多车协同目标跟踪方法

Methods for Multi-Vehicle Cooperative Object Tracking

地面无人系统中多车信息融合技术是提升系统环境感知能力的重要途径。针对单车传感器存在视野遮挡及盲区导致的目标跟踪不连续不稳定问题,提出一种集中式多车协同感知的结果级融合系统模型。该系统模型采用激光雷达作为车辆感知传感器,对不同车辆构建的环境栅格地图在主控端采用D-S证据理论进行融合、得到全局静态环境地图,完成多车协同感知环境模型的构建。在此环境模型基础上设计一种多车协同目标检测与跟踪方法,采用极大值抑制的方法解决检测目标融合冲突;设计一种级联动态目标匹配与跟踪管理方法,完成目标预测跟踪并将结果下发给各车。由两辆无人车组成的实车系统测试结果表明:当出现目标遮挡时,所提多车协同目标检测与跟踪架构相对于单车感知在环境表征上能够获得更全面的目标信息,跟踪目标未出现漏检,未发生跳变;跟踪器输出位置状态结果与检测结果误差较小,能够对所跟踪目标的状态进行准确估计,跟踪轨迹保持连续,有效提高了单车环境感知视野。

Multi-vehicle information fusion technology is an important way to improve the perception of the environment of ground unmanned systems.To address the problem of discontinuous and unstable object tracking in single-vehicle sensors caused by vision occlusion and blind spots,a result-level fusion system model for centralized multi-vehicle cooperative perception is proposed.The system model uses lidar as the vehicle perception sensor and stands on the D-S evidence theory to fuse the environment grid maps constructed by different vehicles at the main control terminal to obtain a global static environment map.Based on this environment model,a multi-vehicle cooperative object detection and tracking method is designed.First,a maximum value suppression method is used to resolve the fusion conflict of detected objects.Then,a cascaded dynamic object matching and tracking management method is designed to complete object prediction and tracking and send the results to vehicles.The test results of a real-vehicle system composed of two unmanned vehicles suggest that when the object is occluded,the proposed multi-vehicle cooperative object detection and tracking architecture can obtain more comprehensive environmental information of the object than a single-vehicle perception system.No tracking object is missed,and no jump occurs.The error between the tracker s output position state result and the detection result is small.The state of the tracked object can be accurately estimated,and the tracking trajectory remains continuous,thus effectively improving the field of vision of the single-vehicle environment.