To design a good control system for a 6 degree-of-freedom aircraft is not an easy work. The classic linear method such as PID can’t provide a good performance for complex nonlinear systems in some cases. In this paper, an end-to-end policy using Reinforcement Learning (RL) based on Proximal Policy Optimization (PPO) is proposed. This mothed is used to established a mapping from the observed states to the motor signals to control the attitude of quadrotor. In order to evaluate the performance of the end-to-end controller based on PPO, a RL-PID controller is designed to compare with the end-to-end controller in this paper. The result shows that the performance of the end-to-end controller based on PPO is better.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Attitude Control Based on Reinforcement Learning for Quadrotor


    Weitere Titelangaben:

    Lect. Notes Electrical Eng.


    Beteiligte:
    Wu, Meiping (Herausgeber:in) / Niu, Yifeng (Herausgeber:in) / Gu, Mancang (Herausgeber:in) / Cheng, Jin (Herausgeber:in) / Wang, Yao (Autor:in) / Zhang, Weiping (Autor:in) / Mou, Jiawang (Autor:in) / Zheng, Kexin (Autor:in)

    Kongress:

    International Conference on Autonomous Unmanned Systems ; 2021 ; Changsha, China September 24, 2021 - September 26, 2021



    Erscheinungsdatum :

    2022-03-18


    Format / Umfang :

    8 pages





    Medientyp :

    Aufsatz/Kapitel (Buch)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch




    Attitude control of a quadrotor

    Dikmen, I.C. / Arisoy, A. / Temeltas, H. | Tema Archiv | 2009


    Attitude control of a quadrotor

    Dikmen, I. Can / Arisoy, Aydemir / Temeltas, Hakan | IEEE | 2009


    Attitude Control of Quadrotor UAV Based on ADRC

    Wang, Haoyu / Zhuo, Jiajin / Xiong, Xingjia | IEEE | 2023


    Full quaternion based attitude control for a quadrotor

    Fresk, Emil / Nikolakopoulos, George | BASE | 2013

    Freier Zugriff