Abstract: To solve the positioning problem of the visual robot, based on the two-dimensional code, a method for solving camera pose was proposed to realize the precise positioning of the visual robot. Using the open source computer vision libracy（OpenCV）, the standard positions of the two-dimensional code and the camera were calibrated. The image was preprocessed to obtain the accurate edge contour of the two-dimensional code. The loop traversal and the maximum distance algorithm were used to solve the feature points, and the "pixel mutation" algorithm was used to optimize the edge distortion caused by low pixel. The feature points were used to solve the camera position through coordinate transformation, and the method was tested in actual scenes. The experimental results show that the rotation axis error of the algorithm is plus or minus 0.18 meters with the rotation angle error plus or minus 2 degrees and the total translation error plus or minus 0.02 meters, which can realize accurate and effective vision robot positioning.