부산대학교 도서관

Three-dimensional object detection technology in autonomous driving systems is an important component of the environment perception module, which relies on sensors such as LiDAR to measure the distance of the surrounding environment and generate three-dimensional point cloud information. For autonomous driving tasks, three-dimensional object detection and tracking are critical tasks that require the use of the object’s historical trajectory as input. Currently, scholars at home and abroad are conducting research on three-dimensional point cloud object detection. This paper proposes a method for processing the geometric features of point clouds based on the PointNet model. After extracting the geometric features of point cloud information, the feature vector is used as input, and PointNet is used to process the feature information for three-dimensional point cloud object detection, which has high accuracy and speed.