Human Following by an Omnidirectional Mobile Robot Using Maps Built from Laser Range-Finder Measurement

Abstract

This paper deals with map building from laser range finder measurement in an unknown indoor environment and its application to human following by an omnidirectional mobile robot. After reviewing basic strategies of human following by a mobile robot involving simultaneous acquisition of indoor map and robot location acquisition, we implemented “pseudo” odometry, rather than conventional odometry, for the omnidirectional mobile robot, using this information to improve scan-matching calculation accuracy. We then conducted experiments in which the robot followed a pedestrian. We confirmed that the robot could follow different pedestrian trajectories if walking was slow, and that our approach effectively improved scan matching calculation accuracy.