Scholarlink

<p><b>−</b>To solve the “last mile” problem in autonomous driving and curvature discontinuity during parking process, an improved path planning method combining Hybrid A* and geometric curves is proposed. Traditional algorithms such as A* and RRT* often produce paths with excessive length and discontinuous curvature, leading to infeasible maneuvers during low-speed precise parking. First, for global path planning, an evaluation function is established to compare three algorithms, and the Hybrid A* algorithm is selected and optimized using the time elastic band algorithm. Second, the clothoid curve is utilized for local path planning to smoothly transition at the abrupt curvature change in the C-shaped parking path, ensuring compliance with curvature constraints. Third, the vertical parking path is tracked using MPC with PSO-optimized weight coefficients, resulting in a maximum tracking error of 0.0246&#xa0;m for the optimized path. Finally, real vehicle experiments based on the Apollo D-KIT platform validate the effectiveness of the proposed path planning and tracking methods.</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

Vertical Parking Path Planning Method Based on Improved Hybrid A* Algorithm and Geometric Curve

  • 期刊论文

  • 发表日期：

    2026年1月26日

查看全文
  • Leiyan Yu,
  • Xiaobin Liu,
  • Yongpeng Cai,
  • Haijie Huang,
  • Xinyue Wu

摘要

−To solve the “last mile” problem in autonomous driving and curvature discontinuity during parking process, an improved path planning method combining Hybrid A* and geometric curves is proposed. Traditional algorithms such as A* and RRT* often produce paths with excessive length and discontinuous curvature, leading to infeasible maneuvers during low-speed precise parking. First, for global path planning, an evaluation function is established to compare three algorithms, and the Hybrid A* algorithm is selected and optimized using the time elastic band algorithm. Second, the clothoid curve is utilized for local path planning to smoothly transition at the abrupt curvature change in the C-shaped parking path, ensuring compliance with curvature constraints. Third, the vertical parking path is tracked using MPC with PSO-optimized weight coefficients, resulting in a maximum tracking error of 0.0246 m for the optimized path. Finally, real vehicle experiments based on the Apollo D-KIT platform validate the effectiveness of the proposed path planning and tracking methods.