Multi-robot Localization for a Formation System Based on Range Information Only

Localization of the multi-robot formation system without infrastructure is vital for navigation and control when priori knowledge of environment is unavailable. This paper proposes an approach to estimate robots' poses and landmark positions of a multi-robot formation system. Through combining odometry and observation information, the multi-robot localization problem is formulated as a least squares problem, which is solved through Gauss-Newton iteration algorithm. Particularly, two particular scenarios are discussed, which shows that the ambiguity level of the landmark can be avoided comparing with existing range-only SLAM methods. The effectiveness of the method proposed is verified via simulation results.