RF-Agent: Automated Reward Function Design via Language Agent Tree Search
This work addresses the problem of automating reward function design for low-level control tasks, offering a novel method that improves upon existing approaches by better utilizing historical feedback and search efficiency.
The paper tackles the challenge of designing efficient reward functions for low-level control tasks by proposing RF-Agent, a framework that uses language agents and Monte Carlo Tree Search to enhance optimization through better contextual reasoning, achieving outstanding results in 17 diverse tasks.
Designing efficient reward functions for low-level control tasks is a challenging problem. Recent research aims to reduce reliance on expert experience by using Large Language Models (LLMs) with task information to generate dense reward functions. These methods typically rely on training results as feedback, iteratively generating new reward functions with greedy or evolutionary algorithms. However, they suffer from poor utilization of historical feedback and inefficient search, resulting in limited improvements in complex control tasks. To address this challenge, we propose RF-Agent, a framework that treats LLMs as language agents and frames reward function design as a sequential decision-making process, enhancing optimization through better contextual reasoning. RF-Agent integrates Monte Carlo Tree Search (MCTS) to manage the reward design and optimization process, leveraging the multi-stage contextual reasoning ability of LLMs. This approach better utilizes historical information and improves search efficiency to identify promising reward functions. Outstanding experimental results in 17 diverse low-level control tasks demonstrate the effectiveness of our method. The source code is available at https://github.com/deng-ai-lab/RF-Agent.