Tracking People with a 360-Degree Lidar

Abstract

Advances in lidar technology, in particular 360-degree lidar sensors, create new opportunities to augment and improve traditional surveillance systems. This paper describes an initial challenge to use a single stationary 360-degree lidar sensor to detect and track people moving throughout a scene in real-time. The depicted approach focuses on overcoming three primary challenges inherent in any lidar tracker: classification and matching errors between multiple human targets, segmentation errors between humans and fixed objects in the scene, and segmentation errors between targets that are very close together.