Learning

An updated version of this page will be available in March

 

Autonomous robots are able to achieve more and more complex tasks, relying on sensory-motor functions. To better understand their behavior and improve their performance, it becomes necessary to characterize and to model at the global level how robots behave in a given environment

We have developed a general framework for learning, from observation data, the behavior model of a robot when it is performing a given task. The behavior is modeled as a Dynamic Bayesian network (DBN) that can be learned (using a modified version of Expected Maximization with Particle Filtering) and it can be used to improve on line the robot behaviors

Relevant publications

M.FOX , M.GHALLAB , G.INFANTES , D.LONG, Robot introspection through learned hidden Markov models, Artificial Intelligence, Vol.170, N°2, pp.59-113, February 2006

G.INFANTES , F.INGRAND , M.GHALLAB, Learning Behaviors Models for Robot Execution Control. ECAI 2006 The 17th European Conference on Artificial Intelligence, Riva del Garda (Italie), August 2006

N.DO HUU , W.PAQUIER , R.CHATILA, Combining structural description and image-based representation for image, object, and scene recognition. 19th International Joint Conference on Artificial Intelligence (IJCAI'05), Edinburgh (GB), July  2005, pp.1452-1457

W.PAQUIER , R.CHATILA, Learning new representations and goals for autonomous robots. Rapport LAAS N°02404, 2003 IEEE International Conference on Robotics and Automation (ICRA'2003), Taipei (Taïwan),  September 2003, pp.803-808

(not so) Recent theses

N.DO HUU Apprentissage de représentations sensori-motrices pour la reconnaissance d'objet en robotique, Doctorat, Université Paul Sabatier, Toulouse, 1er décembre 2007

G.INFANTES Apprentissage de modèles de comportement pour le contrôle d'exécution et la planification robotique, Doctorat, Université Paul Sabatier, Toulouse, 5 octobre 2006

W. PAQUIER Apprentissage ouvert de représentations et de fonctionnalités en robotique : anayse, modèles et implémentation, Doctorat, Université Paul Sabatier, Toulouse, 19 Mars 2004, 147p.

B. MORISSET Vers un robot au comportement robuste. Apprendre à combiner des modalités sensori-motrices complémentaires, Doctorat, Université Paul Sabatier, Toulouse, 28 Novembre 2002, 158p