This paper studies the problem of vehicle detection and tracking for an autonomous vehicle using 3D Light Detection and Ranging (LiDAR). In order to increase the accuracy of vehicle detection and tracking, a new clustering algorithm is proposed to obtain vehicle candidates from preprocessed point cloud data collected by the LiDAR. A classifier trained by the support vector machine (SVM) algorithm is used to detect vehicles from vehicle candidates. Kalman filter and global nearest neighbor (GNN) algorithm are used to track vehicles, and the accuracy of vehicle detection result is further improved by the aid of tracking results. The proposed method has been verified on a testing platform.

中文翻译：

---

使用 3D LiDAR 进行实时车辆检测和跟踪

本文研究了使用 3D 光检测和测距 (LiDAR) 对自动驾驶车辆进行车辆检测和跟踪的问题。为了提高车辆检测和跟踪的准确性，提出了一种新的聚类算法，从激光雷达采集的预处理点云数据中获取候选车辆。由支持向量机 (SVM) 算法训练的分类器用于从候选车辆中检测车辆。采用卡尔曼滤波和全局最近邻（GNN）算法对车辆进行跟踪，借助跟踪结果进一步提高了车辆检测结果的准确性。所提出的方法已在测试平台上得到验证。