Pattern Recognition and Tracking Dynamic Objects with LIDAR

This paper presents a method to detect moving persons based on the information from a LIDAR type sensor mounted on a mobile robot. The detection of the objects is performed relative to the estimated robot position. The objects of interests are first described with Gaussian Mixture Model (GMM) and later on searched and classified with a Bayesian classifier. The estimated person positions are tracked via the Extended Kalman filter with two kind of motion models for humans.