Intelligent Design and Verification of Aircraft Autonomous Obstacle Avoidance and Collision Avoidance System

With the rapid development of aviation industry, the obstacle avoidance challenges faced by aircraft in flight missions have become more severe. In order to improve flight safety, autonomous obstacle avoidance and collision warning system has become one of the key technologies. In this paper, an intelligent design scheme of autonomous obstacle avoidance and collision warning system for aircraft is proposed and verified. The system design combines high-precision radar, lidar sensor and stereo vision camera, and adopts parallel processing architecture accelerated by Graphics Processing Unit (GPU) to realize real-time processing of collected data and obstacle recognition. By using technologies such as Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) and Depth Q Network, the system can accurately predict the trajectory of obstacles and automatically learn the best obstacle avoidance strategy. The path planning algorithm ensures that the aircraft can effectively avoid obstacles and minimize the deviation from the predetermined route. The experimental results show that the response time of the system is less than 479 ms, the highest obstacle avoidance success rate is 100%, and the path deviation is controlled within 9 cm, which can play an important role in the autonomous obstacle avoidance of aircraft.