Decentralized Localization for Multi-agent Systems Based on Asynchronous Communication

This paper provides a fully distributed algorithm for localization problem of multi-agent systems based on the extended and iterated Kalman filter. The main challenge in distributed collaborative localization is how to accurately handle the information coupling problem between different agents. If the cross-covariance estimation of each agent is inaccurate or even directly ignored, it will lead to inconsistent in the final estimation results. It is even more difficult when sustained synchronous communication between the robots cannot be guaranteed. In this paper, an iterated Kalman filter method combined with relative measurement is studied. This algorithm is not only suitable for asynchronous communication and distributed scenarios, but also can greatly improve the accuracy of localization as compared to the method by only using private measurement. The solution consists of two parts. Firstly, each agents update their location information by motion and private measurements using extended Kalman filter estimator. Secondly, the agent uses the relative measurement information from neighbors to estimate the inter-robot correlations to further update its localization based on iterated Kalman filter estimator. Simulation and experimental results prove the effectiveness of our proposed algorithm.