Thèse Ai-Driven Socio-Affective Touch In Robotics H/F

École nationale supérieure de techniques avancées

Touch is a fundamental channel of human communication, conveying emotions such as comfort, trust, and empathy through subtle variations in pressure, warmth, and motion. Yet, most social robots today lack the ability to feel or express touch in meaningful ways. This thesis aims to develop a new multimodal tactile intelligence framework that enables robots to sense, interpret, and respond to social touch in human-robot interaction. The research project will advance the state of the art in tactile sensing and affective robotics by designing multimodal haptic interfaces capable of measuring distributed pressure, temperature, and friction across the robot's hands and body, and by developing deep learning models that integrate tactile and physiological cues to infer human affective intent. The research addresses the challenge of interpreting the subtle social cues embedded in human touch and translating them into contextually appropriate robot responses that foster trust, cooperation, and smoother integration of robots into human environments. Leveraging AI-driven modeling, the work examines how robots can understand, predict, and adapt to social interaction dynamics.
It is structured around three core objectives: (1) Develop multimodal haptic interfaces that sense and express distributed pressure, friction, and temperature across the robot's body, forming the physical basis for social tactile communication; (2) Recognize human behavior and affective states from tactile and physiological cues using AI and GenAI, enabling context-aware and empathetic robot responses; and (3) Evaluate human perception, acceptance, and socio-affective impact through controlled studies, including handshakes, affectionate contact, and comforting gestures, to assess trust, empathy, and engagement. Together, these components aim to create robots that can accurately interpret, express, and respond to human touch, enhancing social integration and emotional connection.

This work is in collaboration with Prof. Shan Luo from KCL, UK

Human touch is a rich channel for social and emotional communication. In social robotics, enabling robots to perceive and express affective touch is crucial for natural and adaptive interactions. AI and Generative AI provide tools to model, predict, and generate socio-affective robot behaviors.