Cooperative, distributed localization in multi-robot systems: a minimum-entropy approach

In this paper, we consider the problem of localization in a multi-robot system. We present a new approach focused on distribution, scalability, and minimum-uncertainty perception. An Extended Kalman Filter (EKF) is used to update an estimate of the robot poses in correspondence to each sensor measurement. An entropic criterion is used, in order to select optimal measurements that reduce the global uncertainty relative to the estimate of the robot poses. It is shown that, in addition to EKF, also the selection of the optimal measurement can be distributed among the robots, in a scalable fashion. The proposed approach has been validated by simulations and preliminary experimental results.