Cooperative Multi-Robot Object Tracking in Unknown Environment Using Covariance Intersection

This work presents a approach for multiple cooperating mobile robots for moving object tracking in unknown environment. Each robot in the team uses the full covariance extend Kalman filter based algorithm to simultaneously localize the robot and target while building a landmark feature map of the surrounding environment. Meanwhile, in local robot system the covariance intersection based data fusion method is used to fuse information sent by the other robot teammates, those information may contains the location of target and the location of robot itself from other teammate’s point of view. The method is distributed, and let the multi-robot system have the ability of robustness. The results of simulation validate a higher accuracy of our method compared with non-fusion single robot solution.