Scholarlink

Humans are increasingly integrated into system feedback loops to enhance intelligence and assist AI agent learning through various means, including calibration and supervision. This paper introduces a human experience gain model to migrate the human prior experience. Combining the Radial Basis Function (RBF) neural-network (NN) is employed to approximate the nonlinear Hamilton-Jacobi-Bellman (HJB) equation for leader-follower multi-agent system (MASs). Additionally, the optimized control design incorporates an actor-critic architecture responsible for both executing behaviors and evaluating control performance. The primary result is exemplified through an illustrative example, demonstrating the effectiveness of the proposed algorithm.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Reinforcement Learning for Consensus Tracking Control of Multi-agent Systems

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Luning Yang,
  • Jiang Zhao,
  • Chi Pei,
  • Yingxun Wang

摘要

Humans are increasingly integrated into system feedback loops to enhance intelligence and assist AI agent learning through various means, including calibration and supervision. This paper introduces a human experience gain model to migrate the human prior experience. Combining the Radial Basis Function (RBF) neural-network (NN) is employed to approximate the nonlinear Hamilton-Jacobi-Bellman (HJB) equation for leader-follower multi-agent system (MASs). Additionally, the optimized control design incorporates an actor-critic architecture responsible for both executing behaviors and evaluating control performance. The primary result is exemplified through an illustrative example, demonstrating the effectiveness of the proposed algorithm.