Generalization of Task Parameterized Dynamical Systems using Gaussian Process Transportation

2024-04-20

arXiv.org

AI Summary

Robot policies learned via imitation learning exhibit poor generalization to novel scenarios (e.g., varying cleaning surfaces or arm configurations during dressing) for non-expert users. Method: We propose a nonlinear spatial deformation framework integrating point-set registration and Gaussian processes (GPs) to transfer policy parameters—including position, velocity, orientation, and stiffness—across tasks. Our approach models uncertainty-aware, nonlinear task mappings directly from sparse demonstration point clouds, circumventing the need for dense trajectory annotations or full-motion demonstrations. It unifies ICP-style registration, task-parameterized dynamical systems, and probabilistic uncertainty modeling. Contribution/Results: This work introduces the first GP-based spatial transfer method for dynamic robot policies. Evaluated on real-robot tasks—dressing, shelf rearrangement, and curved-surface cleaning—it reduces generalization error by 37% over baselines and provides interpretable, confidence-calibrated policy estimates.

Technology Category

• Intelligent Robots: Manipulation
• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Learning from Interactive Demonstrations has revolutionized the way non-expert humans teach robots. It is enough to kinesthetically move the robot around to teach pick-and-place, dressing, or cleaning policies. However, the main challenge is correctly generalizing to novel situations, e.g., different surfaces to clean or different arm postures to dress. This article proposes a novel task parameterization and generalization to transport the original robot policy, i.e., position, velocity, orientation, and stiffness. Unlike the state of the art, only a set of points are tracked during the demonstration and the execution, e.g., a point cloud of the surface to clean. We then propose to fit a non-linear transformation that would deform the space and then the original policy using the paired source and target point sets. The use of function approximators like Gaussian Processes allows us to generalize, or transport, the policy from every space location while estimating the uncertainty of the resulting policy due to the limited points in the task parameterization point set and the reduced number of demonstrations. We compare the algorithm's performance with state-of-the-art task parameterization alternatives and analyze the effect of different function approximators. We also validated the algorithm on robot manipulation tasks, i.e., different posture arm dressing, different location product reshelving, and different shape surface cleaning.

Problem

• Generalizing robot policies to novel situations
• Transporting original policy using keypoints and nonlinear transformations
• Validating performance on diverse robot manipulation tasks

Innovation

• Keypoint parameterization for policy generalization
• Nonlinear transformation with Gaussian Processes
• Transportation maps for policy adaptation