Decentralized task coordination and active sensing for multi-agent systems under team-wise intermittent communication

Multi-agent systems have demonstrated significant potential in enhancing task efficiency by acting collaboratively and concurrently. However, when inter-agent communication is limited to short-range or intermittent links, achieving effective coordination becomes significantly challenging. The difficulty is further amplified in partially unknown environments, where agents must actively sense the environment and share observations to improve situational awareness. This work presents an online decentralized framework that integrates temporal task coordination, active information gathering, and team-wise intermittent communication for multi-agent systems. The proposed method jointly optimizes the motion plan of each agent to satisfy local temporal logic tasks, selects informative sensing locations to reduce environmental uncertainty, and schedules team-wise communication to ensure timely information exchange under connectivity constraints. Extensive simulations in large-scale scenarios demonstrate the scalability and robustness of the framework in achieving reliable task completion, efficient uncertainty reduction, and resilient team communication.