A hierarchical control framework associated with control algorithms of swarm motion controller are proposed for unmanned surface vehicles (USVs). The controller has the capabilities of being controlled by themselves autonomously and by human operator remotely. To achieve the distributed control of the swarm, the controller is divided into three task levels. The supreme level delivers the human operator's command to the USV swarm remotely, and the middle level deals with the autonomous tasks (i.e. swarm dispersion, obstacles avoidance, and inner-USV collision avoidance), which are defined by specific mathematical functions. The bottom level controls the end effectors (i.e. propellers and rudders) for the USVs are underactuated, and the sliding-mode control method is used to in this level. The Lyapunov method is used to prove the control algorithms theoretically according to the convergence of control error. Numerical simulation results of a group of USVs for a set of autonomous tasks are presented and discussed to verify the effectiveness of the control framework.
In the past decades, formation control of multi-agent has been widely studied by scholars in field of control and robotic, and study object covers satellites, aerocrafts, vessels, and underwater vehicles. During the study processes, three dominating methods for formation control was proposed, i.e. leader-follower, virtual structure, and behavior based method (DO, 2011).
The formation control can preserve desired formation well, but the controller is centralized, and the members are not endowed with the capability of decision-making. Thus, when dealing with some problems, for instance obstacles avoidance and colliding avoidance with fleet's member, the controller might be unstable.
To make multi-agent autonomic, the swarm control has become a focus recently (Bae et al. 2012, Wang 2013, Wai et al. 2011). Be superior to the formation control, the swarm control focuses on attaching robotics with intelligence.