中文摘要：

智能车周围车辆目标参数（位姿、运动速度和几何形状）对智能车路径规划和决策算法而言至关重要.选取三维激光雷达作为传感器,对车辆目标位置、航向角、速度和几何形状进行计算和滤波.针对车辆目标点云受观测点位置和自遮挡等因素的影响,提出一种动态参考点模型用于计算目标速度;对于目标航向角,采用先点云分块聚类后主成分拟合的思路来提高航向角精度;提出一种基于几何形状变化速率的滤波算法来解决几何形状不易测量和复杂多变等问题.分别就车辆目标速度、航向角和几何形状进行实验分析,结果证明基于本文方法的参数计算结果能获得较高精度,满足智能车环境感知要求.

英文摘要：

The parameters of vehicle targets surrounding intelligent vehicle, including the position and pose,movement velocity and geometric shape are important for the path planning and decision making algorithms of intelligent vehicle. With 3D laser radar chosen as sensor,the position, azimuth angle, velocity and geometric shape of vehicle targets are calculated and filtered. In view of that the point cloud of vehicle targets is affected by the posi-tion of observation point and self- occlusion,a dynamic reference point model is proposed for calculating target velocity. A scheme of block clustering first then principal component fitting is adopted for increasing the accuracy of azi-muth angle,and a filtering algorithm based on the changing rate of geometric shape is proposed to tackle the prob-lem that geometric shape is complex,changeable and difficult to measure. Finally experimental analyse are conducted on the speed,azimuth angle and geometric shape of vehicle targets repsectively with a result verifying that with the parameters calculated by the method proposed,the higher accuracy can be achieved and the requirements of en-vironmental perception for intelligent vehicle can be met.