Under the requirement of environmentally friendly propulsion solutions, solar-powered aircraft has been the focus of future development due to its pollution-free characteristics which also provides a platform for distributed electric propulsion system. In this work, a lateral controller based on deep reinforcement learning framework is proposed to track the specified path through calculating commands for multiple propellers and elevator separately. Considering the high aspect ratio of the studied aircraft with distributed electric propulsion system, a 6DOF model is developed for data collection and a class of reward function is designed based on expertise. The simulation results indicate that the trained controller can realizes the path tracking in an end-to-end manner.
Lateral Control of Distributed Propulsion Aircraft Using Deep Reinforcement Learning
Lect. Notes Electrical Eng.
International Conference on Guidance, Navigation and Control ; 2022 ; Harbin, China August 05, 2022 - August 07, 2022
2023-01-31
10 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Aircraft with distributed propulsion actuators aircraft control method
Europäisches Patentamt | 2023
|