基于激光雷达的智能车前方障碍物检测.docVIP

基于激光扫描仪的智能车前方障碍物检测 作者一1, 2 作者二2 作者三1 （西安交通大学综合自动化研究所 西安 710049） （E-mail: xxx@xxx.cn, xxxr@ia.ac.cn） 摘 要： 无人驾驶智能车在非结构化环境下的环境感知是其进行自主导航的关键问题之一[1]。智能车辆要实现在未知环境下的自主导航，必须获取到有效并且可靠的环境信息，其中包括判断哪些区域存在障碍物，哪些区域不存在障碍物。这些也是了实现环境建模、定位、路径规划等工作打好基础。这就要求将障碍物和可行地面数据进行有效分离。传统的地物分离方法是高度阈值法和空间聚类法。高度阈值法直接滤除阈值外数据，这种算法只适用于室内或室外比较平坦的环境，对于越野情况就不再适合。空间聚类法是一种很耗时的算法，如果没有任何预处理，所花费的时间将是不可接受的。鉴于以上原因，本文提出了一种基于坡度检测和聚类分析的障碍物检测方法。同时，为了增强算法的鲁棒性和坡度检测的稳定性，算法中加入了平面条带法和距离滤波方法，收到了较好的效果。 关键词：激光扫描仪，障碍物检测，无人驾驶智能车,环境感知 中图分类号：TP273 Obstacle Detection For Intelligent Vehicle Based on Laser Radar FIRST Author-Aa1,2 SECOND Author-Bb2 THIRD Author-Cc1 (Institute Of Integrated Automation,Xi’an Jiaotong University, Xi’an 710049,China) (E-mail: aas@ia.ac.cn, aas_editor@ia.ac.cn) Abstract The environmental sensing of Unmanned Intelligent Vehicles in unstructured environments is a key issue for autonomous navigation. To achieving the autonomous navigation, we must get effective and reliable environmental information, including determining which areas exist obstructions, which areas doesn’t. These things is for the aim of achieving environmental modeling, locationing, path planning, etc. This is required to sepatate the obstruction datas and the feasible ground datas.The traditional separation method is altitude threshold method and space clustering method. Altitude threshold method is directly to filter external data, this algorithm only applies to indoor or relatively flat outdoor environment, for off-road situation is no longer appropriate. Spatial clustering method is a very time-consuming algorithm, if there is no pre-treatment, the time would be unacceptable.Due to the above reasons,we found a obstacle detection method which is based on slope detecting and cluster analysising.Meanwhile, in order to enhance the robustness and stability of the slope detection algorithm ,we added a flat strip method and distance filtering method, we received a better results. Key words Laser Radar, Obstacle D