SAMOVAR UMR 5157

  • Accueil
  • Accueil
  • Accueil
  • Accueil

CNRS

Rechercher




Accueil > Équipes > ACMES > Séminaires ACMES > Séminaires ACMES : > Séminaires 2017 ACMES

Séminaire ACMES présenté par Nathan Ramoly, le 03/07 à 14h en A003

Séminaire ACMES présenté par Nathan Ramoly, le 03/07 à 14h en A003

La prochaine présentation, dans le cadre des séminaires ACMES, sera faite par Nathan Ramoly.

Quand : le 03/07/2017 à 14h
Où : Salle A003

Titre : Using Semantic Based Experience and User Validation to Prevent Task Failures

Résumé : As the population is getting older, there is a growing need for domestic healthcare.
Rising technologies, including service robots, are enabling new solution. Such robots aim to help people in their everyday life by performing various tasks, from user notification to object manipulation. However, homes are not robot friendly environments, thus, robots may encounter difficulties and may fail some of their tasks, leading to possibly critical situations. The literature often proposes solutions to overcome failure situations, but not for preventing them. In this work, we present LEAF (Learning, Evaluating and Avoiding Failures), a method that allows to prevent task failures by identifying failure causes from the experience. LEAF includes the user in its learning loop and uses a multi-armed bandit solution improved with causal induction to accurately identify causes.

Short Bio : Nathan Ramoly is graduated from ISTY (Institut des Sciences et Techniques des Yvelines) in 2014 and from master COSY (of UVSQ) in the same year. During his engineering schooling, he participated for three years in the RoboCup competition supervised by Vincent Hugel.
From this experience, he was willing to keep working on robotic applications, leading to his current PhD. He is currently a third year Ph.D. student in the Computer Science department of Telecom SudParis, SAMOVAR ACMES team, under the joint supervision of Amel Bouzeghoub and Béatrice Finance. His research topics are Ambient Intelligence, Robotics, Reasoning, Machine learning, Ontology management