A Switching Strategy for Target Tracking by Mobile Sensing Agents

Abstract

We introduce a co-design approach for control and sensing algorithms for autonomous mobile sensing agents to track a moving target. The mobile sensing agents move in a group while taking bearings-only measurements. We construct an extended Kalman filter that provides estimates of the state of the target and investigate the optimal formation of the agent group by maximizing the determinant of Fisher information matrix (FIM). We develop a strategy that allows the agents to form groups with different number of agents. Agents autonomously switch to larger groups only when the convergence rate of smaller groups to the target is not satisfactory. This switching strategy strikes a balance between the complexity of cooperation and performance of sensing algorithms, which may enable more flexibility in autonomy compared to fixed strategies.