Ship stabilization against roll motion caused by uncertainties, such as external waves or wind impact, nonlinear roll damping and parametric variation etc., is an important problem. Among announced approaches, the active fin stabilizer is a very effective and widely adopted procedure; however an accurate model of a whole nonlinear dynamic system is difficult to obtain. On the other hand, the mobile-wheeled platform also has highly nonlinear dynamic characteristics and it is difficult to evaluate the appropriate control effort to steer and balance on a bumpy road. To deal with these two systems, there existing some restricted resemblance between them, an intelligent roll-motion controller is developed in this paper. Firstly, to confront the uncertainties, a hetero-associative neural observer is added into a novel translated fuzzy sliding-mode controller (FSMC) to predict uncertainties. The neural observer can speed up the response and increase the self-tuning capability of the FSMC. In the proposed roll-motion controller, a compact gyroscope and accelerometer are used to detect the rolling conditions; the gathered data are sent to a microcontroller to calculate the command. Finally, to verify the effectiveness of the proposed controller, some preliminary simulations for ship stabilization are provided under the assumption that the sea surface is modeled as a one-dimensional linear free surface. Then, some experimental results are provided for a mobile-wheeled platform steered on a bumpy road. The performance is also compared with a conventional PD controller and a pure FSMC under the same conditions.