Improving Reinforcement Learning Speed for Robot Control. - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2006

Improving Reinforcement Learning Speed for Robot Control.

Résumé

Reinforcement Learning (RL) is an intuitive way of programming well-suited for use on autonomous robots because it does not need to specify how the task has to be achieved. However, RL remains difficult to implement in realistic domains because of its slowness in convergence. In this paper, we develop a theoretical study of the influence of some RL parameters over the learning speed. We also provide experimental justifications for choosing the reward function and initial Q-values in order to improve RLspeed within the context of a goal-directed robot task.
Fichier principal
Vignette du fichier
matignon2006iros.pdf (564.24 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-00333584 , version 1 (23-10-2008)

Identifiants

Citer

Laetitia Matignon, Guillaume J. Laurent, Nadine Le Fort-Piat. Improving Reinforcement Learning Speed for Robot Control.. IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS'06., Oct 2006, Beijing, China. pp.3172-3177, ⟨10.1109/IROS.2006.282341⟩. ⟨hal-00333584⟩
68 Consultations
283 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More