Xiru Gao

EvoHIL: Self-Evolving Reward and Flow-Matched Policy Optimization for Robust Human-in-the-Loop Reinforcement Learning

EvoHIL is presented, a unified framework that adapts the reward model, action generator, and visual do main within a staged human-in-the-loop learning process to improve task success, agreement with human-confirmation labels, motion smoothness, and completion time relative to human-in-the-loop and imitation baselines.