面向非结构化环境的改进APF-RRT*路径规划算法

An Improved APF-RRT* Path Planning Algorithm in Unstructured Environments

摘要: 针对自动驾驶车辆在非结构化环境中路径规划所面临的效率低、安全性差及可行性不足等问题，本文提出了一种融合APF与RRT*的改进规划算法。通过构建参数可调的引力与斥力势场模型，并引入距离加权修正系数，有效抑制了传统APF的局部极小问题。同时将势场梯度信息嵌入RRT*随机采样过程，以优化节点扩展方向，提升算法收敛速度与路径质量。进一步地，考虑车辆动力学约束，确保生成路径符合实际操纵特性。多场景仿真结果表明，所提算法相较于传统RRT*及普通APF-RRT*算法，在避障安全性、路径平滑性方面均有提升，同时保持了具有竞争力的规划效率，验证了其在非结构化环境中的优越性与鲁棒性。

Abstract: Addressing the challenges of low efficiency, poor safety, and insufficient feasibility in path planning for autonomous vehicles within unstructured environments, this paper proposes an improved planning algorithm that integrates the APF method with RRT*. A parameter-adjustable attractive and repulsive potential field model is constructed, and a distance-weighted correction factor is introduced to effectively mitigate the local minimum problem inherent in traditional APF methods. Simultaneously, potential field gradient information is embedded into the RRT* random sampling process to optimize the node expansion direction, thereby enhancing the algorithm convergence speed and path quality. Furthermore, vehicle dynamics constraints are incorporated to ensure the generated path complies with practical maneuvering characteristics. Multi-scenario simulation results demonstrate that, compared with conventional RRT* and standard APF-RRT* algorithms, the proposed algorithm achieves improvements in obstacle avoidance safety and path smoothness while maintaining competitive planning efficiency, effectively validating its superiority and robustness in unstructured environments.