Tracking People with a 360-Degree Lidar

Abstract

Advances in lidar technology, in particular 360-degreelidar sensors, create new opportunities to augment andimprove traditional surveillance systems. This paperdescribes an initial challenge to use a single stationary360-degree lidar sensor to detect and track people movingthroughout a scene in real-time. The depicted approachfocuses on overcoming three primary challenges inherentin any lidar tracker: classification and matching errorsbetween multiple human targets, segmentation errorsbetween humans and fixed objects in the scene, andsegmentation errors between targets that are very closetogether.