| dc.description.abstract | This thesis investigates dense Light Detection and Ranging (LiDAR) sensors as a method for object detection and tracking in stationary infrastructure-like applications. A literature review of existing works is conducted, with discussion and comparisons for other sensing technologies. Additional discussions are made for geometric feature-based methods and end-to-end learning methods for object detection from pointcloud data. Subsequently, theoretical pointcloud spacing models for multi-beam 360 deg LiDAR sensors are developed, with analysis on placement strategies and LiDAR configurations. The thesis continues with an implementation of a geometric feature based object detection method, primarily for vehicles. Several algorithm designs are presented for pointcloud background removal, clustering, orientation detection, tracking, and filtering. Detection and tracking metrics are then established to observe the system's performance on both experimental and simulation datasets. Two datasets collected with a Velodnye VLP-16 sensor on both a highway and urban road segment are utilized for experimentation, while scenarios of light traffic and stop-and-go traffic on a highway are developed in the CARLA simulator to further validate tracking performance. | en |
