毫米波雷达的扩展目标点云聚合算法  

Extended-target point cloud aggregation algorithm for millimeter-wave radar

作　　者：沈磊 刘康 SHEN Lei;LIU Kang(College of Mechanical and Electrical Engineering,China Jiliang University,Hangzhou 310018,China)

机构地区：[1]中国计量大学机电工程学院,浙江杭州310018

出　　处：《现代电子技术》2023年第3期1-5,共5页Modern Electronics Technique

基　　金：中德无人驾驶中心毫米波雷达联合实验室基金项目(03103-211179);浙江省自然科学基金一般项目(LY19F010007)。

摘　　要：低成本毫米波雷达的天线数目少、天线孔径小、雷达功率低等自身因素限制了雷达的DOA估计。较低的角度分辨率导致难以区分位置相近的扩展目标,从而导致多目标跟踪的误判。针对上述问题,提出一种扩展目标点云聚合算法,首先对多个毫米波雷达所探测到的点云进行融合和叠帧处理;其次通过最小二乘法曲线拟合,构建EKF的加速度模型,在预测值处自适应生成波门,得到假设目标关联位置,将此位置与预测值作加权修正处理后聚合点云。文中设计了基于低成本毫米波雷达的室内多目标跟踪对比实验,通过完备度和位置误差指标加以验证。实验结果表明,该算法有效弥补了低成本毫米波雷达角度分辨差以及扩展目标点云辨识度低的缺陷,大幅提升了扩展目标点云的辨识度及跟踪的精确度,对提升低成本毫米波雷达跟踪性能具有重要意义。Low-cost millimeter-wave radar has a small number of antennas,small antenna aperture and low power,which limit its DOA(direction of arrival)estimation.In addition,its low angular resolution makes it difficult to distinguish the extended targets that are located closely,which leads to misjudgment in multi-target tracking.Therefore,an extended target point cloud aggregation algorithm is proposed.The point clouds detected by several millimeter-wave radars are fused and stacked.The acceleration model of EKF(extended Kalman filter)is established by the leastsquare curve fitting.The wave gate is generated adaptively at the predicted value to obtain the associated position of hypothetical target.This position and the predicted value are weighted and corrected to aggregate the point cloud.A comparison experiment of indoor multi-target tracking based on the low-cost millimeter-wave radar was designed to verify the algorithm by completeness and position error indicators.The experimental results show that the algorithm effectively makes up for the defects of low-cost millimeter-wave radar′s poor angular resolution and low recognition of extended target point cloud,and greatly improves the recognition and tracking accuracy of the extended target point cloud.Therefore,the proposed algorithm is of great significance to improve the tracking performance of low-cost millimeter-wave radar.

关 键 词：点云聚合算法 毫米波雷达 角度分辨率 多目标跟踪 曲线拟合 完备度验证 

分 类 号：TN95-34[电子电信—信号与信息处理]