Saved in:
Bibliographic Details
Main Authors: Ma, Hao, Wang, Shijie, Pu, Zhiqiang, Zhao, Siyao, Ai, Xiaolin
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2502.13430
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910835546259456
author Ma, Hao
Wang, Shijie
Pu, Zhiqiang
Zhao, Siyao
Ai, Xiaolin
author_facet Ma, Hao
Wang, Shijie
Pu, Zhiqiang
Zhao, Siyao
Ai, Xiaolin
contents Guiding the policy of multi-agent reinforcement learning to align with human common sense is a difficult problem, largely due to the complexity of modeling common sense as a reward, especially in complex and long-horizon multi-agent tasks. Recent works have shown the effectiveness of reward shaping, such as potential-based rewards, to enhance policy alignment. The existing works, however, primarily rely on experts to design rule-based rewards, which are often labor-intensive and lack a high-level semantic understanding of common sense. To solve this problem, we propose a hierarchical vision-based reward shaping method. At the bottom layer, a visual-language model (VLM) serves as a generic potential function, guiding the policy to align with human common sense through its intrinsic semantic understanding. To help the policy adapts to uncertainty and changes in long-horizon tasks, the top layer features an adaptive skill selection module based on a visual large language model (vLLM). The module uses instructions, video replays, and training records to dynamically select suitable potential function from a pre-designed pool. Besides, our method is theoretically proven to preserve the optimal policy. Extensive experiments conducted in the Google Research Football environment demonstrate that our method not only achieves a higher win rate but also effectively aligns the policy with human common sense.
format Preprint
id arxiv_https___arxiv_org_abs_2502_13430
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Vision-Based Generic Potential Function for Policy Alignment in Multi-Agent Reinforcement Learning
Ma, Hao
Wang, Shijie
Pu, Zhiqiang
Zhao, Siyao
Ai, Xiaolin
Artificial Intelligence
Machine Learning
Guiding the policy of multi-agent reinforcement learning to align with human common sense is a difficult problem, largely due to the complexity of modeling common sense as a reward, especially in complex and long-horizon multi-agent tasks. Recent works have shown the effectiveness of reward shaping, such as potential-based rewards, to enhance policy alignment. The existing works, however, primarily rely on experts to design rule-based rewards, which are often labor-intensive and lack a high-level semantic understanding of common sense. To solve this problem, we propose a hierarchical vision-based reward shaping method. At the bottom layer, a visual-language model (VLM) serves as a generic potential function, guiding the policy to align with human common sense through its intrinsic semantic understanding. To help the policy adapts to uncertainty and changes in long-horizon tasks, the top layer features an adaptive skill selection module based on a visual large language model (vLLM). The module uses instructions, video replays, and training records to dynamically select suitable potential function from a pre-designed pool. Besides, our method is theoretically proven to preserve the optimal policy. Extensive experiments conducted in the Google Research Football environment demonstrate that our method not only achieves a higher win rate but also effectively aligns the policy with human common sense.
title Vision-Based Generic Potential Function for Policy Alignment in Multi-Agent Reinforcement Learning
topic Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2502.13430