… policy results in better target policies. We then turn to a practical approach for incorporating
source policies into diffusion … (Behavioral Refinement via Interpolantbased Diffusion for …

… While IBC is considered state of the art for EBM behavioral cloning, follow up … policies which
are competitive with, and even outperform, diffusion and stochastic interpolant based policies…

… , we adopt an interpolant-based diffusion controller called BRIDGeR [44] to refine VLA-generated
actions using tactile feedback. Unlike conventional diffusion models that start from …

… Diffusion Policy (DP) enables robots to learn complex behaviors by imitating expert
demonstrations through … process, ie utilizing the robot’s observations as the condition to refine …

… As discussed above, these methods were designed for static safety criteria, and the lack the
ability to satisfy temporally extended behaviors. Our work inherits the advantages of diffusion …

… Abstract—Recent advances in diffusion-based robot policies have demonstrated significant
potential in imitating multimodal behaviors. However, these approaches typically require …

… using generative modelling techniques like diffusion, have enabled … Policy, a novel behaviour
cloning approach based on Implicit Maximum Likelihood Estimation (IMLE). IMLE Policy …

… In this study, we do observe the supraconvergence behaviour … using the Tanemura-Ogawa-Ogita
algorithm [73] and subsequent refinement is based on incremental point insertion using …

… diffusion bridges on policy diffusion, we chose NoMaD, the first approach to apply policy
diffusion to … Behavioral refinement via interpolant-based policy diffusion. arXiv preprint arXiv:…

… balance refinement capability and efficiency: too small values limit available expert actions
for refinement, … achieves a success rate exceeding 80% on 31 tasks, whereas Diffusion Policy …