A Robotic Dance System Driven By Music And Its Computational Implementation

As one of the most important directions in artificial intelligence,robotic dance recently has drawn more and more researchers' attention.Its research has a vital significance to both humans and robotics.Many robotic dances are preprogrammed by choreographers for a particular piece of music so that the motions can be smoothly executed and synchronized to the music.But this method requires manual intervention,increasing the burden on choreographers and resulting in the sameness of robotic dances as well as low intelligence level of robots.After the creation of robotic choreography turned to be developed by artificial intelligence methods,many researchers focused on analyzing beat,tempo,rhythm features and so on,ignoring the musical structure,correspondence between the structure of music and dance,and the choreography theory.In this paper,a robotic dance system driven by music has been built.The solution is to guide the robot to dance along with the musical structure,and the choreography should reflect the emotional character of the music.Besides,the connection between dancing motions use chance choreography for references.The main work in this paper are as follows:(1)analyze the musical structure(2)recognize the musical emotion(3)find a novel way to match musical features and dancing motions.In the music data pre-processing stage,MIDI files are parsed to extract pitch,duration,musical tempo and so on.By discriminating the concepts of phrase in different musical literatures,a set of rules are summarized to segment phrase correctly,which are universal in music field.On the basis of the rules,an algorithm is presented to segment phrase automatically.The proposed model is tested against Local Boundary Detection Model.The result suggested that the proposed model is more accurate.The features that can reflect musical emotion are discussed and calculation methods are given to extract musical emotion feature vector.Due to the subjective nature of human perception,fuzzy logic is a more appropriate mathematical tool for emotion detection.In musical emotion recognition stage,fuzzy c-means algorithm is used to divide the music phrases into different clusters.Inspired by chance choreography developed by Cunningham,Hidden Markov Model is used to match the musical emotion sequence with the dancing motion sequence.After the completion of the system,the robot can perform different dances synchronized to different music,reflecting some originality and exhibiting autonomous behavior.This research has a vital academic significance to enrich and develop the computational method of autonomous robotic dance.