TamedPUMA: safe and stable imitation learning with geometric fabrics

2025-03-21

AI Summary

Imitation learning (IL) struggles to simultaneously ensure stability and physical safety in robot task-space motion control. This paper proposes a novel, safety-stable IL paradigm embedded within the Geometric Fabrics framework: for the first time, the IL policy is formulated as the navigation layer of a geometric fabric, integrating hard constraints—including goal convergence, collision avoidance, and joint limit enforcement—thereby enabling theoretically guaranteed stability and real-time verifiable constraint satisfaction. The method builds upon second-order dynamical modeling and constraint-aware task-space IL, supporting high-fidelity trajectory reproduction. Evaluated in simulation and on a real 7-DoF robotic arm, the approach reduces constraint violation rates by over 90% while preserving high fidelity to expert demonstrations, significantly enhancing both safety and motion robustness.

Technology Category

• Intelligent Robots: Learning & Optimization for ROB
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning
• Natural Language Processing: Safety and Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Using the language of dynamical systems, Imitation learning (IL) provides an intuitive and effective way of teaching stable task-space motions to robots with goal convergence. Yet, IL techniques are affected by serious limitations when it comes to ensuring safety and fulfillment of physical constraints. With this work, we solve this challenge via TamedPUMA, an IL algorithm augmented with a recent development in motion generation called geometric fabrics. As both the IL policy and geometric fabrics describe motions as artificial second-order dynamical systems, we propose two variations where IL provides a navigation policy for geometric fabrics. The result is a stable imitation learning strategy within which we can seamlessly blend geometrical constraints like collision avoidance and joint limits. Beyond providing a theoretical analysis, we demonstrate TamedPUMA with simulated and real-world tasks, including a 7-DoF manipulator.

Problem

• Ensures safe imitation learning for robots
• Combines IL with geometric fabrics for stability
• Integrates collision avoidance and joint limits

Innovation

• Combines imitation learning with geometric fabrics
• Ensures safety via geometric constraints blending
• Demonstrates stability in simulated and real tasks