다수 무인선의 대형 유지와 충돌 회피를 위한 군집제어 알고리즘에 관한 연구

초록

In swarm control research, the two most important issues are collision avoidance among moving objects and vehicle-to-vehicle communication (V2V communication). Leader-Follower method, the commonly used in swarm control, has problems about these two function. Data loss in communication or a failure of the leader can cause the entire formation to become uncontrollable. To resolve this problem, this paper suggests the variable Leader-Follower method and collision avoidance process based on the Velocity Obstacle method. Instead of fixing a leader, it selects the leader using an allocation algorithm based on the navigation situation. Also, each ship except the leader, in the swarm choose new ideal point using formation matrix that contains the distance from each ship to ideal points. To avoid static and dynamic obstacles, each ships in the swarm make local path using Velocity Obstacle method(VO) with Quadrangle Ship Domain (QSD). If there is no collision risk, they follow the global path; Otherwise, they plans the local path that can avoid collisions with other vessels and obstacles. To validate the algorithm proposed in this paper, navigation simulations are conducted in the simulation environment based on a MMG type manoeuvring model. It is confirmed that the follower can maintain their formation by formation control and appropriate leader was selected based on the current situation by path following control. The collision avoidance simulation with static, dynamic and complex obstacles ensured that the vessels in the swarm can avoid the collision and they can find the optimum formation.

키워드