一种Kalman最近邻的智能汽车传感器多目标跟踪算法  

Research on Post Processing Method of Intelligent Vehicle Sensors Based on Kalman's NN Algorithm

作　　者：黄文奎 罗峰[1,2] 胡凤鉴[1,2] 

机构地区：[1]同济大学汽车学院,上海201804 [2]同济大学智能型新能源汽车协同创新中心,上海201804

出　　处：《机电一体化》2017年第12期32-37,41,共7页Mechatronics

基　　金：上海市科学技术委员会项目"基于北斗导航的低速智能汽车关键技术研究"(14DZ1104700)

摘　　要：对清扫车的前向雷达和摄像头传感器的性能参数及软件算法进行了介绍,并采用基于Kalman的最近邻算法对车辆前方的障碍进行滤波跟踪,改进性地提出了基于马氏距离与夹角联合关联的判断,完成航迹的起始、跟踪和删除。环境感知系统分别通过模拟仿真数据和真实数据的测试,能够将处理后的环境信息发送给智能决策模块。经实车测试,对于低速运动的无人清扫车有较好的效果。The performance parameters is introduced about vehicle forward radar and camera sensor, and the software algorithm is described. The nearest neighbor algorithm based on Kalman is used to filter and track the obstacles. In addition, how to start or delete the track is followed up. The system can send the processed environment information to intelligent decision through the actual test based on hypothetical data and real data. It is proved to have a good effect for low-speed unmanned sweeper.

关 键 词：环境感知 滤波跟踪 KALMAN 智能汽车 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]