基于多线激光雷达的障碍检测与环境重建

Obstacle detection and map reconstruction using multi-layer laser radar

针对越野环境中自主导航车(ALV)的导航问题,提出了基于卡尔曼滤波理论的障碍检测算法.通过分析多线激光雷达的扫描数据,依次进行候选障碍点提取、非障碍扫描点滤除和聚类分割获取障碍信息,使用卡尔曼滤波算法对目标障碍位置进行跟踪和滤波处理,并结合自主导航车上差分全球定位系统/惯性导航系统(DGPS/INS)的位置和姿态信息进行联合定位,对ALV途经的周围环境进行了三维地图重建.试验结果表明,该算法稳定可靠,跟踪的位置误差可以降低到10 cm以内,有效地解决了越野环境下的障碍检测问题,同时成功地完成了环境场景的重建.

Aimed at the autonomous land vehicle （ALV） navigation problem in a cross-country environment, an obstacle detection algorithm based on Kalman filter theory was proposed. Scan data from multilayer laser radar were analyzed. After the candidate obstacle points were extracted, the non-obstacle scanning points were eliminated, and then the obstacle points were clustered and segmented to retrieve obstacles information. The positions of object obstacles were filtered and tracked using Kalman filter algorithm. By combining the ALV position and pose information from difference global position system/inertial navigation system （DGPS/INS）, a three-dimension environment map along ALV route was rebuilt. Experimental results show that the proposed algorithm is robust and reliable, and that the tracking position errors decrease below 10 cm. The algorithm can solve the obstacle detecting problem in cross-country environment, and rebuild the environment map successfully.