Socio-Affective Touch in Robotics

• Auteur / Autrice : Adnan Saood
• Direction : Adriana Tapus
• Type : Projet de thèse
• Discipline(s) : Informatique, données, IA
• Date : Inscription en doctorat le 01/04/2024
• Etablissement(s) : Institut polytechnique de Paris
• Ecole(s) doctorale(s) : École doctorale de l'Institut polytechnique de Paris
• Partenaire(s) de recherche : Laboratoire : U2IS - Unité d'Informatique et d'Ingénierie des Systèmes
• Partenaire(s) de recherche : Equipe de recherche : SAR - Systèmes Autonomes et Robotique

Mots clés

Mots clés libres

Social robotics

Haptics

Tactile sensors

Machine learning

Résumé

Touch is one of the main ways for humans to coordinate and interact. Sense of Touch can help humans evaluate the properties of objects, such as size, shape, texture, and temperature. In addition, the sense of touch can also be used to detect the slippage of an object, thereby developing human knowledge of the body. The robot's touch sensing (Touch Sensing) can help the robot understand the interactive behavior of objects in the real world. These behaviors depend on their weight and stiffness, the feeling of the surface when touching, the deformation when touching, and the way of moving when being pushed. It is well known from the social sciences and medical literature that stimulation of the skin senses can exert beneficial physiological and psychological effects. Several zones that respond optimally to gentle, slow moving touch are likely to play a direct and significant role in social interactions. The goal of this thesis is the understanding of which human body zones, respond better to different force/velocity stroking/rubbing touch. This work is conducted in a human-robot interaction context. During the thesis a robotic touch system (glove, skin) will be developed. The robotic system will exhibit a socio-affective touch behavior and will adapt its behavior to the context and human user emotional state (based on the valence/arousal spectrum) and profile (personality, preferences, etc.). A model defining social touch will be developed and several machine learning techniques will be used.