Adaptive Risk-Aware Multi-Target Tracking and Monitoring With Network Reconfiguration

We consider a scenario in which a group of robots tracks a group of targets in an open space. In particular, the robots are heterogeneous, and the targets are adversarial, capable of attacking the robots and severing the links between the robots and their corresponding sensors. Additionally, each robot is required to estimate the state of the targets individually on the basis of the communication graph. We propose a framework that adaptively balances accuracy and safety while automatically repairing the communication graph. Our framework follows a two-stage strategy: In the first stage, we assess the entire team to determine if repair is necessary. If necessary, by quantifying the team’s observability using the trace of the Grammian matrix, we propose a computationally efficient repair strategy. In the second stage, safety and accuracy are quantified, with the sensing margin serving as the dynamic weight to guide robot coordination. To validate the effectiveness of our work, we simulated a monitoring and tracking task and compared our network reconfiguration strategy with greedy and One-Hop-Grammian-based methods. The simulation results demonstrate the effectiveness and efficiency of our approach.