Anti-impact drilling robot is the key equipment for underground rock burst relief in coal mine and the accurate position and attitude determination is the basis and premise for realizing unmanned pressure relief operation. By analysing the characteristics of mobile robot and coal mine equipment positioning technology at home and abroad, a position and attitude calculation method of anti-impact drilling robot based on spatial array inertial units and visual image is proposed, as shown in Fig. 1. The data fusion model of spatial array inertial units is established to derive the fusion equations of angular velocity and proportional acceleration, and the position and attitude calculation process of spatial array inertial units is designed. By analysing the actual operating conditions of the anti-impact drilling robot, the curve motion trajectory is simulated, and the position and attitude calculation method is simulated and analysed. The simulation results show that the proposed method is better than the original method and the differential method in terms of displacement cumulative error and attitude angle average absolute error. On this basis, a binocular vision position estimation model based on continuous frame images is established, and an error elimination method based on visual correction is designed. An experimental platform for monitoring the position and position of drilling robot is built, and the feasibility and superiority of the proposed method are verified. This method can continuously and reliably monitor the position and attitude of anti-impact drilling robot and has high popularization and application value.