RMA: Rapid Motor Adaptation for Legged Robots

Ashish Kumar1
Zipeng Fu2
Deepak Pathak2
Jitendra Malik1,3

1 University of California at Berkeley, 2 Carnegie Mellon University,3 Facebook AI Research


Paper             Supplementary


Successful real-world deployment of legged robots would require them to adapt in real-time to unseen scenarios like changing terrains, changing payloads, wear and tear. This paper presents Rapid Motor Adaptation (RMA) algorithm to solve this problem of real-time online adaptation in quadruped robots. RMA consists of two components: a base policy and an adaptation module. The combination of these components enables the robot to adapt to novel situations in fractions of a second. RMA is trained completely in simulation without using any domain knowledge like reference trajectories or predefined foot trajectory generators and is deployed on the A1 robot without any fine-tuning. We train RMA on a varied terrain generator using bioenergetics-inspired rewards and deploy it on a variety of difficult terrains including rocky, slippery, deformable surfaces in environments with grass, long vegetation, concrete, pebbles, stairs, sand, etc. RMA shows state-of-the-art performance across diverse real-world as well as simulation experiments. Method Video: https://youtu.be/qKF6dr_S-wQ





Selected Ourdoor Runs:




Selected Indoor Runs:




Rocky and Pebble Terrain:




Unstable and Loose Ground:




Vegetation and Grass Patch and Hiking Stairs:




Oil Slip Test:




Plank Test:




Foam and Mattress:




Step-Up and Step-Down:




Comparison with A1 controller:

A1 controller:

RMA:

Acknowledgments

We would like to thank Jemin Hwangbo for helping with the simulation platform, and Koushil Sreenath and Stuart Anderson for helpful feedback during the course of this project. We would also like to thank Claire Tomlin, Shankar Sastry, Chris Atkeson, Aravind Sivakumar, Ilija Radosavovic and Russell Mendonca for their high quality feedback on the paper. This research was part of a BAIR-FAIR collaborative project, and recently also supported by the DARPA Machine Common Sense program. This webpage template was borrowed from some colorful folks.