Trajectory Planning and Decision-Making for Multi-Robot Systems with Robust and Resilient Connectivity Maintenance and Human-Robot Collaboration

This work addresses the problems of motion planning and decision-making for multi-robot systems with emphasis on the maintenance of communication network properties such as resiliency and connectivity. Planners based on mixed-integer programming and model predictive control (MPC) are proposed, enabling the group of robots to perform tasks while maintaining network connectivity considering potential bounded disturbances and robot failures. The requirement of both omnidirectional and directional line of sight between agents for communication links to be formed is studied. A novel algorithm for human-robot collaborative teams, leveraging a combination of deep inverse reinforcement learning and MPC, is also proposed.