Aiming at trajectory tracking of the visionguided AGV with twodimension code, a fuzzy PID rectification algorithm based on optimal deviation path is proposed. Firstly, the kinematics equation of the AGV is established, using the lateral deviation and course deviation as the input variables of the control system. Secondly, the optimal deviation path and optimal control equation are obtained based on the optimal deviation transformation strategy by introducing the Hamilton optimal control function. Finally, the fuzzy PID controller’s parameters are updated according to pose error between the AGV and the optimal deviation path, to achieve the differential velocity adjusting real time. As a result, the AGV drives along the optimal deviation path, and achieves optimal control. The experimental results show that the proposed method can eliminate the lateral and course deviation smoothly and quickly. In extreme deviation, the four kinds of membership interval deviation correction results of this method are 2.38, 2.54, 3.29 and 4.43 mm. The results are all less than 5 mm, and the rectification distance is less than 1.2 m, and trajectory tracking accuracy is 3.2 mm.Not only the navigation precision improves, but also the stability of the operation and driving capability of the nontrack AGV are met.