To detect a ramp and guide a tracked robot to complete the ramp-climbing task in an unknown environment, four tracked robot control modes are proposed in this paper. The modes are the exploration mode, aligning mode, calculating the tilt angle mode, and climbing mode. This self-made tracked robot only equips with a Kinect depth sensor; unfortunately, the sensing depth data, in general, incorporate with measuring uncertainties. To attenuate effects from uncertainties, this paper employs the type-1 fuzzy controller to take responsibility of speed control and direction control, and the interval type-2 fuzzy decision strategy to stabilize unnecessary variations occurred in the moment of switching between the four modes. Finally, experiments of autonomous ramp-climbing illustrate the performance of the proposed control scheme for the tracked robot.