マルチロボットシステムにおける分散メッシュネットワーク形成のための粘菌にヒントを得たアルゴリズム

This work proposes a bio-inspired approach for decentralized coordination in multi-robot systems, applied to a simulated port scenario. The methodology integrates the Robot Operating System (ROS) with the Stage simulator, enabling modeling of a port environment with three autonomous robots, each capable of navigation and obstacle avoidance. The main contribution is a connectivity module inspired by Physarum polycephalum, which manages the mesh network and allows decentralized task sharing whenever connections exist. The algorithm adapts continuously to robot movement and environmental changes, ensuring efficient communication when possible and autonomous operation when disconnected. Experiments confirmed that robots relying only on local perception can form and maintain a functional network. Results showed connections established in less than two seconds on average and reconfigured almost instantly after fault, demonstrating resilience. New robots were integrated in only 0.092 seconds on average, validating scalability. A comparison between scenarios with and without communication revealed a 17.87% efficiency improvement, reducing execution time from 621 to 510 seconds thanks to dynamic load balancing. In summary, the study demonstrates the feasibility of a bio-inspired solution for decentralized coordination in multi-robot systems, capable of generating efficient, resilient, and adaptable communication networks, essential for cooperation in real-world environments. A demonstration video is available at https://youtu.be/ZGPswbfeRKA.