Merging Gaussian Distributions for Object Localization in Multi-Robot Systems

Abstract

We present a method for representing, communicating, and fusing distributed, noisy, and uncertain observations of an object by multiple robots. The approach relies on re-parameterization the two-dimensional Gaussian distribution to correspond more naturally to a robot's observation space. The approach enables two or more observers to achieve greater effective sensor coverage of the environment and improved accuracy in object position estimation. We demonstrate empirically that, using our approach, more observers achieve more accurate object position estimates. The method is tested in three application areas: object location, object tracking, and ball position estimation for robot soccer. We provide quantitative evaluation of the technique on mobile robots.