A Study on the Improvement of USV’s Leader-Follower Swarm Control Algorithm through Fault Coping Algorithm

Abstract

In this study, we conducted a swarm control algorithm to overcome the limitations of a single-object system. Among various swarm control methods, this study selected the leader-follower swarm control method, which is easy to mathematically interpret the algorithm and theoretically can increase the number of followers infinitely. In addition, in order to overcome the phenomenon that the swarm formation collapses when the leader has a breakdown, which is mentioned as a disadvantage of the leader-follower swarm control algorithm, a fault coping algorithm was developed to detect and respond to failures so that the mission can continue. In this study, an actual marine robot(Unmanned Surface Vehicle) was developed to verify the proposed algorithm. USV is largely divided into a power system and a communication system. Among them, a coulometer is installed in the power system to detect failures of the USV and protect other equipment in the event of a failure. When this occurs, the switch connected to the coulometer cuts off the circuit to which the battery is connected, cutting off the power to other equipment such as the sensor or the main controller. USV’s communication system adopts a decentralized method rather than a centralized method to reduce the communication load of the entire system, shorten data sharing time, and is designed to be operable even in environments where infrastructure construction is difficult, such as a server room. The developed USV is loaded with a basic navigation algorithm such as a waypoint algorithm, etc. for missions such as exploration, and a leader-follower swarm control algorithm for swarm control. In addition, data such as voltage, current, thruster RPM, current position, current heading angle, etc. of USV system are acquired and analyzed through sensors mounted on each USV to monitoring the USV failures, if a failure situation(for example, a failure of a specific USV’s thruster) occurs during the mission, the USV in the nearest location takes over the mission of the USV in which the failure occurred according to the fault coping algorithm proposed in this study. The performance of the leader-follower swarm control algorithm and fault coping algorithm was confirmed by conducting the actual sea area test. Prior to the actual sea area test, the sensor performance test and communication speed test were conducted on land to verify the accuracy and communication speed of the sensor installed on the USV. Lastly, it was confirmed that the USV swarm formation was well formed, and when a virtual failure situation was applied to the leader USV, it was confirmed that the fault coping algorithm proposed in this study worked well and normal mission performance was possible.