Fast and Efficient Locomotion via Learned Gait Transitions

Yuxiang Yang, Tingnan Zhang, Erwin Coumans, Jie Tan, Byron Boots
Proceedings of the 5th Conference on Robot Learning, PMLR 164:773-783, 2022.

Abstract

We focus on the problem of developing energy efficient controllers for quadrupedal robots. Animals can actively switch gaits at different speeds to lower their energy consumption. In this paper, we devise a hierarchical learning framework, in which distinctive locomotion gaits and natural gait transitions emerge automatically with a simple reward of energy minimization. We use evolutionary strategies (ES) to train a high-level gait policy that specifies gait patterns of each foot, while the low-level convex MPC controller optimizes the motor commands so that the robot can walk at a desired velocity using that gait pattern. We test our learning framework on a quadruped robot and demonstrate automatic gait transitions, from walking to trotting and to fly-trotting, as the robot increases its speed. We show that the learned hierarchical controller consumes much less energy across a wide range of locomotion speed than baseline controllers.

Cite this Paper


BibTeX
@InProceedings{pmlr-v164-yang22d, title = {Fast and Efficient Locomotion via Learned Gait Transitions}, author = {Yang, Yuxiang and Zhang, Tingnan and Coumans, Erwin and Tan, Jie and Boots, Byron}, booktitle = {Proceedings of the 5th Conference on Robot Learning}, pages = {773--783}, year = {2022}, editor = {Faust, Aleksandra and Hsu, David and Neumann, Gerhard}, volume = {164}, series = {Proceedings of Machine Learning Research}, month = {08--11 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v164/yang22d/yang22d.pdf}, url = {https://proceedings.mlr.press/v164/yang22d.html}, abstract = {We focus on the problem of developing energy efficient controllers for quadrupedal robots. Animals can actively switch gaits at different speeds to lower their energy consumption. In this paper, we devise a hierarchical learning framework, in which distinctive locomotion gaits and natural gait transitions emerge automatically with a simple reward of energy minimization. We use evolutionary strategies (ES) to train a high-level gait policy that specifies gait patterns of each foot, while the low-level convex MPC controller optimizes the motor commands so that the robot can walk at a desired velocity using that gait pattern. We test our learning framework on a quadruped robot and demonstrate automatic gait transitions, from walking to trotting and to fly-trotting, as the robot increases its speed. We show that the learned hierarchical controller consumes much less energy across a wide range of locomotion speed than baseline controllers.} }
Endnote
%0 Conference Paper %T Fast and Efficient Locomotion via Learned Gait Transitions %A Yuxiang Yang %A Tingnan Zhang %A Erwin Coumans %A Jie Tan %A Byron Boots %B Proceedings of the 5th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2022 %E Aleksandra Faust %E David Hsu %E Gerhard Neumann %F pmlr-v164-yang22d %I PMLR %P 773--783 %U https://proceedings.mlr.press/v164/yang22d.html %V 164 %X We focus on the problem of developing energy efficient controllers for quadrupedal robots. Animals can actively switch gaits at different speeds to lower their energy consumption. In this paper, we devise a hierarchical learning framework, in which distinctive locomotion gaits and natural gait transitions emerge automatically with a simple reward of energy minimization. We use evolutionary strategies (ES) to train a high-level gait policy that specifies gait patterns of each foot, while the low-level convex MPC controller optimizes the motor commands so that the robot can walk at a desired velocity using that gait pattern. We test our learning framework on a quadruped robot and demonstrate automatic gait transitions, from walking to trotting and to fly-trotting, as the robot increases its speed. We show that the learned hierarchical controller consumes much less energy across a wide range of locomotion speed than baseline controllers.
APA
Yang, Y., Zhang, T., Coumans, E., Tan, J. & Boots, B.. (2022). Fast and Efficient Locomotion via Learned Gait Transitions. Proceedings of the 5th Conference on Robot Learning, in Proceedings of Machine Learning Research 164:773-783 Available from https://proceedings.mlr.press/v164/yang22d.html.

Related Material