Integrated sensing, communication, and control for multi-agent networked formation control

This paper studies the multi-agent networked formation control in dynamic industrial environment, in which multiple automated guided vehicle (AGV) agents are coordinated by a base station (BS) to collectively perform cooperative transportation tasks. In this system, the limited local sensing capability at each agent and their frequent interactions may cause large synchronization errors and high closed-loop latency, degrading the networked formation control performance. To address these challenges, we propose a new communication-sensing enhanced multi-agent formation control strategy based on the idea of integrated sensing, communication, and control (ISCC). First, we establish an ISCC system design to accurately capture the interdependencies among sensing, communication, and control in networked formation control. Then, we design a dynamic obstacle avoidance risk map using the conditional value at risk, which quantifies the collision risks under communication latency, thus helping to reserve sufficient time for smooth obstacle avoidance and reduce material extrusion risks during emergency braking. Next, we formulate the multi-agent formation control problem as a partially observable Markov decision process, which is solved via the multi-agent proximal policy optimization (MAPPO) by exploiting the global ISCC states. Furthermore, to decrease the extra overhead for ISCC states interaction, we design a dynamic communication cycle allocation mechanism via global states, which effectively balances the synchronization precision and communication overhead. In addition, we employ a heterogeneous framework to mitigate gradient conflicts and boost control efficiency for heterogeneous agents. Simulation results reveal that our strategy improves the synchronous performance via reducing the error by at least 59.9% compared to baselines, significantly reducing closed-loop latency and traveling time.