Joint-Health-Aware Reinforcement Learning for Robust Robot Control Under Actuator Degradation
Joint-Health-Aware Reinforcement Learning for Robust Robot Control Under Actuator Degradation
This paper introduces a joint-health-aware reinforcement learning framework designed to maintain robust robot control despite actuator degradation. The method encodes actuator condition in a continuous joint-health vector and models degradation as multiplicative attenuation of actions, using a HealthNet estimator to infer per-joint health from observations. Evaluated on Walker, Go2, Cassie, and Panda-based tasks under diverse degradation masks, the approach outperforms nominal baselines, with task-level locomotion returns improving by up to 250.1% and manipulation success rates by up to 15.2 percentage points relative to the nominal baseline.
The study presents a reinforcement learning framework that models continuous actuator degradation via a joint-health vector and redistributes control effort, evaluated on simulated locomotion and manipulation tasks with reported improvements up to 928.0% in return under high-degradation conditions.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Environment
- SimulationOpen source S3
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ten without explicitly modeling continuous per-joint residual actuation capability. We propose a joint-health-aware reinforcement learning framework for robust robot control under simulated actuator degradation. Our framework encodes actuator condition in a continuous joint-health vector and models degradation as multiplicative attenuation of executed actions. A dedicated HealthNet estimator infers per-joint health f
Open source S3
locomotion returns improve by up to 250.1\% and manipulation success rates by up to 15.2 percentage points; in the high-degradation Cassie group, return increases from 3.94 to 40.50 (+928.0\%). Ablations support the contribution of per-joint health estimation, action-degradation modeling, and health-conditioned control.
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org