基于支持向量机的局部路径规划算法  

A local path planning method based on support vector machine

作　　者：诸葛程晨 许劲松 唐振民[1] ZHUGE Chengchen;XU Jinsong;TANG Zhenmin(School of Computer Science and Engineering Nanjing University of Science and Technology,Nanjing 210094,China;School of Electrical and Data Engineering Sydney University of Technology,Sydney 2007,Australia)

机构地区：[1]南京理工大学计算机科学与技术学院,江苏南京210094 [2]悉尼科技大学大数据中心

出　　处：《哈尔滨工程大学学报》2019年第2期323-330,共8页Journal of Harbin Engineering University

基　　金：国家自然科学基金项目(61403202;61371040);中国博士后科学基金面上项目(2014M561654);高等学校博士学科点专项科研基金项目(20133219120035);核高基国家重大专项(2015ZX01041101)

摘　　要：针对在非结构化道路环境下地面自主车辆的路径规划问题,本文提出了一种基于64激光雷达以及支持向量机的局部路径规划算法。利用非线性支持向量机分类器在栅格地图上提取出安全的路径;并在多帧投影数据上使用RANSAC算法提取出路径,并使用三次多项式进行描述,从而计算出道路曲率;结合地面自主车辆自身状态在RANSAC路径上选取控制点,并使用贝塞尔曲线拟合出最终路径。该算法能够有效地从局部栅格地图中提取道路,以弥补基于视觉的道路检测算法在受到恶劣光照、天气影响时的不足。通过实车试验验证了所提出的方法的有效性和正确性。To solve the path-planning problem of unmanned ground vehicles(UGV)in an unstructured road environment,a local path-planning method based on multiple Lidar and support vector machine(SVM)is proposed in this paper.First,a safe path is extracted from the grid map by nonlinear SVM.A path described with a cubic polynomial can be extracted by using RANSAC algorithm on the multi-frame projection data and the path can be used to calculate the road curvature.Then,the control points are selected by considering the state of the UGV and the final path is generated by using Bezier curve fitting.The proposed method can effectively extract the path from the grid map when the performance of the vision-based road detection algorithms is limited by poor lighting and weather.Finally,the validity and correctness of the proposed algorithm is verified on a real vehicle.

关 键 词：非结构化道路 地面自主车辆 路径规划 支持向量机 RANSAC算法 局部路径 栅格地图 

分 类 号：TP249[自动化与计算机技术—检测技术与自动化装置]