Self-Localization for Distributed Systems based on the Relative Measurement Information

This paper studies the cooperative relative localization of multiple agents for a distributed system in two dimensions. The system is composed of anchor agents which can obtain absolute location information in the global coordinate system and unknown agents that can obtain location information relative to neighbor agents in the onbody coordinate system. The relative measurement relation and communication topological relation between agents is represented by a time-varying directed graph. We firstly give the classification criteria of which agents in the system can be measured. The classification criterion we proposed has less time and space complexity. Secondly, we propose a Relative-based Distributed Kalman Filter, which can be applied to relative measurement agents and effectively address the need for confidentiality between agents and the scalability of the system through limited information transmission. Thirdly, we apply the algorithm to the agents classified in the first step. Some simulation results demonstrate the feasibility of the algorithm.