analiza aiqbench
This article details CARO, a novel reinforcement learning framework designed to enhance the robustness of quadruped robots through contact-agnostic residual observation. By embedding a fixed-base Euler-Lagrange model directly into the control loop, the system achieves zero-shot adaptation to out-of-distribution payloads and terrain changes without relying on torque sensors or explicit contact estimation. This represents a significant advancement in sim-to-real transfer capabilities for autonomous locomotion systems.
CARO: Contact-Agnostic Residual Observation for Zero-Shot Robust Quadruped Locomotion
arXiv:2608.24217v1 Announce Type: new Abstract: 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.
Abstract: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.
Robotics (cs.RO)
arXiv:2608.24217 [cs.RO]
(or arXiv:2608.24217v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2608.24217 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 209 cuvinte
Subiecte conexe
TAM: Torque Adaptation Module for Robust Motion Transfer in Manipulation
arXiv · 6 iun. 2026
TAM: Torque Adaptation Module for Robust Motion Transfer in Manipulation
arXiv · 9 sept. 2026
The Machines that Make the Machines
NVIDIA · 7 oct. 2026
OATH: Object-Agnostic Tracking for Dexterous Hands, Enabling Rapid Physics-based Retargeting
arXiv · 14 oct. 2026