RK-MPC: Residual Koopman Model Predictive Control for Quadruped Locomotion in Offroad Environments

Sriram S. K. S. Narayanan, Umesh Vaidya · 2026

Read RK-MPC: Residual Koopman Model Predictive Control for Quadruped Locomotion in Offroad Environments by Sriram S. K. S. Narayanan, Umesh Vaidya (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-04221.txt.

Description

Narayanan and Vaidya build a controller that helps four-legged robots walk blind over rough ground. Their RK-MPC method pairs a simple template model with a small data-learned correction for contact and terrain surprises, then runs the combined model inside a predictive controller at 500 hertz on board. Tests in simulation and on a Unitree Go1 robot cover grass, gravel, snow and ice, and the correction improves prediction over standard Koopman baselines.

Catalog record

CREATORSriram S. K. S. Narayanan, Umesh Vaidya
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE87090
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.04221

Browse the full library · About the Tower of Babel