Scholarlink

Traditional parameter tuning of cascade controllers heavily depends on practical experiments and expert experience, which can be both time-consuming and complex. This study proposes a feedforward and cascade feedback PID control scheme to address vehicle trajectory-following challenges and develops a Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm for adaptive tuning of controller parameters. A range of scenarios and control strategies are simulated and evaluated through joint simulations in Matlab/Simulink and Carsim. The simulation results demonstrate that the proposed method effectively adjusts controller coefficients online through learning, reduces calibration time, and enhances trajectory-following performance and adaptability compared to alternative control approaches.

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

Adaptive Cascade Control for Vehicle Trajectory Following Based on Deep Reinforcement Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yong Lu,
  • Diange Yang,
  • He Tian,
  • Kun Jiang,
  • Runfeng Li

摘要

Traditional parameter tuning of cascade controllers heavily depends on practical experiments and expert experience, which can be both time-consuming and complex. This study proposes a feedforward and cascade feedback PID control scheme to address vehicle trajectory-following challenges and develops a Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm for adaptive tuning of controller parameters. A range of scenarios and control strategies are simulated and evaluated through joint simulations in Matlab/Simulink and Carsim. The simulation results demonstrate that the proposed method effectively adjusts controller coefficients online through learning, reduces calibration time, and enhances trajectory-following performance and adaptability compared to alternative control approaches.