Search papers, labs, and topics across Lattice.
The CARO framework introduces a contact-agnostic residual observation approach for enhancing policy adaptation in quadruped locomotion without relying on torque sensors or explicit contact estimation. By integrating a fixed-base Euler鈥揕agrange model into the reinforcement learning control loop, CARO allows the policy to adapt in real-time to dynamic disturbances and variations in the environment. The method demonstrates significant improvements in zero-shot robustness during simulation and sim-to-real transfer tasks, outperforming traditional approaches in challenging scenarios such as payload shifts and abrupt terrain changes.
Zero-shot robustness in quadruped locomotion can be achieved without specialized sensors or complex adaptation supervision, revolutionizing how we approach dynamic environments.
We propose CARO, a contact-agnostic residual observation framework for policy adaptation. CARO embeds a fixed-base Euler--Lagrange model into the reinforcement learning control loop and constructs a torque-level residual observation without requiring torque sensors, explicit contact estimation, or vision-based measurements of the floating-base position and linear velocity. A disturbance observer extracts a structured signal representing dynamics mismatch, while the policy learns to exploit this feedback for online adaptation. CARO is trained under the same terrain, command, and domain-randomization conditions as the nominal policy, without specialized disturbance curricula or additional adaptation supervision. Nevertheless, it achieves substantially improved zero-shot robustness in simulation and sim-to-real transfer tasks involving out-of-distribution payloads, center-of-mass shifts, terrain geometries, abrupt dynamics changes, and elevated-platform landings.
