To solve the diving control problem of an autonomous underwater vehicle(AUV), the backstepping method based on feedback gain is proposed for the controller design. Firstly, part of coupled nonlinear terms can be eliminated through choosing the controller’s parameter with guaranteed stability. Model uncertainties including nonlinear hydrodynamic damping terms and external disturbance in pitch motion are compensated by neural network(NN) controller online, while the approximation error of the neural network can be eliminated by the adaptive robust compensator to increase the convergence learning rate of neural network. Both neural network strength and adaptive laws are obtained through the Lyapunov stability theorem, and uniformly ultimately bounded(UUB) can be guaranteed for the closed-loop system. Finally, simulation results show the effectiveness of the proposed control scheme.