Frontier-Based Exploration and Rendezvous for Multi-Robot Teams in Communication-Restricted Environments: an Evaluation

Multi-robot rendezvous and exploration are fundamental challenges in the domain of mobile robotic systems. This paper addresses multi-robot rendezvous within an initially unknown environment where communication is only possible after the rendezvous. Traditionally, exploration has been focused on rapidly mapping the environment, often leading to suboptimal rendezvous performance in later stages. We adapt a standard frontier-based exploration technique to integrate exploration and rendezvous into a unified strategy, with a mechanism that allows robots to re-visit previously explored regions thus enhancing rendezvous opportunities. We validate our approach in 3D realistic simulations using ROS, showcasing its effectiveness in achieving faster rendezvous times compared to exploration strategies.

We consider the rendezvous problem for a team of autonomous mobile robots in the challenging setting of a communication-restricted and initially unknown indoor environment. We assume that the robots start from different arbitrary locations, but no map of the environment is available to any of them and no pre-determined meeting location or coordination strategy has been agreed upon. The communication in the environment is restricted to occur only after a rendezvous.

The difficulty of the rendezvous problem is augmented by the fact that the MRS needs to simultaneously perform an online exploration task, as the environment is unknown. The widely used frontier-based exploration strategy is in contrast with what is required by an efficient rendezvous, which, in principle, does not strictly require building a complete map.

We propose an extension of the classical Frontier-Based Exploration (FBE) biased towards Rendezvous (hence the acronym FBR). Frontiers are created not only for their potential contribution to the map's expansion but also to increase the opportunity of meeting a teammate. To do so, we introduce an exploration trace with an information decay mechanism that allows each robot to forget about parts of the environment that have been explored and mapped, creating hallucinated frontiers. In this way, the robot is encouraged to backtrack on previously explored areas facilitating accidental rendezvous. When robot meet, they from a cluster merging their exploration trace and sharing the frontiers. When a new cluster is formed, the leader continue the exploration strategy with hallucinated frontiers, while the other members are followers. The rendezvous is completed when all robots are inside the same cluster.

We validate our approach with an extensive experimental campaign in realistic simulations, considering sensors noise and robot failures. We perform more than 180 real-time runs with a Turtlebot-3 burger exploring large environments simulated with Gazebo. The results demonstrate that our method encourages the robots to backtrack on previously-visited location, facilitating accidental rendezvous. Robots using FBR perform a rendezvous with a speed-up of approx. 50% with respect to a classic frontier-based approach (FBE).