Intelligent-assisted detection-based mission reliability enhancement method for unmanned aerial vehicles in mining caverns

As the demand for unmanned aerial vehicle (UAV) technology in mine cave inspection missions increases, the complexity of the mine environment poses a great challenge to the safe flight of UAVs. Traditional hazardous state inspection methods rely on sensor data and preset rules, but the reliability of existing systems is insufficient when facing complex events such as wind speed fluctuations and cave wall collisions. To solve this problem, this paper proposes an intelligent assisted detection system ba sed on flight state invariants, which combines multimodal sensor data, deep learning and reinforcement learning techniques, and is capable of detecting and responding to complex events in the mine cave environment in real time. The flight data in the mine cave environment are collected by Airsim simulator, and 50 cave wall collisions and 50 turbulent wind introduction experiments are conducted. The experimental results show that the detection accuracy of the intelligent assisted detection system in complex environments is 78% higher than that of traditional methods, with a significant reduction in missed and false alarms, and a 30% decrease in the false positive rate. Compared with Kalman Filter and traditional methods, the intelligent-assisted detection system performs better in complex flight environments and is able to effectively deal with dynamically changing mine cave environments. This study provides a new idea to improve the reliability of UAVs in complex environments, which has important theoretical value and practical application significance.