Two-layer virtual leader-following: An adaptive cooperative path following control for crowded UUV swarm subjected to constraints

This paper investigates an adaptive cooperative path following of a crowded unmanned underwater vehicle (UUV) swarm in the presence of constraints, including uncertain dynamics, external disturbance and input saturation. To satisfy a realistic crowded UUV swarm, a two-layer virtual leader-following (TLVL) is developed to keep the swarm along a commonly desired path while avoiding collision and preserving connectivity, where the desired path is the first-layer virtual leader and the center of mass (COM) for entire swarm is the second-layer virtual leader, and all UUV are followers whose interactions are only affected by partial information from their neighbors. For keeping UUVs connected in a group, the COM is estimated by using a consensus algorithm. Unlike the conventional line-of-sight (LOS) without considering unknown and time-varying sideslip angle, a COM-LOS guidance with finite-time sideslip-angle observer is designed to improve the convergence of cross-track error between the first-layer and second-layer virtual leaders, while three bounded potential fields are incorporated to guarantee without any collisions and loss communicating-links. To deal with uncertain dynamics and external disturbance, a robust adaptive tracking controller is proposed incorporating the disturbance observer, neural network, auxiliary scheme and command filter into TLVL pattern designed to track an integrated desired surge velocity and heading, where an auxiliary scheme is introduced into a switching manner to compensate effects of input saturation. The results about system stability are given by Lyapunov analysis. It has been demonstrated that the proposed scheme can drive the UUV swarm to follow a desired path and all error signals in the closed-loop system can converge to a small neighborhood around origin. The numerical simulation studies are finally given to illustrate the effectiveness of the proposed approach.