This paper presents the architecture of an automated UAV control system that includes an onboard computer with machine learning capabilities, sensor modules (GPS, camera, LiDAR, IMU), and communication tools for telemetry transmission. The system’s operation model is based on the use of neural networks for target detection, trajectory prediction, and obstacle avoidance. Particular attention is given to reinforcement learning algorithms, which allow the control system to independently improve its behavior based on action evaluation. The implementation potential of such systems in military applications, reconnaissance missions, and emergency monitoring is discussed. The integration of AI into control systems significantly enhances the efficiency of UAVs, reduces operator workload, and enables autonomous execution of complex tasks in real-time 1,2.
Gulenko et al. (Tue,) studied this question.