Consistent multirobot localization using heuristically tuned extended Kalman filter

Probabilistic algorithms have widely been used with significant success in single-robot localization as well as mapping. However, when it comes to distributed, multirobot systems, probabilistic algorithms have a tendency to quickly converge to inconsistent, often overly optimistic estimates, whenever interdependencies in such systems are ignored. This paper presents a solution to consistent, decentralized, multirobot localization using a heuristically tuned Extended Kalman Filter. Extensive simulations show that the proposed solution is able to significantly improve the consistency of pose estimates for each robot in a system while maintaining the computational complexity of the classical Extended Kalman Filter.