COOPERATIVE PEOPLE TRACKING WITH MULTIPLE GROUND LASER SCANNERS

Abstract

This paper presents laser-based people tracking using two networked 64-layer laser scanners (LSs) fixed in an environment. The tilt angles, heights, and relative poses of the two LSs are calibrated based on the laser-scanned data. After the position data of people are extracted from the laser-scanned data using a background subtraction method, poses and sudden behaviours of people, such as dashing and falling down, are estimated based on the interacting-multimodel (IMM) estimator. Experimental results validate the effectiveness of the proposed method.