MHER: Model-based Hindsight Experience Replay
This addresses sample efficiency issues in multi-goal RL for robotics and simulation applications, representing an incremental improvement over existing relabeling methods.
The paper tackles the challenge of sparse rewards in multi-goal reinforcement learning by proposing Model-based Hindsight Experience Replay (MHER), which uses environmental dynamics to generate virtual goals for more efficient experience utilization, achieving significantly higher sample efficiency in point-based and robotics tasks.
Solving multi-goal reinforcement learning (RL) problems with sparse rewards is generally challenging. Existing approaches have utilized goal relabeling on collected experiences to alleviate issues raised from sparse rewards. However, these methods are still limited in efficiency and cannot make full use of experiences. In this paper, we propose Model-based Hindsight Experience Replay (MHER), which exploits experiences more efficiently by leveraging environmental dynamics to generate virtual achieved goals. Replacing original goals with virtual goals generated from interaction with a trained dynamics model leads to a novel relabeling method, model-based relabeling (MBR). Based on MBR, MHER performs both reinforcement learning and supervised learning for efficient policy improvement. Theoretically, we also prove the supervised part in MHER, i.e., goal-conditioned supervised learning with MBR data, optimizes a lower bound on the multi-goal RL objective. Experimental results in several point-based tasks and simulated robotics environments show that MHER achieves significantly higher sample efficiency than previous model-free and model-based multi-goal methods.