研究業績

This paper presents a distributed navigation algorithm for robotic swarm systems based on model predictive control (MPC). The proposed algorithm addresses sudden robot failures, in which a robot suddenly comes to a halt due to hardware malfunctions or similar causes. Such failures may disrupt swarm connectivity and hinder mission execution, requiring navigation strategies that enable continued swarm operation. The proposed approach considers a leader–follower scenario in which a leader guides follower robots. Each follower determines its motion using only local information from robots within its sensing range, while satisfying practical constraints such as maximum sensing distance and speed. Within a distributed MPC framework, each robot autonomously performs collision avoidance, connectivity maintenance, connectivity recovery after failures, and connectivity strengthening. Simulation and experimental results demonstrate that the proposed method enables the swarm to restore connectivity after failure and continue navigation while avoiding collisions, even when a failed robot remains in the environment as an obstacle.