Global Planning for Multi-Robot Communication Networks in Complex Environments

In this paper, we consider networks of mobile robots responsible for servicing a collection of tasks in complex environments, while ensuring end-to-end connectivity with a fixed infrastructure of access points. Tasks are associated with specific locations in the environment, are announced sequentially, and are not assigned a priori to any robots. Information generated at the tasks is propagated to the access points via a multihop communication network. We propose a distributed, hybrid control scheme that dynamically grows tree networks, rooted at the access points, with branches that connect robots that service individual tasks to the main network structure. To achieve this goal, the robots switch between different roles related to their functionality in the network. The switching process is tightly integrated with distributed optimization of the communication variables and motion planning in complex environments, giving rise to the proposed distributed hybrid system. Our proposed scheme results in an efficient use of the available robots and also allows for global planning by construction, a task that is particularly challenging in complex environments.