Accurate and automatic surveying of beacon positions for a laser guided vehicle

In this paper we outline a system for automatic surveying of beacon positions for an autonomous guided vehicle. The vehicle measures the bearings to retroreeective beacons using a laser scanner. As the vehicle moves, asynchronous measurements to unidentiied beacons are obtained. The goal of the system is to automatically group those measurements that belong to the same beacons, to eliminate all erroneous measurements and to estimate the two-dimensional map of the beacon positions. Grouping of the beacons is achieved in a RANSAC type of algorithm. The map of the beacons is estimated using maximum likelihood estimation using bearing measurements and odometry. Recursive approximations of the estimation problem is used for increased speed.