Real-Time Adaptation of a Robotic Joke Teller Based on Human Social Signals

Abstract

Humor is an essential element of human-human communication. Consequently, robots in the role of companions should exploit its potential as well to make interactions more enjoyable. Using a robot as an entertainer requires finding out what kind of humor its audience prefers. However, it is a challenging task for a social robot to learn what users prefer without bothering them by repeatedly asking questions. In this paper, we present an approach based on Reinforcement Learning that enables a robot to continuously adapt to the users' humor preferences without requiring them to explicitly provide feedback. Instead, we designed the robot to analyze the user's ideomotor social cues. We evaluated our approach in a scenario involving a Reeti robot acting as an entertainer. In this role, it is telling different types of jokes, (possibly) underlining its performance with grimaces and sounds. The adaptation process is accomplished only by using the audience's vocal laughs and visual smiles, but no other form of explicit feedback.