Mapping Music onto Robot Joints for Autonomous Choreographies: PCA-Based Approach

Abstract

A dance choreography is the kinematic interpretation of the music that translates the sound into rhythmic and synchronized movements. The dance makes the sounds visible through body language and the performer's movements, enhancing and augmenting the audience's experience. This work introduces an autonomous approach to augment artistic performance through music-driven choreographies tailored for robots: well-known methodologies to analyze musical pieces are integrated with a novel approach based on principal component analysis to map the audio tracks onto the robot actuation space. To validate our approach's effectiveness, we test the adaptability to various music genres and styles, evaluating the arose coherence between robot movement and audio tracks in terms of dissimilarity of audio-movement dynamic wrapping before and after the proposed mapping. The average distance reduction observed in the dataset of one hundred choreographies is equal 55.38 ± 6.38%.