Almanac
paper

UniIntervene: Agentic Intervention for Efficient Real-World Reinforcement Learning

paperactiveprovisionaluniintervene-agentic-intervention-for-efficient-real-world-reinforcement-learning-09d9248f·1 events·first seen 6d ago

Aliases: UniIntervene: Agentic Intervention for Efficient Real-World Reinforcement Learning

Co-occurring entities

More like this (12)

Recent events (1)

5arXiv · cs.LG·6d ago·source ↗

UniIntervene: Agentic model reduces human interventions in real-world robot RL by 57%

UniIntervene is a proposed agentic intervention model for human-in-the-loop reinforcement learning (HiL-RL) that autonomously detects unproductive exploration and recovers robot policies toward high-value states, replacing the bulk of human corrections. The system uses future-conditioned action-value estimation, a temporal value-risk critic, and a goal-conditioned recovery policy drawing from a memory of past interventions. Experiments on real-world robotic manipulation tasks show a 57% reduction in human interventions and an 8.6% improvement in average success rate over state-of-the-art HiL-RL baselines.