학술논문
Online Learning of Unknown Dynamics for Model-Based Controllers in Legged Locomotion
Document Type
Periodical
Author
Source
IEEE Robotics and Automation Letters IEEE Robot. Autom. Lett. Robotics and Automation Letters, IEEE. 6(4):8442-8449 Oct, 2021
Subject
Language
ISSN
2377-3766
2377-3774
2377-3774
Abstract
The performance of a model-based controller can severely suffer when its model inaccurately represents the real world dynamics. We propose to learn a time-varying, locally linear residual model along the robot’s current trajectory, to compensate for the prediction errors of the controller’s model. Supervised learning is performed online, as the robot is running in the unknown environment, using data collected from its immediate past. We theoretically investigate our method in its general formulation, then apply it to a bipedal controller derived from the full-order dynamics of virtual constraints, and a quadrupedal controller derived from a simplified model of contact forces. For a biped in simulation, our method consistently outperforms the baseline and a recent learning-based method. We also experiment with a 12 kg quadruped in simulation and real world, where the baseline fails to walk with 10 kg of payload but our method succeeds.