Learning an Expert Skill-Space for Replanning Dynamic Quadruped Locomotion over Obstacles

David Surovik, Oliwier Melon, Mathieu Geisert, Maurice Fallon, Ioannis Havoutis
Proceedings of the 2020 Conference on Robot Learning, PMLR 155:1509-1518, 2021.

Abstract

Function approximators are increasingly being considered as a tool for generating robot motions that are temporally extended and express foresight about the scenario at hand. While these longer behaviors are often necessary or beneficial, they also induce multimodality in the decision space, which complicates the training of a regression model on expert data. Motivated by the problem of quadrupedal locomotion over obstacles, we apply an approach that disentangles modal variation from task-to-solution regression by using a conditional variational autoencoder. The resulting decoder is a regression model that outputs trajectories based on the task and a real-valued latent mode vector representing a style of behavior. With the task consisting of robot-relative descriptions of the state, the goal, and nearby obstacles, this model is suitable for receding-horizon generation of structured dynamic motion. We test this approach, along with a trajectory library baseline method, for producing sustained locomotion plans that use a generalized gait. Both options strongly bias planned footholds away from obstacle regions, while the multimodal regressor is far less susceptible to violating kinematic constraints. We conclude by identifying further prospective benefits of the continuous latent mode representation, along with targets for future integration into a hardware-deployable pipeline including perception and control.

Cite this Paper

BibTeX
@InProceedings{pmlr-v155-surovik21a, title = {Learning an Expert Skill-Space for Replanning Dynamic Quadruped Locomotion over Obstacles}, author = {Surovik, David and Melon, Oliwier and Geisert, Mathieu and Fallon, Maurice and Havoutis, Ioannis}, booktitle = {Proceedings of the 2020 Conference on Robot Learning}, pages = {1509--1518}, year = {2021}, editor = {Kober, Jens and Ramos, Fabio and Tomlin, Claire}, volume = {155}, series = {Proceedings of Machine Learning Research}, month = {16--18 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v155/surovik21a/surovik21a.pdf}, url = {https://proceedings.mlr.press/v155/surovik21a.html}, abstract = {Function approximators are increasingly being considered as a tool for generating robot motions that are temporally extended and express foresight about the scenario at hand. While these longer behaviors are often necessary or beneficial, they also induce multimodality in the decision space, which complicates the training of a regression model on expert data. Motivated by the problem of quadrupedal locomotion over obstacles, we apply an approach that disentangles modal variation from task-to-solution regression by using a conditional variational autoencoder. The resulting decoder is a regression model that outputs trajectories based on the task and a real-valued latent mode vector representing a style of behavior. With the task consisting of robot-relative descriptions of the state, the goal, and nearby obstacles, this model is suitable for receding-horizon generation of structured dynamic motion. We test this approach, along with a trajectory library baseline method, for producing sustained locomotion plans that use a generalized gait. Both options strongly bias planned footholds away from obstacle regions, while the multimodal regressor is far less susceptible to violating kinematic constraints. We conclude by identifying further prospective benefits of the continuous latent mode representation, along with targets for future integration into a hardware-deployable pipeline including perception and control.} }
Endnote
%0 Conference Paper %T Learning an Expert Skill-Space for Replanning Dynamic Quadruped Locomotion over Obstacles %A David Surovik %A Oliwier Melon %A Mathieu Geisert %A Maurice Fallon %A Ioannis Havoutis %B Proceedings of the 2020 Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2021 %E Jens Kober %E Fabio Ramos %E Claire Tomlin %F pmlr-v155-surovik21a %I PMLR %P 1509--1518 %U https://proceedings.mlr.press/v155/surovik21a.html %V 155 %X Function approximators are increasingly being considered as a tool for generating robot motions that are temporally extended and express foresight about the scenario at hand. While these longer behaviors are often necessary or beneficial, they also induce multimodality in the decision space, which complicates the training of a regression model on expert data. Motivated by the problem of quadrupedal locomotion over obstacles, we apply an approach that disentangles modal variation from task-to-solution regression by using a conditional variational autoencoder. The resulting decoder is a regression model that outputs trajectories based on the task and a real-valued latent mode vector representing a style of behavior. With the task consisting of robot-relative descriptions of the state, the goal, and nearby obstacles, this model is suitable for receding-horizon generation of structured dynamic motion. We test this approach, along with a trajectory library baseline method, for producing sustained locomotion plans that use a generalized gait. Both options strongly bias planned footholds away from obstacle regions, while the multimodal regressor is far less susceptible to violating kinematic constraints. We conclude by identifying further prospective benefits of the continuous latent mode representation, along with targets for future integration into a hardware-deployable pipeline including perception and control.
APA
Surovik, D., Melon, O., Geisert, M., Fallon, M. & Havoutis, I.. (2021). Learning an Expert Skill-Space for Replanning Dynamic Quadruped Locomotion over Obstacles. Proceedings of the 2020 Conference on Robot Learning, in Proceedings of Machine Learning Research 155:1509-1518 Available from https://proceedings.mlr.press/v155/surovik21a.html.

Related Material