Scholarlink

<p>The artificial potential field (APF) method used in path planning is prone to falling into local optimum solutions, exhibiting goal unreachability, and failing to incorporate road boundary potential fields. To address these issues, an intelligent vehicle’s autonomous lane-changing behavior decision-making and motion planning method based on the combination of the APF method and quadratic programming (QP) is proposed. At the level of path pre-planning, an improved APF method is utilized to preprocess potential collision areas. At the level of path re-planning, the QP method is employed to refine the uncertain areas obtained from preprocessing, resulting in a safe, collision-free vehicle travel path that satisfies dynamic constraints. Simulation results demonstrate that this method successfully avoids all obstacles. Compared to the APF algorithm before improvement, the accuracy in static and dynamic obstacle scenarios increases by 6% and 22%. Additionally, the trajectory smoothness increases by approximately 30%, and the average time required within one calculation cycle is reduced by 4.8&#xa0;ms. Real-vehicle tests further confirm that this method exhibits superior scene generalization performance.</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

Decision planning for intelligent vehicles in obstacle avoidance using APF-QP methods

  • 期刊论文

  • 发表日期：

    2025年11月2日

查看全文
  • Minrui Ma,
  • Bin Huang,
  • Liutao Ma,
  • Xu Yang

摘要

The artificial potential field (APF) method used in path planning is prone to falling into local optimum solutions, exhibiting goal unreachability, and failing to incorporate road boundary potential fields. To address these issues, an intelligent vehicle’s autonomous lane-changing behavior decision-making and motion planning method based on the combination of the APF method and quadratic programming (QP) is proposed. At the level of path pre-planning, an improved APF method is utilized to preprocess potential collision areas. At the level of path re-planning, the QP method is employed to refine the uncertain areas obtained from preprocessing, resulting in a safe, collision-free vehicle travel path that satisfies dynamic constraints. Simulation results demonstrate that this method successfully avoids all obstacles. Compared to the APF algorithm before improvement, the accuracy in static and dynamic obstacle scenarios increases by 6% and 22%. Additionally, the trajectory smoothness increases by approximately 30%, and the average time required within one calculation cycle is reduced by 4.8 ms. Real-vehicle tests further confirm that this method exhibits superior scene generalization performance.