Mobile robot trajectory learning using absolute and relative localization data

Abstract

This paper describes the trajectory learning system for an autonomous mobile robot designed for structured indoor environments. Absolute localization data are used by a trajectory learning module to create trajectory files which will be used during the task execution. The authors present the robot localization system which makes use of motion estimation provided by odometry and inertial sensing and of angular position measurements of artificial beacons. The learning system has been implemented on an experimental robot and tested in an office environment.