A consensus algorithm for common reference frame estimation in networked multi-agent systems

This paper is concerned with the design of a novel discrete-time consensus algorithm to estimate a common reference frame in a multi-agent system by performing decentralized agreement on a common reference point on a plane without relative pose measurements. The problem of common reference frame estimation on a 2-D plane can be divided into two subproblems: the estimation of a common origin and the estimation of a common direction. This preliminary paper focuses of the first subproblem and points to other approaches presented in the literature to solve the second subproblem. The proposed algorithm is based only on local interactions among the agents which exploit only measurements of relative distances and angles with respect to each other but are unable to estimate their relative attitude during the estimation process. Our approach extends the state of the art by achieving bounded estimation errors in the case of moving agents that simultaneously change their relative and unknown attitude with respect to each other.