Scholarlink

<p>This paper studies the optimal path tracking control problem for autonomous vehicles subject to multiple constraints and obstacle avoidance requirements. To address prescribed constraints, an adaptive backstepping tracking controller is developed using a unified barrier Lyapunov function approach, which ensures the tracking errors, longitudinal/lateral velocity, and yaw rate are all strictly confined within the predefined ranges. Specifically, a tunnel prescribed performance function is used to quantitatively regulate the overshoot and settling time of the tracking error, and a dynamic auxiliary term is incorporated to relax the original constraint boundaries, thereby prioritizing safety during obstacle avoidance maneuvers. Furthermore, the obstacle avoidance constraint is formulated using barrier functions integrated into an extended value function as a risk penalty term. A critic-only adaptive dynamic programming (ADP) scheme is then introduced to obtain an optimal safety modified policy, which adjusts the tracking controller with minimal intervention to guarantee obstacle avoidance behavior and accurate path tracking. Meanwhile, a fixed-time learning law is presented to relax the persistent excitation (PE) condition and satisfy the real-time computational and fast dynamic response requirements for obstacle avoidance. Simulation results confirm the effectiveness and superior performance of the proposed control strategy.</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-based optimal path tracking control for obstacle avoidance of autonomous vehicles with multiple prescribed constraints

  • 期刊论文

  • 发表日期：

    2026年8月3日

查看全文
  • Xin Zhou,
  • Heng Wang,
  • Zhonghua Miao,
  • Yulong Wang,
  • Qing Li

摘要

This paper studies the optimal path tracking control problem for autonomous vehicles subject to multiple constraints and obstacle avoidance requirements. To address prescribed constraints, an adaptive backstepping tracking controller is developed using a unified barrier Lyapunov function approach, which ensures the tracking errors, longitudinal/lateral velocity, and yaw rate are all strictly confined within the predefined ranges. Specifically, a tunnel prescribed performance function is used to quantitatively regulate the overshoot and settling time of the tracking error, and a dynamic auxiliary term is incorporated to relax the original constraint boundaries, thereby prioritizing safety during obstacle avoidance maneuvers. Furthermore, the obstacle avoidance constraint is formulated using barrier functions integrated into an extended value function as a risk penalty term. A critic-only adaptive dynamic programming (ADP) scheme is then introduced to obtain an optimal safety modified policy, which adjusts the tracking controller with minimal intervention to guarantee obstacle avoidance behavior and accurate path tracking. Meanwhile, a fixed-time learning law is presented to relax the persistent excitation (PE) condition and satisfy the real-time computational and fast dynamic response requirements for obstacle avoidance. Simulation results confirm the effectiveness and superior performance of the proposed control strategy.