Multi-Loco: Unifying Multi-Embodiment Legged Locomotion via Reinforcement Learning Augmented Diffusion

Shunpeng Yang, Zhen Fu, Zhefeng Cao, Guo Junde, Patrick Wensing, Wei Zhang, Hua Chen
Proceedings of The 9th Conference on Robot Learning, PMLR 305:1030-1048, 2025.

Abstract

Generalizing locomotion policies across diverse legged robots with varying morphologies is a key challenge due to differences in observation/action dimensions and system dynamics. In this work, we propose \textit{Multi-Loco}, a novel unified framework combining a morphology-agnostic generative diffusion model with a lightweight residual policy optimized via reinforcement learning (RL). The diffusion model captures morphology-invariant locomotion patterns from diverse cross-embodiment datasets, improving generalization and robustness. The residual policy is shared across all embodiments and refines the actions generated by the diffusion model, enhancing task-aware performance and robustness for real-world deployment. We evaluated our method with a rich library of four legged robots in both simulation and real-world experiments. Compared to a standard RL framework with PPO, our approach - replacing the Gaussian policy with a diffusion model and residual term - achieves a 10.35% average return improvement, with gains up to 13.57% in wheeled-biped locomotion tasks. These results highlight the benefits of cross-embodiment data and composite generative architectures in learning robust, generalized locomotion skills.

Cite this Paper

BibTeX
@InProceedings{pmlr-v305-yang25a, title = {Multi-Loco: Unifying Multi-Embodiment Legged Locomotion via Reinforcement Learning Augmented Diffusion}, author = {Yang, Shunpeng and Fu, Zhen and Cao, Zhefeng and Junde, Guo and Wensing, Patrick and Zhang, Wei and Chen, Hua}, booktitle = {Proceedings of The 9th Conference on Robot Learning}, pages = {1030--1048}, year = {2025}, editor = {Lim, Joseph and Song, Shuran and Park, Hae-Won}, volume = {305}, series = {Proceedings of Machine Learning Research}, month = {27--30 Sep}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v305/main/assets/yang25a/yang25a.pdf}, url = {https://proceedings.mlr.press/v305/yang25a.html}, abstract = {Generalizing locomotion policies across diverse legged robots with varying morphologies is a key challenge due to differences in observation/action dimensions and system dynamics. In this work, we propose \textit{Multi-Loco}, a novel unified framework combining a morphology-agnostic generative diffusion model with a lightweight residual policy optimized via reinforcement learning (RL). The diffusion model captures morphology-invariant locomotion patterns from diverse cross-embodiment datasets, improving generalization and robustness. The residual policy is shared across all embodiments and refines the actions generated by the diffusion model, enhancing task-aware performance and robustness for real-world deployment. We evaluated our method with a rich library of four legged robots in both simulation and real-world experiments. Compared to a standard RL framework with PPO, our approach - replacing the Gaussian policy with a diffusion model and residual term - achieves a 10.35% average return improvement, with gains up to 13.57% in wheeled-biped locomotion tasks. These results highlight the benefits of cross-embodiment data and composite generative architectures in learning robust, generalized locomotion skills.} }
Endnote
%0 Conference Paper %T Multi-Loco: Unifying Multi-Embodiment Legged Locomotion via Reinforcement Learning Augmented Diffusion %A Shunpeng Yang %A Zhen Fu %A Zhefeng Cao %A Guo Junde %A Patrick Wensing %A Wei Zhang %A Hua Chen %B Proceedings of The 9th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2025 %E Joseph Lim %E Shuran Song %E Hae-Won Park %F pmlr-v305-yang25a %I PMLR %P 1030--1048 %U https://proceedings.mlr.press/v305/yang25a.html %V 305 %X Generalizing locomotion policies across diverse legged robots with varying morphologies is a key challenge due to differences in observation/action dimensions and system dynamics. In this work, we propose \textit{Multi-Loco}, a novel unified framework combining a morphology-agnostic generative diffusion model with a lightweight residual policy optimized via reinforcement learning (RL). The diffusion model captures morphology-invariant locomotion patterns from diverse cross-embodiment datasets, improving generalization and robustness. The residual policy is shared across all embodiments and refines the actions generated by the diffusion model, enhancing task-aware performance and robustness for real-world deployment. We evaluated our method with a rich library of four legged robots in both simulation and real-world experiments. Compared to a standard RL framework with PPO, our approach - replacing the Gaussian policy with a diffusion model and residual term - achieves a 10.35% average return improvement, with gains up to 13.57% in wheeled-biped locomotion tasks. These results highlight the benefits of cross-embodiment data and composite generative architectures in learning robust, generalized locomotion skills.
APA
Yang, S., Fu, Z., Cao, Z., Junde, G., Wensing, P., Zhang, W. & Chen, H.. (2025). Multi-Loco: Unifying Multi-Embodiment Legged Locomotion via Reinforcement Learning Augmented Diffusion. Proceedings of The 9th Conference on Robot Learning, in Proceedings of Machine Learning Research 305:1030-1048 Available from https://proceedings.mlr.press/v305/yang25a.html.

Related Material