Neuro-adaptive distributed formation tracking control of under-actuated unmanned surface vehicles with input quantization

Abstract

This paper addresses distributed formation control of multiple under-actuated unmanned surface vehicles (USVs) subject to input quantization, in addition to the unknown dynamics caused by external sea disturbances and internal model uncertainties. A two-level distributed guidance and neuro-adaptive quantized control architecture is presented to achieve a time-varying formation regardless of the input quantization. Specifically, at the kinematic level, an extended state observer (ESO)-based distributed guidance law is developed to track a time-varying trajectory where the ESO is adopted to estimate the unavailable linear velocity and rate of turn (ROT) of neighboring USVs. At the dynamic level, by using a linear time-varying model to deal with the difficulty caused by quantization and the radial basis function neural networks (RBFNNs) to identify the unknown dynamics, a neuro-adaptive quantized control law is developed where no information on the parameters of quantizers is required. The stability of the proposed two-level formation control architecture is proven on the basis of input-to-state stability, and all signals in the closed-loop system are uniformly ultimately bounded. Simulation results demonstrate the effectiveness of the proposed neuro-adaptive quantized control method for USVs.