IMPROVING OBJECT LOCALIZATION THROUGH SENSOR FUSION APPLIED TO SOCCER ROBOTS

Abstract

This paper introduces a method for representing, communicating and fusing distributed, noisy and uncertain observations of objects by multiple robots. The approach relies on re-parameterization of two-dimensional Gaussian distri- butions that are used to represent the positions of all players and the ball. The approach enables two or more observers to achieve greater eective sensor coverage of the environment and improved accuracy in object position estimation. We demonstrate empirically that, using this approach, more observers achieve more accurate object position estimates. Two dierent procedures for merging Gaussian distributions were implemented and tested in the RoboCup Soccer Server, a simulated environment for robotic soccer. We also present the first results obtained with middle-size league robots.