Detecting and tracking of host people on a slave mobile robot for service-related tasks

Abstract

This paper aims to solve detecting and tracking of pedestrian in indoor service areas such as library, museum, or department store, using a laser range finder on a mobile robot for service tasks. It is much more difficult than implementing detection and tracking using fixed sensor because the mobile robot must localize itself correctly when distinguishing static objects and moving objects. Therefore, first, in this paper, the robot odometry is corrected by a scan-matching algorithm with outlier filtering. Second, moving objects are detected by comparing the present scan with a local occupancy grid map. Third, for each moving object, an Extended Kalman Filter is maintained to track it. The proposed algorithm has been extensively tested for the detection and tracking of host people in free space for application such as people following.