Abstract: Aiming at problems of insufficient positioning and control accuracy of tunneling robot in environment with high dust and low illumination of coal mine, a deviation correction control system of rapid tunneling robot based on vision measurement is designed. The explosion-proof camera installed on rapid tunneling robot is used to collect image of laser pointing instrument at the rear, and the collected image information is transmitted to explosion-proof computer through Ethernet. The explosion-proof computer is used to preprocess the image and then the pose of the rapid tunneling robot is calculated according to pose calculation model. The calculated pose information is compared with designed roadway axis information, and the pose deviation between the pose of the rapid tunneling robot and the roadway design axis is calculated. According to type and size of the pose deviation, different control strategies are adopted to calculate corrected control quantity and control instructions are output. The control command controls spool movement of solenoid proportional valve to control telescopic oil cylinder. According to the different expansion amount of the telescopic cylinder in each area, the pose of the rapid tunneling robot is adjusted to realize deviation correction control. The experimental results show that the average deviation of vision meosurement accuracy in X direction is 21.334 mm, that in Y direction is 34.154 mm, that of yaw angle is 0.493 ° and the error of deviation correction control in X direction is less than 30 mm, which meets requirements of actual working conditions.