Adaptive neural-network sliding mode cascade architecture of longitudinal tracking control for unmanned vehicles

Unmanned vehicles have drawn wide attention due to their intrinsic capacity of performing routine tasks for industry, conducting military missions and improving traffic safety. However, since the longitudinal dynamic system of unmanned vehicles inherently has the uncertain nonlinearities and time-varying behavior, longitudinal tracking control is reviewed as a challenging work in the exploitation of unmanned vehicles to deal with the features of uncertain nonlinearities and parametric time varying. In this paper, an adaptive nonlinear cascade control architecture is presented to design the longitudinal speed tracking control for unmanned vehicle, which is a nonholonomic system. Firstly, a upper-level model predictive control law is presented to produce a desired and smooth acceleration in real time, and the saturation characteristic is introduced to limit the accelerations within the range of given values. Then, a lower-level adaptive neuro-network sliding mode control (ANN-SMC) law is presented for dynamically tracking the desired acceleration, in which the uncertain term and the variable structure control term are adaptively adjusted by the neuro-networks, and the stability of proposed ANN-SMC control system is proved by the Lyapunov theory. Finally, simulation and experimental results demonstrate the feasibility and effectiveness of proposed control approach.