Bridging Adaptivity and Safety: Learning Agile Collision-Free Locomotion Across Varied Physics

2025-01-08

AI Summary

Legged robots struggle to simultaneously achieve agility and safety in unknown, time-varying physical environments—e.g., low-friction terrain or dynamic payloads. To address this, we propose an adaptive control framework integrating online physical parameter estimation, conditional reach-avoid (RA) value networks, and dual-strategy (agility/recovery) coordination with seamless switching. This work is the first to jointly model these three components and introduces an on-policy fine-tuned robust parameter estimator to mitigate distributional shift. In simulation, our approach improves safety by 50% while significantly increasing average forward velocity. Hardware experiments demonstrate a 19.8% speed gain over ABS—a state-of-the-art baseline—with collision frequency reduced to 42.4% of ABS’s. The framework successfully handles challenging real-world uncertainties, including unmodeled 8 kg payloads and unknown ground friction coefficients.

Technology Category

• Intelligent Robots: State Estimation
• Multiagent Systems: Multiagent Systems under Uncertainty
• Natural Language Processing: Safety and Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Real-world legged locomotion systems often need to reconcile agility and safety for different scenarios. Moreover, the underlying dynamics are often unknown and time-variant (e.g., payload, friction). In this paper, we introduce BAS (Bridging Adaptivity and Safety), which builds upon the pipeline of prior work Agile But Safe (ABS)(He et al.) and is designed to provide adaptive safety even in dynamic environments with uncertainties. BAS involves an agile policy to avoid obstacles rapidly and a recovery policy to prevent collisions, a physical parameter estimator that is concurrently trained with agile policy, and a learned control-theoretic RA (reach-avoid) value network that governs the policy switch. Also, the agile policy and RA network are both conditioned on physical parameters to make them adaptive. To mitigate the distribution shift issue, we further introduce an on-policy fine-tuning phase for the estimator to enhance its robustness and accuracy. The simulation results show that BAS achieves 50% better safety than baselines in dynamic environments while maintaining a higher speed on average. In real-world experiments, BAS shows its capability in complex environments with unknown physics (e.g., slippery floors with unknown frictions, unknown payloads up to 8kg), while baselines lack adaptivity, leading to collisions or. degraded agility. As a result, BAS achieves a 19.8% increase in speed and gets a 2.36 times lower collision rate than ABS in the real world. Videos: https://adaptive-safe-locomotion.github.io.

Problem

• Legged Robot
• Adaptive Mobility
• Friction and Load Variation

Innovation

• BAS Method
• Agile Obstacle Avoidance
• Physical Parameter Estimation