Laser Tracking and Classification of Multiple Objects

Abstract

Tracking people is a popular problem in machine vision. In this paper, we describe a method for real-time tracking of moving people with a laser range finder - a sensor rapidly gaining popularity in robotics. We show that with planar laser data we can track and distinguish between animate and inanimate objects, by using heuristic quantities. We also show how our method can be extended to estimate the path of objects behind occlusions. We present initial results on object classi cation based on their movement signatures and a Kalman Filter formulation tailored for laser tracking.