A few lessons learned in reinforcement learning for quadcopter attitude control - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2021

A few lessons learned in reinforcement learning for quadcopter attitude control

Nicola Bernini
  • Fonction : Auteur
Mikhail Bessa
  • Fonction : Auteur
Rémi Delmas
  • Fonction : Auteur
Arthur Gold
  • Fonction : Auteur
Romain Pennec
  • Fonction : Auteur
François Sillion
  • Fonction : Auteur

Résumé

In the context of developing safe air transportation, our work is focused on understanding how Reinforcement Learning methods can improve the state of the art in traditional control, in nominal as well as non-nominal cases. The end goal is to train provably safe controllers, by improving both training and verification methods. In this paper, we explore this path for controlling the attitude of a quadcopter: we discuss theoretical as well as practical aspects of training neural nets for controlling a crazyflie 2.0 drone. In particular we describe thoroughly the choices in training algorithms, neural net architecture, hyperparameters, observation space etc. We also discuss the robustness of the obtained controllers, both to partial loss of power for one rotor and to wind gusts. Finally, we measure the performance of the approach by using a robust form of a signal temporal logic to quantitatively evaluate the vehicle's behavior.
Fichier principal
Vignette du fichier
3447928.3456707 (1).pdf (994.74 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03800411 , version 1 (06-10-2022)

Identifiants

Citer

Nicola Bernini, Mikhail Bessa, Rémi Delmas, Arthur Gold, Eric Goubault, et al.. A few lessons learned in reinforcement learning for quadcopter attitude control. HSCC '21: 24th ACM International Conference on Hybrid Systems: Computation and Control, May 2021, Nashville, United States. pp.1-11, ⟨10.1145/3447928.3456707⟩. ⟨hal-03800411⟩
17 Consultations
132 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More