A submerged body that moves near a free surface needs to keep its attitude and position to accomplish its missions, which are required to validate the performance of a designed controller before sea trial. Hydrodynamic maneuvering coefficients are generally obtained by experiments or computational fluid dynamics, but these coefficients suffer from uncertainty. Environmental loads such as wave excitation, current, and suction forces act on the submerged body when it moves near the free surface. Therefore, a controller for the submerged body should be robust to parameter uncertainty and environmental loads. In this paper, six-degree-of-freedom equations of motion for the submerged body are constructed. An adaptive control method based on the neural network and proportional–integral–derivative controller is used for the depth controller. Simulations are performed under various depth and environmental conditions, and the results show the effectiveness of the designed controller.