Distributed consensus-based observer design for target state estimation with bearing measurements

Abstract

This paper introduces a novel distributed consensus-based observer design that enables a group of agents in an undirected communication network to solve the problem of target tracking, where the target is modelled as a chain of integrators of arbitrary order. Each agent is assumed to know its own position and simultaneously measure bearing vectors relative to the target. We start by introducing a general continuous time observer design tailored to systems whose state dynamics are modelled as chains of integrators and whose measurement model follows a particular nonlinear but observer-suited form. This design leverages a correction term that combines innovation and consensus components, allowing each agent to broadcast only a part of the state estimate to its neighbours, which effectively reduces the data flowing across the network. To provide uniform global exponential stability guarantees, a novel result for a class of nonlinear closed-loop systems in a generalized observer form is introduced and subsequently used as the main tool to derive stability conditions on the observer gains. Then, by exploring the properties of orthogonal projection matrices, the proposed design is used to solve the distributed target tracking problem and provide explicit stability conditions that depend on the target-agents geometric formation. Practical examples are derived for a target modelled as first-, second-, and third-order integrator dynamics, highlighting the design procedure and the stability conditions imposed. Finally, numerical results showcase the properties of the proposed algorithm.