Resilient Decentralized Cooperative Localization for Multisource Multirobot System

Abstract

Although cooperative localization (CL) is fundamental to multirobot systems, current algorithms suffer from the tracking of interdependencies, information fusion from multiple sources, and restriction to specific measurement models. To improve the accuracy of localization algorithms for multirobot systems and reduce the impact of uncertainty in multisource measurement information, this article proposes a resilient decentralized CL (RDCL) algorithm. We modify the measurement update procedure of the traditional decentralized CL (DCL) algorithm to track inter-robot correlations and ensure the independence of the measurement update procedure of the elemental filters. We use optimal information fusion algorithms to fuse multisource information, and determine the overall estimate of every robot through a weighted sum of multisource estimates, thereby achieving accurate localization. To enhance the robustness of the multirobot localization system, an online validation module is added to validate the multisource estimates. The proposed CL framework is decentralized and not restricted to specific models. Simulations results show that the proposed algorithm improves localization accuracy and resilience of the multirobot system compared to existing CL algorithms. Experimental results using real-world dataset demonstrate that our proposed algorithm can achieve a localization accuracy with an average root mean square error (ARMSE) of 0.68 m, and it is 34% better than that of the traditional DCL algorithm.