基于栅格地图的智能车辆运动目标检测

运动目标检测是智能车辆动态环境感知中的关键问题。栅格地图是一种实用的环境感知方法。以激光雷达作为传感器,针对基于贝叶斯框架和证据理论的4种栅格地图更新算法,分别提出了不同的运动目标检测方法,并通过仿真和实验对4种方法在栅格地图构建和运动目标检测中的性能进行了对比研究。实验结果表明,通过对比当前时刻局部栅格地图和上一个时刻的全局栅格地图,原始的贝叶斯推理算法和证据理论框架中的经典Dempster组合规则能够清楚完整地检测到运动目标,并能滤除静态障碍物和空白区域中的测量噪声,性能优于修正的贝叶斯推理算法和冲突信息按比例重分配规则。

Moving object detection for intelligent vehicles based on occupancy grid map

Moving object detection is a key point in dynamic environment perception. Occupancy grid mapping is a practical approach for the dynamic environment perception. Different moving object detection methods in the occupancy grid map are proposed aiming at four uncertain reasoning algorithms in the Bayesian framework and the evidential framework. The moving object detection methods are comparably analyzed by simulation, and they are also tested on real range data of the outdoor dynamic environment acquired by a multi layer laser scanner. Experiment results show that, by comparing with the local map at the current moment and the global map at the former moment, the original Bayesian inference and the classical Dempster’s rule of combination are able to clearly and wholly detect the moving objects, and filter out the measurement noise in static objects and free space.