The accurate auto-position is on the basis of many smart actions for mobile robots. The higher cost performance has much significance to the producing process of brainpower robots. This paper works on a differential wheels mobile robot platform. And it researches on the mobile robot's auto-position in a complementary manner of information fusion. The sensing equipments contain an Inertial Measurement Unit (IMU) and a speedometer. This preliminary fusion process deals with the robotic movement velocity and angular velocity on data plane. And then, the further fusion on estimation plane, takes advantages of Unscented Kalman Filter (UKF) to estimate the position and orientation of the robot. The system test proves that the position method referred in this paper is able to meet the application. In addition, the real-time and accuracy has obtained a satisfactory effect.