Robot Dance based on Online Automatic Rhythmic Perception

Rhythmic perception is intimately related to the production of rhythm in the form of music or dance movements. In this paper we present an architecture and the development of a robotic system using humanoid robots, based on the Lego Mindstorms NXT, which tries to simulate the human rhythmic perception from audio signals, and its reactive behavior in the form of dance. To do so we used a rhythmic perception model based on Marsyas, an open source software framework for audio processing, from which we construct a MarSystem with the needed blocks to perform onset feature detection. The model’s output is sent to the robot control application in real-time, via sockets, shaped in three rhythmic events, representing soft, medium and strong onsets in the music. The dance movements can be dynamically defined trough an interactive interface and are performed by the robot in a reactive manner to these rhythmic events’ occurrence. These movements also depend on two kinds of sensorial events, namely the color stepped on the floor or the proximity to some kind of obstacle. This interactive robot control keeps the dynamism manifested by the human behavior, granting spontaneous and chaotic dance movements in synchronism to music, without any previous knowledge of it.