Designing an Emotionally-Intelligent Assistant of a Virtual Dance Creator

Abstract

Intelligent agents and co-robots, or cobots, become increasingly popular today as creators of digital art, including robotic or virtual dancing. Arguably, the creativity of such tools is linked to their social-emotional intelligence. In this work we question this hypothesis, extending the general paradigm of an emotionally-intelligent creative assistant (Samsonovich [1]) to virtual dance creation. For this purpose, a semantic map of dance patterns is constructed. Transitions between dance patterns are selected among local transitions on the map, following general rules. The outcome is judged by subjects as a more confident dance, compared to control conditions, when the semantic map was not used. In the proposed creative assistant of a choreographer, the state of emotional coherence of the cobot-assistant and the human user is maintained dynamically. Using the semantic map and M-schemas, the assistant will suggest variants of dance continuation, based on the current emotional state of the human choreographer and the appraisals of choices. It is expected that this approach, combining efforts of the human and the automaton working together in a state of emotional coherence, will be more user-favored and will yield higher productivity and creativity, compared to more traditional tools for virtual dance generation.