Scholarlink

<p>This paper investigates the tracking control problem for hydraulic excavators with state constraints. First, an optimal control method is developed using the actor-critic reinforcement learning (RL) framework. In this framework, the critic is responsible for evaluating the system performance, and the actor is responsible for executing control actions. Subsequently, to ensure that state variables remain within predetermined safety bounds throughout the learning process, the Barrier Lyapunov Function (BLF) is combined with the optimized backstepping (OB) approach. Furthermore, a novel Lyapunov function is designed to eliminate the impact of unknown control gain, simplifying the controller design and improving robustness. By employing the Lyapunov stability theory for stability analysis, it is verified that all closed-loop system signals are semi-globally ultimately bounded. Finally, the efficacy of the devised control scheme is ultimately substantiated through numerical simulations and experimental trials conducted on the hydraulic excavator platform. To ensure repeatability, our codes are open-sourced on Github: <a href="https://github.com/JiannanChen/RL-based-OBC-of-UHE.git">https://github.com/JiannanChen/RL-based-OBC-of-UHE.git</a>.</p>

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

Safe reinforcement learning for tracking control of uncertain hydraulic excavators

  • 期刊论文

  • 发表日期：

    2025年7月4日

查看全文
  • Kuo Chen,
  • Linyu Tao,
  • Jiannan Chen,
  • Changchun Hua,
  • Qian Li

摘要

This paper investigates the tracking control problem for hydraulic excavators with state constraints. First, an optimal control method is developed using the actor-critic reinforcement learning (RL) framework. In this framework, the critic is responsible for evaluating the system performance, and the actor is responsible for executing control actions. Subsequently, to ensure that state variables remain within predetermined safety bounds throughout the learning process, the Barrier Lyapunov Function (BLF) is combined with the optimized backstepping (OB) approach. Furthermore, a novel Lyapunov function is designed to eliminate the impact of unknown control gain, simplifying the controller design and improving robustness. By employing the Lyapunov stability theory for stability analysis, it is verified that all closed-loop system signals are semi-globally ultimately bounded. Finally, the efficacy of the devised control scheme is ultimately substantiated through numerical simulations and experimental trials conducted on the hydraulic excavator platform. To ensure repeatability, our codes are open-sourced on Github: https://github.com/JiannanChen/RL-based-OBC-of-UHE.git.