Motion Priors Reimagined: Adapting Flat-Terrain Skills for Complex Quadruped Mobility

Zewei Zhang, Chenhao Li, Takahiro Miki, Marco Hutter
Proceedings of The 9th Conference on Robot Learning, PMLR 305:3762-3777, 2025.

Abstract

Reinforcement learning (RL)-based legged locomotion controllers often require meticulous reward tuning to track velocities or goal positions while preserving smooth motion on various terrains. Motion imitation methods via RL using demonstration data reduce reward engineering but fail to generalize to novel environments. We address this by proposing a hierarchical RL framework in which a low-level policy is first pre-trained to imitate animal motions on flat ground, thereby establishing motion priors. A subsequent high-level, goal-conditioned policy then builds on these priors, learning residual corrections that enable perceptive locomotion, local obstacle avoidance, and goal-directed navigation across diverse and rugged terrains. Simulation experiments illustrate the effectiveness of learned residuals in adapting to progressively challenging uneven terrains while still preserving the locomotion characteristics provided by the motion priors. Furthermore, our results demonstrate improvements in motion regularization over baseline models trained without motion priors under similar reward setups. Real-world experiments with an ANYmal-D quadruped robot confirm our policy’s capability to generalize animal-like locomotion skills to complex terrains, demonstrating smooth and efficient locomotion and local navigation performance amidst challenging terrains with obstacles.

Cite this Paper

BibTeX
@InProceedings{pmlr-v305-zhang25j, title = {Motion Priors Reimagined: Adapting Flat-Terrain Skills for Complex Quadruped Mobility}, author = {Zhang, Zewei and Li, Chenhao and Miki, Takahiro and Hutter, Marco}, booktitle = {Proceedings of The 9th Conference on Robot Learning}, pages = {3762--3777}, 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/zhang25j/zhang25j.pdf}, url = {https://proceedings.mlr.press/v305/zhang25j.html}, abstract = {Reinforcement learning (RL)-based legged locomotion controllers often require meticulous reward tuning to track velocities or goal positions while preserving smooth motion on various terrains. Motion imitation methods via RL using demonstration data reduce reward engineering but fail to generalize to novel environments. We address this by proposing a hierarchical RL framework in which a low-level policy is first pre-trained to imitate animal motions on flat ground, thereby establishing motion priors. A subsequent high-level, goal-conditioned policy then builds on these priors, learning residual corrections that enable perceptive locomotion, local obstacle avoidance, and goal-directed navigation across diverse and rugged terrains. Simulation experiments illustrate the effectiveness of learned residuals in adapting to progressively challenging uneven terrains while still preserving the locomotion characteristics provided by the motion priors. Furthermore, our results demonstrate improvements in motion regularization over baseline models trained without motion priors under similar reward setups. Real-world experiments with an ANYmal-D quadruped robot confirm our policy’s capability to generalize animal-like locomotion skills to complex terrains, demonstrating smooth and efficient locomotion and local navigation performance amidst challenging terrains with obstacles.} }
Endnote
%0 Conference Paper %T Motion Priors Reimagined: Adapting Flat-Terrain Skills for Complex Quadruped Mobility %A Zewei Zhang %A Chenhao Li %A Takahiro Miki %A Marco Hutter %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-zhang25j %I PMLR %P 3762--3777 %U https://proceedings.mlr.press/v305/zhang25j.html %V 305 %X Reinforcement learning (RL)-based legged locomotion controllers often require meticulous reward tuning to track velocities or goal positions while preserving smooth motion on various terrains. Motion imitation methods via RL using demonstration data reduce reward engineering but fail to generalize to novel environments. We address this by proposing a hierarchical RL framework in which a low-level policy is first pre-trained to imitate animal motions on flat ground, thereby establishing motion priors. A subsequent high-level, goal-conditioned policy then builds on these priors, learning residual corrections that enable perceptive locomotion, local obstacle avoidance, and goal-directed navigation across diverse and rugged terrains. Simulation experiments illustrate the effectiveness of learned residuals in adapting to progressively challenging uneven terrains while still preserving the locomotion characteristics provided by the motion priors. Furthermore, our results demonstrate improvements in motion regularization over baseline models trained without motion priors under similar reward setups. Real-world experiments with an ANYmal-D quadruped robot confirm our policy’s capability to generalize animal-like locomotion skills to complex terrains, demonstrating smooth and efficient locomotion and local navigation performance amidst challenging terrains with obstacles.
APA
Zhang, Z., Li, C., Miki, T. & Hutter, M.. (2025). Motion Priors Reimagined: Adapting Flat-Terrain Skills for Complex Quadruped Mobility. Proceedings of The 9th Conference on Robot Learning, in Proceedings of Machine Learning Research 305:3762-3777 Available from https://proceedings.mlr.press/v305/zhang25j.html.

Related Material