To stabilize the autonomous underwater vehicle (AUV) in the depth plane, an adaptive two-layer supervisory controller has been designed. The high nonlinearity, time-varying, and uncertain dynamics of this robot make its control a complex problem. With the aim of depth stabilization, the linear model of the vehicle in the XZ plane is decomposed into two inner and outer models to be controlled by a dual PID controller. In the second control layer two interval type-2 fuzzy controllers adjust PID coefficients adaptively to stabilize the depth position and pitch velocity of the vehicle. As the simulation results, this combined adaptive control method guarantees the robustness performance of the controlled system against the big model parameter perturbations and measurement disturbances. However, PID and PID type-1 fuzzy controllers in some senarios cannot stabilize the position of the robot in depth.