A safety guaranteed formation control problem for waterjet unmanned surface vehicles (USV) under the consideration of unmodeled dynamics, environmental disturbances, input saturation, and output constraints is addressed in this paper. The leader-follower method and line-of-sight (LOS) strategy are employed to achieve the desired formation structure. Firstly, two tan-type barrier Lyapunov functions (BLF) are designed to limit the formation errors always within a predefined range to avoid collisions and keep communications, it can also relax the assumption of each USV's initial position to random. Besides, a bioinspired neurodynamics method is involved to avoid the appearance of the virtual velocity differential terms in the control law, which is different from the traditional dynamic surface method and makes it easier to implement in practice. Further, adaptive neural networks (NN) are used to approximate unmodeled dynamics and an auxiliary dynamics system is introduced to handle the input saturation. It is proved that the uniform ultimate boundedness of all signals in the control system can be achieved. Finally, simulation studies show the effectiveness of the proposed control law. INDEX TERMS USV, leader-follower formation, LOS, bioinspired, constraint. I. INTRODUCTION Research on multiple agent system has grown markedly in recent years since the single-vehicle has limited fuel and will increase the time to complete the task. A class of underactuated system waterjet unmanned surface vehicles has received considerable attention due to its rapidity, flexibility and efficiency such as minesweeping, transport strategic materials, rescue missions, seabed mapping, ocean floor survey and so on [1]-[3]. This type of robot with two inputs (surge force and yaw torque) and three outputs (position and orientation) is also called a nonholonomic system, the control strategy for the full-actuated system can not be used directly to the underactuated system. It is worth noting that most works in the formation control for USV [4]-[7] only consider the formation pattern, i.e., the stable-state errors. We can predict that even though the USV moves in the desired pattern, if the extern environments The associate editor coordinating the review of this manuscript and approving it for publication was Valentina E. Balas .