Autonomous deviation-correction algorithm with low computational requirements for mobile equipment in underground coal mines

As coal mining advances toward intelligent and manpower-reduction operation, the need for autonomous relocation of underground mobile equipment is becoming increasingly evident. To address the high computational-resource requirements of existing roadway navigation methods, the difficulty in adapting these methods to the limited computing power of underground explosion-proof computers, and their limited adaptability to complex roadway scenarios, this study proposed an autonomous deviation-correction algorithm with low computational requirements for underground mobile equipment in coal mines. The algorithm fused spatial difference and temporal increment features in measurement signals from multi-angle distance sensor arrays on both sides of the vehicle and directly updated the vehicle's linear and angular velocities to sense lateral offset, heading deviation, and changes in the roadway boundary ahead and perform deviation correction accordingly. To improve control performance, a multi-scenario comprehensive evaluation function was constructed, and a genetic algorithm was used to optimize the sensor installation angles and control parameters. Analysis showed that, with a fixed number of sensors, the per-cycle computational complexity of the algorithm was O(1). Simulation results showed that, with sensor measurement noise and vehicle dynamic response lag considered, the vehicle was able to complete autonomous travel with deviation correction through a 200 m comprehensive roadway. Except in local transition regions near sections with abrupt roadway changes, the lateral deviation did not exceed 15 cm. Prototype testing showed that, at a control update frequency of approximately 4 Hz, the vehicle was able to complete autonomous travel with deviation correction in a 10 m straight simulated roadway. This result preliminarily demonstrates the engineering feasibility of the proposed algorithm.