Adaptive UAV maneuvering control system based on dynamic inversion and long-short-term memory network

The traditional control framework of UAV(unmanned aerial vehicle) platform is mainly applicable for stable flight. But there are still some deficiencies in solving the singularity and strong coupling problems occur in aggressive maneuver flight. This paper systematically studies the UAV's maneuver flight control method, and proposes a novel quaternion based adaptive dynamic inversion control method and a maneuver motion command generation method based on long-short-term memory network. To verify the effectiveness of the above mentioned methods, an autonomous UAV flight simulation system is established. This research helps to understand and realize the maneuver flight of fixed-wing UAVs.