Borehole pressure relief is the primary measure to control rock bursts in high in-situ stress mines. Accurate measurement of drilling tool attitude of the drilling robot for rockburst prevention is the premise to ensure the drilling hole position and pressure relief effect. Therefore, this article proposes a drilling tool attitude calculation method based on inertial sensing assembly and BP neural network. By designing the spatial array layout of inertial sensing assembly (spatial array IMU), the data fusion model and the attitude calculation model of spatial array IMU are formulated, which could realize the high-precision calculation results of drilling tool attitude. On this basis, the error compensation method of inertial sensing units based on the BP neural network is proposed and the error compensation model of drilling tool attitude calculation is established. The feasibility of spatial array IMUs calculation and error compensation method is evaluated by analyzing the drilling tool simulation motion. Finally, the drilling tool attitude monitoring experimental platform of the drilling robot is established to compare and analyze the drilling tool calculation results of different methods. Experimental results show that after the error compensation of the BP neural network model, the attitude accuracy of the drilling tool calculated by the proposed method is significantly improved, and the average errors of azimuth, inclination, and roll angle are 0. 099°, 0. 079°, and 0. 045°, respectively. The compensation measures effectively restrain the drift and error accumulation of inertial sensing units, and there is no divergence in the error curve of drilling tool attitude calculation. Therefore, this method can continuously and reliably monitor the drilling tool attitude of drilling robot for rockburst prevention s and has high popularization and application value.