A Vehicle Detection and Tracking Approach Using Probe Vehicle LIDAR Data

Abstract

Detection, identification and tracking of multiple moving targets haveimportant applications in transportation and vehicle control areas. Inthis paper we present our approach to detect, recognize and track thevehicles within the detection region of a moving probe vehicle, based onthe data collected by multiple sensors, including LIDAR and GPS. Thispaper develops a methodology to group the LIDAR measurements intotargets, classify the targets as vehicles or fixed objects, and trackthe vehicular targets within lanes using a Kalman-filter. One importantfeature of this approach is that we track all of the observations inworld coordinates, allowing us to average over many samples and ideallymany runs to differentiate between the fixed objects (road boundaries)and moving objects (vehicles).