Decentralized extended Kalman filter of multi-robot position

Abstract

Abstract This paper will discuss the estimation of the multi-robot position. It may be difficult for all sensors to submit their observations due to communication constraints on the multi-robot system. Thus, in this paper, we will estimate the position of each robot with limited communication between robots using decentralized estimation. Since the multi-robot system is non-linear, the extended Kalman filter is used as an estimation method. The combination of decentralized estimation and extended Kalman filter is called a decentralized extended Kalman filter. In the multi-robot system, each robot is represented as a node that makes its prediction. Assumed that some of the nodes will interact with their closest neighbors, so the measurement information will be transferred. Based on simulation results, the decentralized extended Kalman filter algorithm runs well for estimation in a multi-robot position.