33th Congress of the International Council of the Aeronautical Sciences

06.2 - Flight Dynamics and Control (UAV related)

USING DEEP REINFORCEMENT LEARNING TO IMPROVE THE ROBUSTNESS OF UAV LATERAL-DIRECTIONAL CONTROL

R. Wang¹, Z. Zhou¹, X. Zhu¹, L. Zheng, Delft University of Technology, Netherlands; ¹Northwestern Polytechnical University, China

Inspired by PID and LQG control method, this paper presents an improved TD3 deep reinforcement learning Lat-Dir flight control law design method. Flight simulations are carried out in the nominal state and deviation state respectively. The results show that the TD3 controller has the advantages of clear physical meaning, simple structure of policy network and strong performance robustness.


View Paper