CaltechAUTHORS
  A Caltech Library Service

Learning Controller Gains on Bipedal Walking Robots via User Preferences

Csomay-Shanklin, Noel and Tucker, Maegan and Dai, Min and Reher, Jenna and Ames, Aaron D. (2021) Learning Controller Gains on Bipedal Walking Robots via User Preferences. . (Unpublished) https://resolver.caltech.edu/CaltechAUTHORS:20210304-095906281

[img] PDF - Submitted Version
See Usage Policy.

5MB

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20210304-095906281

Abstract

Experimental demonstration of complex robotic behaviors relies heavily on finding the correct controller gains. This painstaking process is often completed by a domain expert, requiring deep knowledge of the relationship between parameter values and the resulting behavior of the system. Even when such knowledge is possessed, it can take significant effort to navigate the nonintuitive landscape of possible parameter combinations. In this work, we explore the extent to which preference-based learning can be used to optimize controller gains online by repeatedly querying the user for their preferences. This general methodology is applied to two variants of control Lyapunov function based nonlinear controllers framed as quadratic programs, which have nice theoretic properties but are challenging to realize in practice. These controllers are successfully demonstrated both on the planar underactuated biped, AMBER, and on the 3D underactuated biped, Cassie. We experimentally evaluate the performance of the learned controllers and show that the proposed method is repeatably able to learn gains that yield stable and robust locomotion.


Item Type:Report or Paper (Discussion Paper)
Related URLs:
URLURL TypeDescription
http://arxiv.org/abs/2102.13201v1arXivDiscussion Paper
ORCID:
AuthorORCID
Csomay-Shanklin, Noel0000-0002-2361-1694
Tucker, Maegan0000-0001-7363-6809
Reher, Jenna0000-0002-8297-3847
Ames, Aaron D.0000-0003-0848-3177
Additional Information:This research was supported by NSF NRI award 1924526, NSF award 1932091, NSF CMMI award 1923239, NSF Graduate Research Fellowship No. DGE-1745301, and the Caltech Big Ideas and ZEITLIN Funds.
Funders:
Funding AgencyGrant Number
NSFECCS-1924526
NSFCNS-1932091
NSFCMMI-1923239
NSF Graduate Research FellowshipDGE-1745301
Caltech Big Ideas FundUNSPECIFIED
Zeitlin Family Discovery FundUNSPECIFIED
Record Number:CaltechAUTHORS:20210304-095906281
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20210304-095906281
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:108310
Collection:CaltechAUTHORS
Deposited By: Tony Diaz
Deposited On:04 Mar 2021 21:32
Last Modified:04 Mar 2021 21:32

Repository Staff Only: item control page