Vision based navigation for golf ball collecting mobile robot

This paper describes a vision-based navigation method for an autonomous mobile robot which can be used to collect golf balls in the golf driving range. The relative position and directions between the golf ball and the mobile robot can be determined using wide-view images from the camera mounted on top of golf driving range. The camera information is sent to the main computer which makes map and the shortest path to the destination and then motion commands can be sent to the mobile robot wirelessly. Navigation system uses stereo camera images and Monte Carlo localization (MCL) to determine the location and orientation of the platform, from which it can plan a path to collect golf balls effectively. The prototype robot is tested in a real driving range and results from experiments are presented.