Localization of a mobile robot with beacons taking erroneous data into account

The authors analyze the operation and performance of a navigation system for a mobile inspection robot, using an ultrasonic/infrared telemeter with active beacons. A realistic statistical model has been established from a theoretical study of measurement errors and has been confirmed by a preliminary experiment. The optimization of the mobile position using all the beacon data and a reweighted least-squares criterion is considered. A method for tracking the robot trajectory is proposed. For each set of data a coherence test is conducted, with the possibility of detecting and rejecting occasional erroneous measurements. Numerical simulation results show that this unsophisticated principle of localization is well suited to the intended application: an accuracy of about 0.1 m is obtained for robot position with very noisy measurements