【作者】 高宇轩； 顾波； 樊卫华； 俞方罡； 吕强华；

【Author】 GAO Yuxuan;GU Bo;FAN Weihua;YU Fanggang;LV Qianghua;School of Automation, Nanjing University of Science and Technology;China Construction Eighth Engineering Division Third Construction Co., Ltd.;

【机构】 南京理工大学自动化学院； 中建八局第三建设有限公司；

【摘要】 针对传统Voronoi图路径规划算法在处理大规模地图时计算复杂度高、路径不够平滑等问题,提出了一种基于智能截取和B样条平滑的优化算法。首先,建立了包含代价地图截取、代价地图下采样、死胡同路径剪枝和B样条平滑的完整算法框架;随后,设计了基于起终点位置的动态截取策略,通过计算边界框和安全边距保护实现计算量的大幅减少;接着提出了递归式死胡同路径识别与剪枝算法,有效移除Voronoi图中的无效分支;最后,采用3阶B样条曲线对路径进行平滑处理,提高路径的连续性和可执行性。实验结果表明,优化算法在保证路径安全性的同时,将规划时间减少了84.9%,路径平滑度提升了80%以上,为移动机器人的实时导航提供了有效的解决方案。更多还原

【Abstract】 Aiming at the problems of high computational complexity and insufficient path smoothness in traditional Voronoi diagram path planning algorithms when processing large-scale maps, an optimization algorithm based on intelligent cropping and B-spline smoothing is proposed. Firstly, a complete algorithm framework including costmap cropping, costmap downsampling, dead-end pruning and B-spline smoothing is established. Then, a dynamic cropping strategy based on start-goal positions is designed, achieving significant computational reduction through bounding box calculation and safety margin protection. Subsequently, a recursive dead-end identification and pruning algorithm is proposed to effectively remove invalid branches in Voronoi diagrams. Finally, 3rd-order B-spline curves are used for path smoothing to improve path continuity and executability. Experimental results show that the optimized algorithm reduces planning time by 84.9% while ensuring path safety, improves path smoothness by 80%, providing an effective solution for real-time navigation of mobile robots.更多还原