Decentralized Conflict Resolution for Navigation in Swarm Robotics

The main challenge in multi-robot navigation is the resolution of navigation conflicts between agents caused by intersecting paths. The contributions of this paper are: The creation of a robot framework that allows us to study the emergent behavior of a swarm of robots in a navigation task that involves frequent conflicts, a baseline strategy for decentralized conflict resolution and a more advanced strategy called Decentralized Collective Conflict Resolution (DCCR). To resolve conflicts, the DCCR strategy uses communication between robots and collective decision-making to create planning constraints, while the baseline strategy creates implicit priorities from communication timing. To show the effectiveness of our methods, we use a swarm of up to ten TurtleBot3 robots applying the DCCR and the baseline strategies in simulated and real-world experiments. In scenarios with more than six robots, the baseline strategy is outperformed by the DCCR strategy, which has more resilience against congestion.