Distributed Estimation for Coordinated Target Tracking in a Cluttered Environment

The purpose of this paper is to gain insight into several fundamental questions that arise whenever a distri buted control system is endowed with communication capabilities . These fundamental questions include: what should each vehi cle say?, how frequently should they talk?, and how much does communication really help? These questions are evaluated here with respect to a target tracking task in which multiple pursuit vehicles estimate the state of a target vehicle. Thi s task has three main components: communication, estimation, and control. On each communication event, one pursuit vehicle h as the opportunity to broadcast information to all other pursu ers. This information always includes its own state (assumed to b e known), but can additionally include the most recent targetstate observations or the current target state estimate. Estimat ion is achieved with an Extended Kalman Filter on each pursuer and the controller is behavior-based. Simulation results showthat communication always improves distributed estimate resul ts. Which information to transmit depends on available bandwidth, and more frequent communication yields better estimates. Results also show how coordinated control can be beneficial to target tracking in a cluttered environment.