Tube-sheet positioning robot is used for the heat transfer tube maintenance work of steam generator in high-temperature gas-cooled reactor nuclear power plant. As the carrier of pipe-plugging tools, the robot is equipped with various pipe-plugging execution units and guided to the designated working pipe hole to implement pipe plugging process. According to the construction requirements, a tube-sheet positioning robot was studied herein. Limited by the inspected objects and construction conditions, the tube-sheet positioning robot adopted modular and lightweight design. In order to eliminate the influences of the stress deformation of the tube-sheet structure, the motion error of the robot itself and the deformation error caused by the attitude adjustment during the working process of the tube-sheet positioning robot, a machine vision-aided positioning design scheme was adopted to eliminated the above-mentioned nonlinear errors and improve the position control accuracy of the tube-sheet positioning robot. The designed robot prototype was tested on a steam generator simulator. The results show that the application of machine vision technology for target positioning and motion feedback control is an effective means for nonlinear error compensation. This method has high control accuracy and relatively simple as well as reliable control system.