Search papers, labs, and topics across Lattice.
CARO introduces a novel contact-agnostic residual observation framework that enhances policy adaptation for quadruped locomotion by integrating a fixed-base Euler鈥揕agrange model into the reinforcement learning control loop. This approach allows for torque-level residual observation without the need for torque sensors or explicit contact estimation, enabling the policy to adapt to dynamic changes in the environment effectively. The framework demonstrates significant improvements in zero-shot robustness during simulation and real-world transfer tasks, outperforming existing methods in handling various disturbances and out-of-distribution scenarios.
Achieving zero-shot robustness in quadruped locomotion without the need for contact sensors or specialized adaptation supervision is a game changer for real-world applications.
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.
