24, pp. 889-910, 2005. [36]G. L. Foresti and F. A. Pellegrino, Automatic visual recognition of deformable objects for grasping and manipulation, IEEE Transaction on Systems, Man, and Cybernetics: Applications and Reviews, vol. 34, pp. 325-333, 2004. [37]H. Wakamatsu, S. Hirai and K. Iwata, Static analysis of deformable object grasping based on bounded force closure, IEEE International Conference on Robotics and Automation, vol.4, pp. 3324-3329, 1996. [38]R. J. Anderson and M. W. Spong, Asymptotic stability for force reflecting teleoperators with time delays, in, 1989. IEEE International Conference on Robotics and Automation, vol.3, pp. 1618-1625, 1989. [39]S. Stramigioli, C. Melchiorri and S. Andreotti, A passivity-based control scheme for robotic grasping and manipulation, The 38th Conference on Decision and Control, Phoenix, Arizona, USA, 1999. [40]B. Hannaford and J.-H. Ryu, Time domain passivity control of haptic interfaces, IEEE International Conference on Robotics and Automation, vol.2, pp. 1863-1869, 2001. [41]J.-H. Ryu, D.-S. Kwon and B. Hannaford, Control of a flexible manipulator with noncollocated feedback: time-domain passivity approach, IEEE Transactions on Robotics, vol. 20, pp. 776-780, 2004. [42]T. Wada, S. Hirai, S. Kawamura and N. Kamiji, Robust manipulation of deformable objects by a simple PID feedback, IEEE International Conference on Robotics and Automation, vol.1, pp. 85-90, 2001. [43]V. D. Nguyen, Constructing force-closure grasps, IEEE International Conference on Robotics and Automation, pp. 1368-1373, 1986. [44]J. Ponce and B. Faverjon, On computing three-finger force-closure grasps of polygonal objects, Fifth International Conference on Advanced Robotics, vol.2, pp. 1018-1023, 1991. [45]J. E. Colgate and N. Hogan, Robust control of dynamically interacting systems, International Journal of Control, vol. 48, pp. 65 - 88, 1988. [46]J. J.-H. Ryu and C. Preusche, Stable bilateral control of teleoperators under time-varying communication delay: time domain passivity approach, IEEE International Conference on Robotics and Automation, pp. 3508-3513, 2007.
- 53. 3 Novel Assistive Robot for Self-Feeding Won-Kyung Song and Jongbae Kim Korea National Rehabilitation Research Institute, Korea National Rehabilitation Center Korea 1. Introduction Assistive robots, with which users can interact directly, have attracted worldwide attention. They can assist people with disabilities and older persons in the activities of daily living. Assistive robots could be employed for improving quality of life as they can be adjusted according to demographic changes. There are several crucial issues to be considered with regard to these robots, such as customizing them according to the specific culture of the users as well as ensuring cost-effectiveness (Mann, 2005). In Korea, the official number of registered people with disabilities due to illnesses, injuries, and the natural aging process has already exceeded two million (Employment Development Institute, 2009). More than one-third of these disabled people are the elderly. Moreover, due to longer life spans and a decline in birthrate, the elderly make up over 10% of the population in Korea. As a result, effective caregiving with restricted resources is an urgent problem. In order to achieve efficient caregiving for people with disabilities and elderly persons, caregivers should physically interact with the people. For example, caregivers have to assist people in performing the routine activities of their daily lives, such as eating, changing clothes, changing their posture, moving from one location to another, and bathing. Among these activities, eating meals is one of the most essential daily activities. In this regard, caregivers must interact with people frequently to assist with food selection, feeding interval, etc. Existing robotic technologies can be utilized to take over the functions of the caregivers. Thus, assistive robots represent one of the solutions by which disabled or elderly people can receive support for performing the activities of daily life. The design of assistive robots to help with self-feeding depends strongly on the specific culture of the user. Korean food consists chiefly of boiled rice, soup, and side dishes such as Kimchi. The procedure of having a meal is as follows: the user eats the boiled rice first and then the side dishes. These steps are performed repetitively. In comparison with foreign boiled rice, Korean boiled rice sticks together very well after cooking. Handling this sticky boiled rice can be problematic. In addition, Korean soup includes meat, noodles, and various vegetables, thus the existing feeding robots find it difficult to handle Korean foods. Various assistive robots have been developed since the late 1980s, as shown in Fig. 1. Handy1 (Topping & Smith, 1999) is an assistive robot for daily activities such as eating, drinking, washing, shaving, teeth cleaning, and applying make-up. Handy1 consists of a five-DOF (degree-of-freedom) robot, a gripper, and a tray unit. The major function of Handy1 is to help with eating. Handy1 allows a user to select food from any part of the tray. A cup is attached to enable users to drink water with their meal. The walled columns of a
- 54. Robotic Systems – Applications, Control and Programming 44 food dish serve an important purpose: the food can be scooped on to the dish without any resultant mixing of food items. (a) Handy1 (b) Winsford feeder (c) Neater Eater (d) My Spoon (e) Meal Buddy (f) Mealtime Partner Dining System Fig. 1. Feeding systems The Winsford feeder (Sammons Preston; Hermann et al., 1999) is a mechanical self-feeding system. It uses a mechanical pusher to fill a spoon and a pivoting arm to raise the spoon to the user’s mouth that is at a preset position. The plate is rotated to place more food in front of the pusher. The user can choose from two input devices: a chin switch and a rocker switch. Neater Eater (Neater Solutions) has two versions: a manual–operation-type and an automatic-operation-type system. Neater Eater consists of a two-DOF arm and one dish. Two types of food can be present on the one dish. The manual type can be used to suppress the tremors of a user’s upper limbs while he or she eats. My Spoon (Soyama et al., 2003) is suitable in the case of Japanese food. It consists of a five- DOF manipulator, a gripper, and a meal tray. The meal tray has four rectangular cells. My Spoon combines several pre-programmed motions: automatic operation, semiautomatic operation, and manual operation. The semiautomatic operation allows a user to select food. The manual operation can change the position in which the food is held. The input device can be selected from among the following: the chin joystick, reinforcement joystick, and switch. The end-effector of the robotic arm has one spoon and one fork, which move together to realize the grasping motion. During the grasping process, the gap between the spoon and the fork changes and thus the end-effector grasps the food. Then the robot moves to a predefined position in front of the user’s mouth, and the fork moves backward to enable the user to eat the food off the spoon. Meal Buddy (Sammons Preston) has a three-DOF robotic arm and three bowls that can be mounted on a board using magnets. After the system scoops the food, the robotic arm scrapes the surplus food off the spoon with the rod on the bowls.
- 55. Another Random Scribd Document with Unrelated Content
- 56. quick audience) in the open square there, near the end of the street where Mr. Sly’s King’s Arms and Royal Hotel stands.” “Doctor Marigold” was published in 1865, seven years after Dickens’s visit. But he not only remembered the King’s Arms, but also Mr. Sly, the proprietor, who thus became immortalised in a Dickens story. Mr. Sly evidently was a popular man in the town, and his energy and good nature were much appreciated. That this was so, the following paragraph bears witness: It is recorded as an historical fact that, on the marriage of H.R.H. the Prince of Wales, the demonstration made in Lancaster exceeded any held out of the Metropolis. The credit of this success is mainly due to Mr. Sly, who proposed the programme, which included the roasting of two oxen whole, and a grotesque torchlight procession. The manner in which the whole arrangements were carried out was so satisfactory to the inhabitants of the town and neighbourhood that, at a meeting held a short time after the event, it was unanimously resolved to present Mr. and Mrs. Sly with a piece of plate, of a design suitable to commemorate the event. The sum required was subscribed in a few days, the piece of plate procured, and the presentation was made in the Assembly Rooms on the 9th of November by the High Sheriff, W. A. F. Saunders, Esq., of Wennington Hall, in the presence of a numerous company. In its palmy days the King’s Arms was a prominent landmark for travellers en route to Morecambe Bay, Windermere, the Lakes, and Scotland. It was erected in 1625, and in the coaching era was the head hotel in the town for general posting purposes, and was the most suitable place for tourists to break their journey going North, or in returning. Consequently, it was one of the most important in the North of England. The inn the two idle apprentices entered at Hesket Newmarket “to drink whiskey and eat oat-cake” is not named, but it has been identified with a house which is no longer an inn. At the time of the story it was called the Queen’s Head, and was quite a prominent
- 57. hostelry in the town, the innkeeper of which is described as having “a ruddy cheek, a bright eye, a well-knit frame, an immense hand, a cheery, outspeaking voice, and a straight, bright, broad look. He had a drawing-room, too, upstairs, which was worth a visit to the Cumberland Fells.” “The ceiling of this drawing-room,” we are further told, “was so crossed and re-crossed by beams of unequal lengths, radiating from a centre, in a corner, that it looked like a broken starfish. The room was comfortably and solidly furnished with good mahogany and horsehair. It had a snug fireside, a couple of well-curtained windows, looking out upon the wild country behind the house. What it most developed was an unexpected taste for little ornaments and knick- knacks, of which it contained a most surprising number,” which Dickens goes on to describe in his own whimsical manner. Hesket has not altered very much, we understand, since those days, and the inn itself remains, not as an inn, but as a private house, and the room where the oat-cake and whiskey were served still has its crossed and re-crossed beams of unequal length. From this inn, and under the guidance of the landlord, the two idle apprentices mounted Carrock—with what disastrous effects to Mr. Idle on the way down, readers of the story well know. On again reaching the inn, under uncomfortable circumstances, they remained only a few hours, and continued the tour to Wigton in a covered carriage. Here, Mr. Idle was “melodramatically carried to the inn’s first floor and laid upon three chairs.” The King’s Arms is said to be the Wigton inn referred to, but no details are given of it in the book. Their next halting place was Allonby, where they put up at the Ship. Thomas Idle, we are informed, “made a crab-like progress up a clean little bulk-headed staircase, into a clean little bulk-headed room, where he slowly deposited himself on a sofa, with a stick on either hand of him, looking exceedingly grim,” and both partook of
- 58. dinner. The little inn is described as delightful, “excellently kept by the most comfortable of landladies and the most attentive of landlords.” It still exists, and, “as a family and commercial hotel and posting-house commanding extensive views of the Solway Firth and the Scottish Hills,” is apparently little altered since Dickens and Collins visited it. Its Dickensian associations are cherished by the owner to-day, who shows with pride the room occupied by the two literary giants. After their visit to Lancaster, already referred to, the two idle apprentices went on to Doncaster, and arrived there in the St. Leger Race week. They put up at the Angel Hotel, where they had secured rooms, which Dickens described as “very good, clean and quiet apartments on the second floor, looking down into the main street.” His own room was “airy and clean, little dressing-room attached, eight water-jugs ... capital sponge bath, perfect arrangement, exquisite neatness.” Doncaster during the race week is described as a collection of mad people under the charge of a body of designing keepers, horse-mad, betting-mad, vice-mad. But the two novelists managed to find it enticing enough to remain there a week. The Angel Hotel was often called the Royal because Queen Victoria stayed there in 1851. It was built in 1810, has always been a celebrated hotel, and was a busy coaching-inn in those days. It remains much as it was when Thomas Idle lay in the room for a week with his bad ankle and his friend Francis Goodchild went roaming around the city with his usual observant eyes.
- 59. I CHAPTER XIV Sketches by Boz and The Uncommercial Traveller THE GOAT AND BOOTS—THE BLUE LION AND STOMACH WARMER— THE RED HOUSE—THE FREEMASONS’ TAVERN—THE EAGLE— OFFLEY’S—THE RAINBOW—THE ALBION—THE FLOWER-POT—THE BULL’S HEAD—THE DOLPHIN’S HEAD—THE LORD WARDEN HOTEL— THE CRISPIN AND CRISPIANUS N Dickens’s minor writings there are mentioned many inns, taverns and coffee-houses, some merely fictitious with fanciful names, others whose fame has been recorded in the social history of their times. Sketches by Boz is fairly well supplied in this respect, but none of them is described at any length; indeed, scarcely anything but the names are mentioned, and those only in passing. In the second chapter of “Our Parish,” we are introduced to the new curate who became so popular with the ladies that their enthusiasm for him knew no bounds. It culminated, we are told, when “he spoke for one hour and twenty-five minutes at an anti-slavery meeting at the Goat and Boots.” A proposal was forthwith set on foot to make him a presentation, and this, in the shape of a splendid silver ink-stand engraved with an appropriate inscription, was publicly handed to him at a special breakfast at the aforementioned Goat and Boots, “in a neat little speech by Mr. Gubbins, the ex-churchwarden, and acknowledged by the curate in terms which drew tears into the eyes of all present—the very waiters melted.” The Goat and Boots was no doubt a highly respectable hostelry, but its whereabouts is “wropped in mystery.” So is the Blue Lion and Stomach Warmer, except that we are told that it was at Great
- 60. Winglebury, and we know that Great Winglebury was a fictitious name for Rochester. But which was the inn that received this whimsical name at the hands of the novelist under whose roof Horace Hunter penned his challenge to that base umbrella-maker Alexander Trott, we are unable to state. On the other hand, the Winglebury Arms where Alexander Trott was staying at the time was the Bull Hotel, Rochester.[3] The Red House, Battersea, casually mentioned in the chapter on “The River” as the “Red-us,” was a popular tavern and tea-gardens in those days and notorious for its pigeon-shooting; indeed, tradition has it that it took the lead in the quality and quantity of the sport, and that the crack shots assembled there to determine important matches. It was also famous as the winning-post of many a boat race from Westminster Bridge, and was the place “where all the prime of life lads assembled,” the joy and fun of which is vividly described by Dickens in the chapter referred to. It was a red-bricked building, and a prominent landmark of what was then known as Battersea Fields, the one-time scene of many a duel. The Cross Keys mentioned in the chapter on “Omnibuses” we have already referred to when dealing with Great Expectations; whilst for particulars of the Golden Cross, the busy coaching-inn mentioned in “Hackney Coach Stands,” and in “Early Coaches,” we must refer the reader to “The Inns and Taverns of Pickwick.” The Freemasons’ Tavern in the chapter on “Public Dinners” does not receive much attention from Dickens. He is describing the public dinner given in aid of the “Indigent Orphans Friends’ Benevolent Institution,” and no reference beyond the use of the name is made to the building itself. The tavern still stands to-day, and no doubt more glorious in its splendour than it was on the occasion of the public dinner Dickens refers to. It is used to-day for similar purposes, the ceremony and atmosphere at which being little changed from what it was then. It is interesting to note that in the same building a farewell dinner was given Dickens on the eve of his departure for America in 1867, with Lord Lytton in the chair.
- 61. The chapter devoted to the story of Miss Evans and the Eagle, recalls the notorious tavern immortalised in the famous jingle: Up and down the City Road, In and out the Eagle, That’s the way the money goes— Pop goes the weasel! and the chronicle of Miss Jemima Evans’s visit to the highly famed pleasure-resort will contribute more towards retaining the Eagle on the recording tablets of history than the contemporary rhymster’s poetic effort. It was in 1825 that the Eagle Tavern turned its saloon into what was the forerunner of the music hall, and was the making of many a well-known singer. It was to this gay spot in London that Mr. Samuel Wilkins took Miss Jemima Evans, with whom he “kept company.” They were joined in the Pancras Road by Miss Ivins’s lady friend and her young man. We do not attempt to identify the Crown where they stayed on the way to taste some stout, and are content with the knowledge that they reached the rotunda where the concert was held, and to remind our readers of the impression it had on Miss J’mima Ivins and Miss J’mima Ivins’s friend, who both exclaimed at once “How ’ev’nly!” when they were fairly inside the gardens. Dickens’s description of the place will convey some idea of its splendour: “There were the walks, beautifully gravelled and planted—and the refreshment boxes, painted and ornamented like so many snuff- boxes—and the variegated lamps shedding their rich light upon the company’s heads—and the place for dancing ready chalked for the company’s feet—and a Moorish band playing at one end of the gardens—and an opposition military band playing away at the other. Then, the waiters were rushing to and fro with glasses of negus, and glasses of brandy and water, and bottles of ale, and bottles of stout; and ginger-beer was going off in one place, and practical jokes were going off in another; and people were crowding to the door of the Rotunda; and, in short, the whole scene was, as Miss J’mima Ivins, inspired by the novelty, or the stout, or both, observed, ‘One of
- 62. dazzling excitement.’ As to the concert room, never was anything half so splendid. There was an orchestra for the singers, all paint, gilding, and plate-glass; and such an organ!... The audience was seated on elevated benches round the room, and crowded into every part of it; and everybody was eating and drinking as comfortably as possible.” THE EAGLE TAVERN PLEASURE GARDENS From an old Print What happened to our friends there, and how the trouble over the waistcoat and whiskers was adjusted, is not our business here. The printed account must be read elsewhere. But we have quoted what is perhaps one of the best pictures of this famous resort extant.
- 63. Ultimately, the Rotunda was turned into the Grecian Theatre, and was not demolished until 1901. By then, of course, the real glory of the Eagle had departed and succeeding generations of Jemima Evanses and their young men friends had sought other glittering palaces for their pleasures. There are two taverns mentioned in the following paragraph appearing in the chapter on Mr. John Dounce: “There was once a fine collection of old boys to be seen round the circular table at Offley’s every night, between the hours of half-past eight and half-past eleven. We have lost sight of them for some time. There were, and may be still for aught we know, two splendid specimens in full blossom at the Rainbow Tavern in Fleet Street, who always used to sit in the box nearest the fire-place, and smoked long cherry-stick pipes which went under the table with the bowls resting on the floor.” Offley’s, long ago demolished, was a noted tavern in its day, and, according to Timbs, enjoyed great and deserved celebrity, though short-lived. It was situated at No. 23 Henrietta Street, Covent Garden, and its fame rested on Burton ale and the largest supper- room in the neighbourhood. It had a certain dignity about it, and eschewed “pictures, placards, paper-hangings, or vulgar coffee-room finery,” in order that its customers should not be disturbed in their relish of the good things provided. Of these good things may be mentioned Offley’s chop, which was thick and substantial. The House of Commons chop was small and thin, and Honourable Members sometimes ate a dozen at a sitting. “Offley’s chop was served with shalots shred and warmed in gravy, and accompanied by nips of Burton ale, and was a delicious after-theatre supper.” There was a large room upstairs with wines really worth drinking, and withal Offley’s presented a sort of quakerly plainness, but solid comfort. There was singing by amateurs one day a week, and, to prevent the chorus waking the dead in their cerements in St. Paul’s churchyard opposite, the coffee-room window was double.
- 64. Upon other evenings, there came to a large round table (a sort of privileged place) a few well-to-do, substantial tradesmen from the neighbourhood, and this was the little coterie to which Dickens refers. The Rainbow, also mentioned in the quotation above, was the second house in London to sell coffee and was at one time kept by a Mr. Farr, who was prosecuted for the nuisance caused by the odious smell in the roasting of the berry. In later years (about 1780) the tavern was kept by Alexander Moncrieff, grandfather of the author of “Tom and Jerry,” and was known as the Rainbow Coffee-House. In those days the coffee-room had a lofty bay-window at the south end, looking into the Temple; the room was separated from the kitchen only by a glazed partition. In the bay was a table for the elders, amongst whom doubtless were the “grand old boys” Dickens speaks of as being always there, puffing and drinking away in great state. Everybody knew them, and it was supposed by some people that they were both “immortal.” In the chapter “Making a Night of It,” we learn that Mr. Potter, in his “rough blue coat with wooden buttons, made upon the fireman’s principle, in which, with the addition of a low-crowned, flower-pot, saucer-shaped hat,” created no inconsiderable sensation at the Albion in Little Russell Street, and divers other places of public and fashionable resort. “Making a Night of It” is no doubt mainly reminiscent of a merry evening in the business life of Dickens, and possibly the Albion was one of the favourite resorts of his, and of his co-clerk, Potter. In their day, the Albion was favoured by the theatrical profession and all those associated with things theatrical, and also by those young men who hung on the skirts of actors. Dickens used the Albion in the ’fifties. In a letter to W. H. Wills (1851) there are instructions to order a plain cold supper at Simpson’s, the Albion, by Drury Lane Theatre, for the next play night. “I would merely have cold joints, lobsters, salad, and plenty of
- 65. clean ice,” he says. “Perhaps there might be one hot dish, as broiled bones. But I would have only one, and I would have it cheap.” The play referred to was “Not so Bad as we Seem,” which Dickens and his friends were rehearsing for the Guild of Literature and Art. The supper was to be paid for at so much per head, “not including wines, spirits or beers, which each gentleman will order for himself.” Mr. Percy FitzGerald tells of another evening when Dickens took his friends to the Albion. It was the occasion of Hollingshead’s revival of “The Miller and his Men,” and Dickens was determined to be there. He gave a little dinner party at “the good old Albion,” and all were in great spirits, seated in one of the “boxes” or eating pews as they might be called, and then crossed over the Drury Lane Theatre afterwards. In the chapter devoted to “Mr. Minns and his Cousin,” in giving instructions as to the best way for Mr. Augustus Minns to get to Mr. Budden’s in Poplar Walk, the latter says, “Now mind the direction; the coach goes from the Flower Pot in Bishopsgate Street, every half-hour. When the coach stops at the Swan, you’ll see, immediately opposite you, a White House.” The Flower Pot was a coaching inn of some distinction in those days, for not only did the coaches ply between it and the north-east of London, but the inn was also the starting point of the Norwich coach and others to the eastern counties. The Swan was at Stamford Hill, and, beyond that it was the scheduled stopping-place for coaches, to and from London, we can find no record of its history. The innumerable references to inns and taverns in The Uncommercial Traveller are for the most part purely imaginary. Even when it is clear that Dickens is describing something he actually saw and experienced, he has taken the precaution, in this book, to disguise the inn’s name and whereabouts. There are several such in the chapter entitled “Refreshments for Travellers,” a chapter made up of a series of complaints and adverse criticisms verging on the brink of libel. For instance:
- 66. “Take the old-established Bull’s Head with its old-established knife- boxes on its old-established sideboards, its old-established flue under its old-established four-post bedsteads in its old-established airless rooms, its old-established frouziness upstairs and downstairs, its old-established cookery, and its old-established principles of plunder. Count up your injuries, in its side-dishes of ailing sweetbreads in white poultices, of apothecaries’ powders in rice for curry, of pale stewed bits of calf ineffectually relying for an adventitious interest on forcemeat balls. You have had experience of the old-established Bull’s Head stringy fowls, with lower extremities like wooden legs sticking up out of the dish; of its cannibalistic boiled mutton, gushing horribly among its capers, when carved; of its little dishes of pastry—roofs of spermaceti ointment erected over half an apple or four gooseberries. Well for you if you have yet forgotten the old-established Bull’s Head fruity port; whose reputation was gained solely by the old-established price the Bull’s Head put upon it, and by the old-established air with which the Bull’s Head set the glasses and d’oyleys on, and held that Liquid Gout to the three-and-sixpenny wax candle, as if its old-established colour hadn’t come from the dyers.” Had that inn been properly named at the time, the proprietor’s ire would have been raised, with serious consequences. Then there is the chapter on “An Old Stage-Coaching House,” whose title seemed to augur well for our purpose. Yet, although it is interesting as picturing the decay of coaching and how it resulted on a coaching town, there is nothing by which we can fix the name of the town, and so identify the Dolphin’s Head there. It had been a great stage-coaching town in the great stage-coaching times, and the ruthless railways had killed and buried it. That is all we are told about its whereabouts. “The sign of the house was the Dolphin’s Head. Why only head I don’t know; for the Dolphin’s effigy at full length, and upside down— as a dolphin is always bound to be when artistically treated, though I suppose he is sometimes right side upward in his natural condition—
- 67. graced the sign-board. The sign-board chafed its rusty hooks outside the bow-window of my room, and was a shabby work. No visitor could have denied that the dolphin was dying by inches, but he showed no bright colours. He had once served another master; there was a newer streak of paint below him, displaying with inconsistent freshness the legend, By J. Mellows. “Pursuing my researches in the Dolphin’s Head, I found it sorely shrunken. When J. Mellows came into possession, he had walled off half the bar, which was now a tobacco shop with its own entrance in the yard—the once glorious yard where the post-boys, whip in hand and always buttoning their waistcoats at the last moment, used to come running forth to mount and away. A ‘Scientific Shoeing-Smith and Veterinary Surgeon’ had further encroached upon the yard; and a grimly satirical Jobber, who announced himself as having to let ‘A neat one-horse fly, and a one-horse cart,’ had established his business, himself, and his family, in a part of the extensive stables. Another part was lopped clean off from the Dolphin’s Head, and now comprised a chapel, a wheelwright’s, and a Young Men’s Mutual Improvement and Discussion Society (in a loft); the whole forming a back lane. No audacious hand had plucked down the vane from the central cupola of the stables, but it had grown rusty and stuck at Nil: while the score or two of pigeons that remained true to their ancestral traditions and the place had collected in a row on the roof- ridge of the only outhouse retained by the Dolphin, where all the inside pigeons tried to push the outside pigeon off. This I accepted as emblematical of the struggle for post and place in railway times.” There are, however, at least three inns we have been able to trace: the Blue Boar, London (dealt with in a previous chapter), the Crispin and Crispianus at Strood, and The Lord Warden Hotel at Dover. The latter is referred to in the chapter entitled “The Calais Night Mail” as follows: “I particularly detest Dover for the self-complacency with which it goes to bed. It always goes to bed (when I am going to Calais) with a more brilliant display of lamp and candle than any other town. Mr.
- 68. and Mrs. Birmingham, host and hostess of the Lord Warden, are my much esteemed friends, but they are too conceited about the comforts of that establishment when the Night Mail is starting. I know it is a good house to stay at, and I don’t want the fact insisted upon in all its warm bright windows at such an hour. I know The Warden is a stationary edifice that never rolls or pitches, and I object to its big outline seeming to insist upon that circumstance, and, as it were, to come over me with it, when I am reeling on the deck of the boat. Beshrew the Warden likewise for obstructing that corner, and making the wind so angry as it rushes round. Shall I not know that it blows quite soon enough without the officious Warden’s interference?” The Lord Warden was evidently built on the site of the Ship, as we have already noted in the chapter devoted to A Tale of Two Cities. The Crispin and Crispianus at Strood is mentioned in the chapter on “Tramps.” The tramp in question is a clockmaker, who, having repaired a clock at Cobham Hall, and paid freely for it, says, “We should be at liberty to go, and should be told by a pointing helper to keep round over yonder by the blasted oak, and go straight through the woods till we should see the town lights right before us.... So should we lie that night at the ancient sign of the Crispin and Crispianus, and rise early next morning to be betimes on tramp again.” The Crispin and Crispianus is a very old-fashioned inn still standing just outside Strood. It is a long building with an overhanging upper floor built with wood. How long the present house has existed we cannot tell, but its hanging sign speaks of St. Crispin’s Day, 1415, and it is said that it may probably have had its origin from the Battle of Agincourt fought on that day. Mr. Harper thinks the sign older than that, and probably was one of the very many religious inn-signs designed to attract the custom of thirsty wayfarers to Becket’s shrine.
- 69. The brothers Crispin and Crispian were members of a noble family in ancient Rome, who, professing Christianity, fled to Gaul and supported themselves by shoemaking in the town of Troyes. They suffered martyrdom in Soissons in A.D. 287. The sign represents the patron saints of the shoemaking fraternity, as these holy brothers are designated, at work on their cobblers’ bench, and is understood to have been faithfully copied from a well-known work preserved to this day at the church of St. Pantaleon at Troyes. THE CRISPIN AND CRISPIANUS Drawn by C. G. Harper The inn’s interior is typical of those to be found in country villages, with its sanded floor of the parlour, and wooden settles with arms at
- 70. each corner. One of these corners is said to have been the favourite seat of Dickens, for it is known that he sometimes called at the inn as he drew near the end of one of his long walks, either alone or with friends, for refreshments. It was an inn, as he said elsewhere, that no thirsty man was known to pass on a hot summer’s day.
- 71. I CHAPTER XV Christmas Stories and Minor Writings THE MITRE INN—THE SALISBURY ARMS—THE PEAL OF BELLS—THE NUTMEG-GRATER—THE DODO—THE PAVILIONSTONE HOTEL—HEN AND CHICKENS—THE SWAN N the First Branch of “The Holly Tree,” in Christmas Stories, there are many inns far and wide referred to, and reminiscences associated with each recalled. These reminiscences may be personal to Dickens or merely of an imaginary nature. The Holly Tree Inn itself is real enough, and has been identified as the George, Greta Bridge, referred to in our chapter on Nicholas Nickleby. There is no doubt, either, that the inn in the cathedral town where Dickens went to school was the Mitre Inn at Chatham. “It was the inn where friends used to put up,” he says, “and where we used to go to see parents, and to have salmon and fowls, and to be tipped. It had an ecclesiastical sign—the Mitre—and a bar that seemed the next best thing to a bishopric, it was so snug. I loved the landlord’s daughter to distraction—but let that pass. It was in that inn that I was cried over by my rosy little sister, because I had acquired a black eye in a fight. And though she had been, that Holly Tree night, for many a long year where all tears are dried, the Mitre softened me yet.”
- 72. THE MITRE INN, CHATHAM From an engraving The Mitre Inn and Clarence Hotel still exists at Chatham, very much as it was in Dickens’s childhood days when his family lived in Ordnance Terrace. It was kept in those days by a Mr. Tribe, who was a friend of John Dickens, and the two families met there and enjoyed many friendly evenings when Dickens and his sister, as he has told us, mounted on a dining-table for a stage, would sing some old sea-songs together. He had a clear treble voice then, but, recalling these incidents many years afterward, said, “he must have been a horrible little nuisance to many unoffending grown-up people who were called upon to admire him.” The Mitre Inn was described in 1838 as being the Manor House, and the first posting-house of the town. It is also on record that, at the close of the eighteenth century, Lord Nelson used to reside there when on duty at Chatham, and that the room he occupied was known as “Nelson’s Cabin” till recent times. William the Fourth, when
- 73. Duke of Clarence, used to stay there, hence the added word of Clarence to the sign. The Salisbury Arms at Hatfield where Mr. and Mrs. Lirriper went upon their wedding-day, “and passed as happy a fortnight as ever happy was,” adjoined the little post-office there, and now exists as a private house. Mr. Lirriper’s youngest brother also had a sneaking regard for the Salisbury Arms, where he enjoyed himself for the space of a fortnight and left without paying his bill, an omission Mrs. Lirriper rectified in the innocent belief that it was fraternal affection which induced her unprincipled brother-in-law to favour Hatfield with his presence. It is believed that Dickens and Phiz stayed the night of October the 27th, 1838, at the Salisbury Arms, when they made their excursion to the West Country. The scene of the first four chapters of “A Message from the Sea,” is laid in “Steepways, North Devon, England,” the name Dickens gives to Clovelly, and the story opens with a faithful and unmistakable description of one of the most beautiful and quaintest villages in England. To it comes Captain Jorgan to unravel a sea mystery, but no reference is made to his staying at the inn there. The task he has set himself, however, eventually takes him to another adjacent village, which Dickens calls Lanrean. There he puts up at the King Arthur’s Arms, to identify which we must first identify Lanrean. That Dickens had a certain village near Clovelly in mind, there is little doubt, for he and Wilkie Collins, who collaborated in writing the story, went there for the purpose. Their description of Clovelly being so accurate and meticulous, it is only natural that Lanrean has a prototype, and, if found, the original of King Arthur’s Arms would be forthcoming. The original of the Peal of Bells, the village ale-house, in “Tom Tiddler’s Ground,” on the other hand, has been discovered, for Mr. Traveller seeking Mr. Mopes the Hermit, naturally had to go where Mr. Mopes the Hermit located himself. This we know to have been
- 74. near Stevenage, and F. G. Kitton identified the ale-house as the White Hart there, where Dickens called on his way to see Lucas, the original of Mr. Mopes, to enquire of the landlord, old Sam Cooper, the shortest route to his “ruined hermitage” some five miles distant. No particular coffee-houses were, we suspect, intended for the Slamjam Coffee-House or the Admiral Nelson Civic and General Dining Rooms, mentioned in “Somebody’s Luggage”; nor can we hope to identify the George and the Gridiron, where the waiters supported nature by what they found in the plates, “which was, as it happened, and but too often thoughtlessly, immersed in mustard,” or what was found in the glasses, “which rarely went beyond driblets and lemons.” No name either is given to the inn in “Mugby Junction” where the traveller arrived at past three o’clock on a tempestuous morning and found himself stranded. Having got his two large black portmanteaux on a truck, the porter trundled them on “through a silent street” and came to a stop. When the owner had shivered on the pavement half an hour, “what time the porter’s knocks at the inn door knocked up the whole town first, and the inn last, he groped his way into the close air of a shut-up house, and so groped between the sheets of a shut-up bed that seemed to have been expressly refrigerated for him when last made.” It is known that Mugby stood for Rugby, but that is all. The particular shut-up inn, if it ever had any original, has not, so far as we are aware, been discovered. In A Christmas Carol we are told that Scrooge “took his melancholy dinner in his usual melancholy tavern; and having read all the newspapers, and beguiled the rest of the evening with his banker’s book, went home to bed.” There were many taverns in the city of London at which Scrooge might have dined, and it may be that Baker’s Chop-House in Change Alley, as has been suggested, was the one he chose. It is no longer
- 75. a chop-house, having a year or so back been taken over by a city business company, and the building added to their premises. But it had been for a century or more a noted city chop-house, where, up to the last, meals were served on pewter plates, and other old-time customs were retained. It was one of those city houses, of which some still exist happily, where the waiters grow old in the service of their customers. Baker’s had at least one such waiter, known familiarly as James, who pursued his calling there for thirty-five years, and became famous by having his portrait painted in oils and hung in the lower room, where it remained until the end of the career of the house as a tavern. Perhaps old Scrooge was one of his special customers. The Nutmeg-Grater, the inn kept by Benjamin Britain in “The Battle of Life,” has no real prototype, but such an inn as described would entice any country rambler into its cosy interior. It was “snugly sheltered behind a great elm tree, with a rare seat for idlers encircling its capacious bole, addressed a cheerful front towards the traveller, as a house of entertainment ought, and tempted him with many mute but significant assurances of a comfortable welcome. The ruddy sign-board perched up in the tree, with its golden letters winking in the sun, ogled the passer-by, from among the leaves, like a jolly face, and promised good cheer. The horse trough, full of clear, fresh water, and the ground below it sprinkled with droppings of fragrant hay, made every horse that passed prick up his ears. The crimson curtains of the lower rooms, and the pure white hangings in the little bedrooms above, beckoned Come in! with every breath of air. Upon the bright green shutters, there were golden legends about beer and ale, and neat wines, and good beds, and an affecting picture of a brown jug frothing over at the top. Upon the window- sills were flowering plants in bright red pots, which made a lively show against the white front of the house; and in the darkness of the doorway there were streaks of light, which glanced off from the surface of bottles and tankards”——
- 76. An ideal picture of an inn any traveller would love to encounter and sample. Reprinted Pieces would form a happy hunting-ground for tracking down inns and public-houses mentioned in its pages if one were so minded. Few of them would prove to be of any importance if discovered, but the task would have its excitement and interest. Take for instance the chapter devoted to the Detective Police. No doubt the taverns used by the criminals which the police had to visit were real houses, as the detectives whom Dickens interviewed were real persons. In this chapter alone there is the Warwick Arms, through which, and the New Inn near R., Tally-Ho Thompson the horse stealer was tracked and captured; the “little public-house” near Smithfield, used by journeymen butchers, and those concerned in “the extensive robberies of lawns and silks”; and the Setting Moon in the Commercial Road, where Simpson was arrested in a room upstairs. Then there is the extinct inn, the Dodo, in one of the chiefest towns of Staffordshire—the pivot of the chapter on “A Plated Article.” Which is the town, and which is the inn referred to, we know not. But Dickens’s description of it is very minute: “If the Dodo were only a gregarious bird,” he says, “if he had only some confused idea of making a comfortable nest, I could hope to get through the hours between this and bedtime, without being consumed by devouring melancholy. But the Dodo’s habits are all wrong. It provides me with a trackless desert of sitting-room, with a chair for every day in the year, a table for every month, and a waste of sideboard where a lonely China vase pines in a corner for its mate long departed, and will never make a match with the candlestick in the opposite corner if it live till Doomsday. The Dodo has nothing in the larder. Even now I behold the Boots returning with my sole in a piece of paper; and, with that portion of my dinner, the Boots, perceiving me at the blank bow-window, slaps his leg as he comes across the road, pretending it is something else. The Dodo excludes
- 77. the outer air. When I mount up to my bedroom, a smell of closeness and flue gets lazily up my nose like sleepy snuff. The loose little bits of carpet writhe under my tread, and take wormy shapes. I don’t know the ridiculous man in the looking-glass, beyond having met him once or twice in a dish-cover—and I can never shave him to- morrow morning! The Dodo is narrow-minded as to towels; expects me to wash on a freemason’s apron without the trimming: when I ask for soap, gives me a stony-hearted something white, with no more lather in it than the Elgin marbles. The Dodo has seen better days, and possesses interminable stables at the back—silent, grass- grown, broken-windowed, horseless. This mournful bird can fry a sole, however, which is much. Can cook a steak, too, which is more. I wonder where it gets its sherry? If I were to send my pint of wine to some famous chemist to be analysed, what would it turn out to be made of? It tastes of pepper, sugar, bitter-almonds, vinegar, warm knives, any flat drinks, and a little brandy. Would it unman a Spanish exile by reminding him of his native land at all? I think not. If there really be any townspeople out of the churchyards, and if a caravan of them ever do dine, with a bottle of wine per man, in this desert of the Dodo, it must make good for the doctor next day!” If the Dodo is undiscoverable, the same need not be said of the Pavilionstone Hotel, because we know that Dickens gave that name to the town of Folkestone, in the chapter entitled “Out of Town.” The lion of Pavilionstone, he tells us, is its great hotel, and one sees at once how he manufactured the name, for its hotel was, and is to- day, called the Pavilion. “A dozen years ago, going over to Paris by South-Eastern Tidal Steamer,” the narrative goes on, “you used to be dropped upon the platform of the main line Pavilionstone Station (not a junction then) at eleven o’clock on a dark winter’s night, in a roaring wind; and in the howling wilderness outside the station was a short omnibus which brought you up by the forehead the instant you got in at the door; and nobody cared about you, and you were alone in the world. You bumped over infinite chalk, until you were turned out at a
- 78. strange building which had just left off being a barn without having quite begun to be a house, where nobody expected your coming, or knew what to do with you when you were come, and where you were usually blown about, until you happened to be blown against the cold beef, and finally into bed. At five in the morning you were blown out of bed, and after a dreary breakfast, with crumpled company, in the midst of confusion, were hustled on board a steamboat, and lay wretched on deck until you saw France lingering and surging at you with great vehemence over the bowsprit.” THE LORD WARDEN HOTEL, DOVER See page 253
- 79. Welcome to our website – the perfect destination for book lovers and knowledge seekers. We believe that every book holds a new world, offering opportunities for learning, discovery, and personal growth. That’s why we are dedicated to bringing you a diverse collection of books, ranging from classic literature and specialized publications to self-development guides and children's books. More than just a book-buying platform, we strive to be a bridge connecting you with timeless cultural and intellectual values. With an elegant, user-friendly interface and a smart search system, you can quickly find the books that best suit your interests. Additionally, our special promotions and home delivery services help you save time and fully enjoy the joy of reading. Join us on a journey of knowledge exploration, passion nurturing, and personal growth every day! ebookbell.com