Thèse Humor Generation In Robots Integrating Llms Emotions Gestures And Incongruity Theory H/F

École nationale supérieure de techniques avancées

This thesis investigates the creation of humor-capable robots by combining Large Language Models (LLMs), emotional frameworks, gesture generation, and the principles of Incongruity Theory. The research aims to develop robots that can understand, generate, and express humor through both verbal and non-verbal means, enhancing natural human-robot interactions. Incongruity Theory, which suggests humor arises from unexpected or mismatched elements, provides a foundation for generating humorous content that feels spontaneous and engaging. Additionally, the integration of gesture generation synchronized with emotional and linguistic cues enables robots to deliver humor in a way that mimics human timing and delivery. This multi-faceted approach aims to achieve humor that is contextually appropriate, emotionally resonant, and socially engaging, advancing robots' ability to foster rapport and empathy with users.

Social Robotics; Machine Learning, Social Norms; LLMs