The paper addresses the problem of multi-robot navigation in human shared spaces. We propose a hierarchical framework that combines global path planning, local path planning and reactive strategies, ensuring a safe and socially-aware navigation. We show through several tests and extensive experiments with a real robotic implementation that our combination of solutions delivers excellent results in terms of robustness and performance even in challenging natural scenarios.

Given a team composed of groups of robots with different abilities and different tasks, we propose a control solution that allows maintaining the connectivity between the agents whilst securing the execution of the different tasks assigned to the team. The specific notion of connectivity adopted here is that the different agents have to mutually remain in line-of-sight, and this condition is guaranteed in presence of obstacles. In the computation of the graph topology, based on Minimum Spanning Tree (MST), each agent takes into account the surrounding environment, its own task and the tasks of its neighbours. Our solution permits the execution of heterogeneous tasks and the task dynamic reallocation between the robots when required by the situation. The approach relies on a sound theoretical framework, which is discussed in this letter, while its practical feasibility is shown through simulations and experimental data using real robotic platforms.