Skip to content

Event

Research proposes joint health-aware reinforcement learning framework

1 reports · 1 sources

Haifa Zhang and other authors propose a joint health-aware reinforcement learning framework for robust control of robots under actuator degradation. The framework encodes actuator states with a continuous joint health vector, models degradation as multiplicative attenuation applied to executed actions, and uses HealthNet to estimate the health of each joint from recent observations and actions, enabling the policy to redistribute control effort between healthy and degraded joints. The authors state that it has been evaluated on Walker, Go2, Cassie, and Panda-based ManiSkill manipulation tasks under various degradation masks and severity levels; these results come from simulated degradation environments.

Report timeline

  1. Joint-Health-Aware Reinforcement Learning Framework for Robust Control Under Actuator Degradation

    The study proposes a joint-health-aware reinforcement learning framework that encodes actuator condition in a continuous joint-health vector and uses HealthNet to estimate each joint's health from recent observations and actions, allowing the policy to redistribute control effort across healthy and degraded joints.

    arXiv cs.RO ·

Related topics

Motion controlRobot models and learningSafety and reliabilityResearch papers

Interest trend

Current
3
Peak
4Nov 4 01:00
Past 24 hours
–

The trend compares participants observed throughout the same period, so its range may be smaller than the current interest count. Move or tap to inspect each hour; use the arrow keys on a keyboard.