Intelligent Buildings: Foundation for Intelligent Physical Agents

Francisco J. Mora Lizán. Int. Journal of Engineering Research and Application www.ijera.com
ISSN : 2248-9622, Vol. 7, Issue 5, ( Part -2) May 2017, pp.21-25
www.ijera.com DOI: 10.9790/9622-0705022125 21 | P a g e
Intelligent Buildings: Foundation for Intelligent Physical Agents
Francisco J. Mora Lizán1
, Carlos Rizo Maestre2
,
1
(Departamento de Ciencia de la Computación e IA. Universidad de Alicante. Spain)
2
(Instituto del Agua y Ciencias Ambientales. Universidad de Alicante. Spain)
ABSTRACT
FIPA is an IEEE Computer Society standards organization that promotes agent-based technology and the
interoperability of its standards with other technologies. In the design phase of Intelligent Buildings, it is
essential to manage many services and facilities, to do this, multi-agent systems are a good tool to manage them.
In this paper, we will gereneral description of the features and elements of multiagent systems described by
Foundation for Intelligent Physical Agents (FIPA). Secondly, we will focus on the architectures of these multi-
agent systems. And finally, we will propose a multi-agent system design to see the application in the design of a
detached house where the lighting, air conditioning and security systems will be integrated.
Keywords – Architecture, Intelligent Building, Multiagent Systems
I. INTRODUCTION
In the field of intelligent building design,
we must have mechanisms to integrate sensors,
actuators, different automatic control systems and
information capture. MultiAgent Systems (MAS)
form a community of interdependent elements that
act individually. From the moment that a group of
elements, in this case agents, assume to act in group
forming systems, the capacity to negotiate and to
coordinate different tasks is necessary.
In many cases the use of individual agents
is not just as suitable for all situations that occur in
practice. Solving a problem using an individual
agent causes major constraints. An individual agent
requires an enormous amount of knowledge to solve
complex problems. In the worst case, the problem
can be so complex that an agent cannot find a useful
solution. Even when the individual agent can solve a
problem, it always presents a bottleneck in terms of
speed, reliability, flexibility and modularity. Multi-
agent systems offer a method to avoid the
problematic situations described. In a multi-agent
system, several independent autonomous agents are
active. Each of these agents is dedicated to their own
objectives and only contacts the other agents to
obtain information, or to contribute to a coordinated
solution of a general problem. In both situations,
each individual agent has a specific task for which it
is adequate and whose solution does not exceed its
capabilities. This allows for the processing of
complex problems.
In a building, we can find different
subsystems (security, air conditioning, lighting,
multimedia, control etc), each with specific
objectives and in turn all must work together to
achieve a common goal.
Multi-agent systems provide a great
advantage: they allow the integration of existing
agents into a large system. Therefore, solving a
problem does not require the design and
development of a new specialized agent, instead, the
knowledge of existing agents can be used by
combining them into a multi-agent system and
allowing them to work together to solve the
problem.
The multiagent systems (MAS) constitute a
field of research that compose a wide set of
applications within Intelligent Buildings, their
construction [1], sensor and control of buildings [2],
[3], maintenance of Intelligent Buildings [4]. A
modern approach to Architectural Construction
involves considering MAS as elements linked to the
architectural design of buildings [5]. In this sense, a
multidisciplinary approach to architectural design is
essential [6], the Building Information Models [7]
supplemented with the MASs constitute powerful
systems to holistic design of buildings [8].
II. FOUNDATION FOR INTELLIGENT
PHYSICAL AGENTS. FIPA
FIPA [9] is an IEEE Computer Society
standards organization that promotes agent-based
technology and the interoperability of its standards
with other technologies. FIPA was originally formed
as a Swiss based organization in 1996 to produce
software standards specifications for heterogeneous
and interacting agents and agent based systems.
FIPA, the standards organization for agents and
multi-agent systems was officially accepted by the
IEEE as its eleventh standards committee on 2005.
From the structural point of view, the FIPA standard
defines a series of key elements, shown in Fig. 1.:
1. An Agent Platform (AP) provides the physical
infrastructure in which agents can be deployed.
The AP consists of the machine(s), operating
system, agent support software, FIPA agent
RESEARCH ARTICLE OPEN ACCESS

management components (DF, AMS and MTS)
and agents.
Figure 1. FIPA Management Reference Model.
2. A Directory Facilitator (DF) is an optional
component of the AP, but if it is present, it must
be implemented as a DF service. The DF
provides yellow pages services to other agents.
Agents may register their services with the DF
or query the DF to find out what services are
offered by other agents. Multiple DFs may exist
within an AP and may be federated.
3. An Agent Management System (AMS) is a
mandatory component of the AP. The AMS
exerts supervisory control over access to and
use of the AP. Only one AMS will exist in a
single AP. The AMS maintains a directory of
AIDs which contain transport addresses
(amongst other things) for agents registered
with the AP. The AMS offers white pages
services to other agents. Each agent must
register with an AMS to get a valid AID. There
are three ways in which an agent can be
registered with an AMS: the agent was created
on the AP, the agent explicitly registered with
the AP and the agent migrated to the AP, for
those APs which support agent mobility.
4. A Message Transport Service (MTS) is the
default communication method between agents
on different APs. FIPA is concerned only with
how communication is carried out between
agents who are native to the AP and agents
outside the AP. Agents are free to exchange
messages directly by any means that they can
support.
It should be noted that the concept of an AP
does not mean that all agents resident on an AP have
to be co-located on the same host computer. FIPA
envisages a variety of different APs from single
processes containing lightweight agent threads, to
fully distributed APs built around proprietary or
open middleware standards.
FIPA defines the specification of a
language for the communication between agents
(ACL) in which different types of content are
represented: SL (Semantic Language), CCL
(Constraint Choice Language), KIF (Knowledge
Interchange Format) And RDF ("Resource
Description Framework"). In this paper, we are not
going to extend more in this part of communications,
but we will do it in the structural one.
III. AGENT ARCHITECTURES
There are many types of intelligent agents,
each of them developing the tasks for which it has
been created. And while for some time becomes a
critical resource, others will be able to make more
rational decisions by having enough time and
knowledge to do so. Some agents will be more
complex and intelligent than others. Therefore, there
is no single ideal architecture for intelligent agents.
The concrete structure of the architectures will
depend on the tasks and the environment where they
are developed. We are going to analyze some
architectures that have served as reference and that
can be representative of the wide range of
possibilities that exist.
Reactive architectures are based on a close
relationship between perception and action. They
work well in real-time environments since they are
computationally economical.
Reactive agents do not have a symbolic
model of their environment. The ability to perform
complex reasoning processes is also omitted. The
reason for these restrictions is in the creation of
compact, fault tolerant, and above all, flexible
agents.
The Fig. 2 shows the fundamental
architecture of the reactive agents that correspond to
a simple stimulus / response system. The sensors
collect the information, send it to the corresponding
competition modules, producing a reaction as output
in the same, which is transmitted to the outside by
means of actuators.
Figure 2: Reactive agent architecture
Deliberative architectures follow the
current of symbolic AI, which is based on the
hypothesis of the physical-symbol systems
enunciated by Newell and Simons, according to
which a system of physical symbols capable of

manipulating symbolic structures can exhibit
intelligent behavior. In order to be able to work at
Newell's knowledge level, our problem will be how
to describe the objectives and means of satisfying
them, and how to translate the level of knowledge at
the symbolic level.
Decisions are made using deductive
mechanisms: Pattern matching and various logical
formalisms.
Figure 3: Architecture of deliberative agents
Deliberative agent architectures (Fig. 3) are
usually based on the classical artificial intelligence
planning theory: given an initial state, a set of
operators / plans, and an objective state, the agent's
deliberation is to determine which steps to chain To
achieve its goal, following a top-down approach.
In BDI ("Belief-Desire-Intention")
architectures, decision-making takes place on a
process of reasoning based on the agent's beliefs
about the world and taking into account intentions
and actions.
The basic components of this architecture
are the beliefs, desires and intentions of the agent;
The functions that represent their deliberation; and
the reasoning of ends and means. Practical reasoning
involves two important processes: deciding what
goals to achieve, a process known as deliberation;
and how these goals will be achieved, a procedure
called means-ends reasoning. The decision process
typically begins by trying to understand what
options are available; once this set of alternatives has
been generated, one must choose between them and
commit with one; this chosen option becomes an
intention, which determines the actions of the agent.
Intentions locate the agent's future practical
reasoning; when one has a particular intention, all
those options that are inconsistent with intention are
discarded. In addition, once an intention is adopted,
the agent must persevere ("persist") in it, it must
only rectify it when the reason for which it had the
intention has changed; or when the agent knows with
certainty that he will not be able to comply with it.
Finally, intentions are closely related to
beliefs about the future. When he has an intention,
the agent should at least believe that he has a great
chance of complying with it.
Hybrid Architecture "Touring Machines"
combine deliberative and reactive aspects, by
combining reactive modules with deliberative
modules. The reactive modules are responsible for
processing the stimuli that do not require
deliberation, while the deliberative modules
determine what actions must be performed to satisfy
the local and cooperative objectives of the agents.
The Hybrid Architecture "Touring
Machines" are organized by horizontal layers that
produce activities. This means that each layer
constantly produces suggestions about the actions
that the agent should carry out. The reactive layer
provides a more or less immediate response to
changes in the environment. It is implemented as a
set of action-situation rules as shown in Fig. 4.
Figure 4: Hybrid Architecture "Turing Machine"
The planning layer serves to provide a
proactive agent behavior. Under normal
circumstances, the planning layer is responsible for
deciding what the agent does. Use a set of skeletons
of plans called schemes. These schemes are
essentially hierarchically structured plans that the
agent elaborates at runtime to decide what to do. To
carry out a goal, the planning layer tries to find a
scheme that corresponds to that goal. This schema
will contain sub-goals, which the planning layer uses
to find other schemas that correspond with them.
The modeling layer represents the various
entities of the world (including the agent itself, as
well as other agents). Predicts conflicts between
agents and generates new objectives to resolve these
conflicts. These new objectives are passed on to the
planning layer that looks for schemes that satisfy
them.
The three control layers are embedded in a
control subsystem, which decides which of the
layers will have control over the agent. This control
subsystem is implemented as a set of control rules.
These control rules can suppress sensor information
for some layer or censor the actions of some layer.
"InteRRaP" is an example of a vertically
layered and two-step split agent architecture, as
shown in Fig. 5.

Figure 5: InteRRaP Architecture
As in the "Touring Machines" have three
layers of control. In addition, the purpose of each
"InteRRaP" layer seems to match the corresponding
layer of the "Touring Machine". Each layer is
associated with a knowledge base, for example, a
representation of the world suitable for each layer.
These knowledge bases represent the agent and his
environment at different levels of abstraction. The
knowledge base at the highest level represents the
plans and actions of the other agents in the
environment; the knowledge base of the middle level
represents the plans and actions of the agent himself;
and the bottom-level knowledge base represents
information about the environment. The explicit
introduction of these knowledge bases distinguishes
the "InteRRaP" from the "Touring Machines". The
way of the different layers in "InteRRaP" work to
produce the behavior also differs from the "Touring
Machines". The main difference is the way of the
layers interact with the environment. In "Toruing
Machines" each layer had as input the perceptions of
the environment and as an output an action in the
environment. This introduced the need for a control
module to deal with conflicts. In "InteRRaP", the
layers interact with the rest to achieve the same
purpose.
IV. APPLICATION OFARCHITECTURES.
In our case, we will propose a multi-agent
system for a single-person home, where we will use
3 reactive agents for the safety, air conditioning and
lighting subsystems, and one with the Touring
Machines architecture for global planning and
control as shown in Fig. 6. The system will be
completed by an agent that acts as an external
interface with the users (dashboard), a service agent
(ADF) and an agent control agent (AMS).
Figure 6: Multiagent System in Intelligent Building
The reactive agent of the safety subsystem
receives information directly from the presence, gas,
smoke and water sensors. In the case of detection of
leaks of water or gas will cut the servo valves of
such systems. In the event of a fire, it could start the
alarm and the fire system of the house.
For the air conditioning system, we will use
another reactive agent where its sensors will collect
the data of temperature, humidity of the different
rooms and external wind and interectuará with the
actuator modules of the air conditioning and
awnings.
The latest reactive agent in our system will
be responsible for lighting. This agent will obtain
information of the interior and exterior light, and
indirectly of the presence sensors that the agent of
the security system has obtained. Interact with the
actuators of blinds and interior lighting. We could
also make requests to the air conditioning agent to
act on the awnings, requesting for example to raise
the awning to get more ambient light.
A very important agent in our system will
be with a planning and control agent that uses the
Touring Machine architecture. The mission of this
agent will be to maintain the security of the house, to
satisfy the needs of the inhabitants of the house
(comfort) and to promote energy saving. It will be
very important to define the desired plans and
activities. This agent will send plans and rules to the
different agents of the platform (subsystems).
We could have several agents forming each
subsystem, the ADF agent would be the one that
would inform the rest of the services that each of
these agents possesses. In our example the lighting
agent would ask the ADF if there is an agent in
charge of the service of awnings, the ADF would
answer that there is one, and the air conditioning
agent will make the necessary requests if necessary.
Remember that all communication between
agents will do through the "message transport
system" using the agent communication language
defined by FIPA.

V. CONCLUSION
In this paper, we have developed the
different agent architectures, applying these concepts
in the design of a multi-agent system for a house
where several subsystems (lighting, air conditioning,
safety and control) are integrated. For each of them
an architecture has been selected, attesting to its
characteristics and needs. This system has been
provided with a series of standard agents (Directory
Facilitator, Agent Management System) according to
the FIPA standard whose mission is to register and
disseminate the services.
ACKNOWLEDGEMENTS
This work has been supported by the
Ministerio de Economía y Competitividad (Spain),
project TIN2013-40982-R. Project co-financed with
FEDER funds.
REFERENCES
[1]. Z. Ren and C. . Anumba, “Multi-agent
systems in construction–state of the art and
prospects,” Autom. Constr., vol. 13, no. 3,
pp. 421–434, May 2004.
[2]. S. Sharples, V. Callaghan, and G. Clarke,
“A multi‐agent architecture for intelligent
building sensing and control,” Sens. Rev.,
vol. 19, no. 2, pp. 135–140, Jun. 1999.
[3]. P. Davidsson and M. Boman, “A multi-
agent system for controlling intelligent
buildings,” in Proceedings Fourth
International Conference on MultiAgent
Systems, 2000, pp. 377–378.
[4]. S. Wang, Z. Xu, H. Li, J. Hong, and W.
Shi, “Investigation on intelligent building
standard communication protocols and
application of IT technologies,” Autom.
Constr., vol. 13, no. 5, pp. 607–619, 2004.
[5]. J. K. W. Wong, H. Li, and S. W. Wang,
“Intelligent building research: a review,”
Autom. Constr., vol. 14, no. 1, pp. 143–
159, 2005.
[6]. Z. Ren, F. Yang, N. M. Bouchlaghem, and
C. J. Anumba, “Multi-disciplinary
collaborative building design—A
comparative study between multi-agent
systems and multi-disciplinary optimisation
approaches,” Autom. Constr., vol. 20, no. 5,
pp. 537–549, Aug. 2011.
[7]. E. A. Pärn, D. J. Edwards, and M. C. P.
Sing, “The building information modelling
trajectory in facilities management: A
review,” Autom. Constr., vol. 75, pp. 45–
55, Mar. 2017.
[8]. C. Rizo Maestre and F. J. Mora Lizán,
“Intelligent Buildings: Considerations for
its Design using Multiagent Systems,” Int.
J. Eng. Res. Technol., no. Volume. 6-Issue.
04, April-2017, 2017.
[9]. “Foundation for Intelligent Physical
Agents,” 1996. [Online]. Available:
http://www.fipa.org/. [Accessed: 17-Mar-
2017].

Intelligent Buildings: Foundation for Intelligent Physical Agents

More Related Content

What's hot (15)

Similar to Intelligent Buildings: Foundation for Intelligent Physical Agents (20)

Recently uploaded (20)

Intelligent Buildings: Foundation for Intelligent Physical Agents