Abstract: A substation pillar insulator cleaning robot based on the visual servo control method is designed,which can realize efficient,safe and dead-end cleaning of insulators in the state of no power outage,and also solve the problem of hidden safety hazards of manual cleaning. The whole and important parts of the robot are designed according to the live working environment of the robot in the substation,which can ensure the safety of the live working of the robot and completes the design of the robot visual servo control system. The real-time detection and two-dimensional positioning of the target insulator using the YOLO-v4 detection algorithm achieves a balance between detection accuracy and recognition speed,and the overall recognition rate reaches over 95%. The relative depth information between the insulator and the robot is obtained by the depth camera to convert the two-dimensional coordinates of the plane to obtain the real three-dimensional pose of the insulator to complete the positioning,and the overall positioning accuracy is within 3. 5 cm. According to the D-H method,the coordinate model of the robot connecting rod is established,and then the quintic interpolation method is used to plan the trajectory of the robot,so as to ensure the safe and stable operation of each joint of the robot. A fuzzy-PID controller is designed to control and correct the desired trajectory of the robot to ensure control accuracy and stability. The cleaning operation test and withstand voltage test were carried out in the actual environment of the substation to verify the reliability of the robot visual servo control system for efficient cleaning of different types of insulators in a live working environment. Through data comparison, it is verified that the robot can improve the cleaning efficiency by 5% and reduce the cleaning time by 8 minutes compared with manual operation.