A Music-Driven Dance System of Humanoid Robots

Abstract

Robot dance is an important topic in robotics. Conventional robot dance systems mainly rely on beats or rhythms of music; however, these conventional systems suffer from limited dance styles and less action novelty. In this paper, we instead develop a humanoid robot dance system driven by musical structures and emotions. In the proposed system, a musical phrase and a dance phrase are considered as the basic structural units of music and dance, respectively. A musical phrasing algorithm based on music theories is created to divide a piece of music into a sequence of phrases. When the emotion of each phrase has been recognized, an emotion sequence can be established. Meanwhile, a hidden Markov model (HMM) matches a dance phrase sequence to the emotion sequence. In particular, several concepts of the “chance method” created by choreographer Merce Cunningham are adopted to guide our robot dance system; thus, a dance phrase is choreographed by randomly selecting and combining a number of actions from a predesigned action library. Based on the approach, one music can generate diverse robotic dance motions, showing the novelty and diversity of robot dance. The experiments on our humanoid robot “Alpha1 Pro” show that our robot can do a good job dancing to music according to musical structures and emotions and can be well accepted by various people.