A Relative Localization System for Mobile Robot Formations

This diploma thesis project describes a localization system to measure the relative poses of mobile robots driving in a formation. An overview of existing localization systems is given and several possible measurement methods are discussed. An attempt has been made to characterize the suitability of the Global Positioning System (GPS) for the coordination of a group of mobile robots. A single frequency GPS receiver was found not to be precise enough to avoid collisions of the vehicles. It is also not available indoors, so an additional system based on ultrasonic sensors has been designed and implemented. To find the most suitable localization algorithm, the achievable accuracies were compared in worst case simulations. The major advantage compared to existing systems is the reuse of ultrasonic obstacle detection sensors that are already available on most mobile robots, and that no precise time synchronisation between the robots is required. The sensors can be placed at arbitrary locations on the robot. This minimizes the effort of integrating the system into existing multi-vehicle platforms. The computational complexity is kept low so that even a low cost micro controller is fast enough to run the derived algorithm for real time measurements. The selected hardware design is robust and can survive harsh conditions in outdoor environments. The proposed localization system was installed on the OutdoorMERLIN rovers. With these vehicles several experiments were made to test the reliability and accuracy of the system in real world conditions.