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

Yichao Zhong, Chong Zhang, Tairan He, Guanya Shi
Proceedings of the 7th Annual Learning for Dynamics \& Control Conference, PMLR 283:1498-1511, 2025.

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) 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 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, carrying up 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. More detailed videos of the real-world performance of BAS are listed on our website: https://adaptive-safe-locomotion.github.io

Cite this Paper

BibTeX
@InProceedings{pmlr-v283-zhong25a, title = {Bridging Adaptivity and Safety: Learning Agile Collision-Free Locomotion Across Varied Physics}, author = {Zhong, Yichao and Zhang, Chong and He, Tairan and Shi, Guanya}, booktitle = {Proceedings of the 7th Annual Learning for Dynamics \& Control Conference}, pages = {1498--1511}, year = {2025}, editor = {Ozay, Necmiye and Balzano, Laura and Panagou, Dimitra and Abate, Alessandro}, volume = {283}, series = {Proceedings of Machine Learning Research}, month = {04--06 Jun}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v283/main/assets/zhong25a/zhong25a.pdf}, url = {https://proceedings.mlr.press/v283/zhong25a.html}, 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) 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 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, carrying up 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. More detailed videos of the real-world performance of BAS are listed on our website: https://adaptive-safe-locomotion.github.io} }
Endnote
%0 Conference Paper %T Bridging Adaptivity and Safety: Learning Agile Collision-Free Locomotion Across Varied Physics %A Yichao Zhong %A Chong Zhang %A Tairan He %A Guanya Shi %B Proceedings of the 7th Annual Learning for Dynamics \& Control Conference %C Proceedings of Machine Learning Research %D 2025 %E Necmiye Ozay %E Laura Balzano %E Dimitra Panagou %E Alessandro Abate %F pmlr-v283-zhong25a %I PMLR %P 1498--1511 %U https://proceedings.mlr.press/v283/zhong25a.html %V 283 %X 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) 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 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, carrying up 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. More detailed videos of the real-world performance of BAS are listed on our website: https://adaptive-safe-locomotion.github.io
APA
Zhong, Y., Zhang, C., He, T. & Shi, G.. (2025). Bridging Adaptivity and Safety: Learning Agile Collision-Free Locomotion Across Varied Physics. Proceedings of the 7th Annual Learning for Dynamics \& Control Conference, in Proceedings of Machine Learning Research 283:1498-1511 Available from https://proceedings.mlr.press/v283/zhong25a.html.

Related Material