An Intelligent Piano Playing Algorithm Applied to The Humanoid Robot

Abstract

Piano playing is a highly flexible and intelligent artistic task for humanoid robot due to a large number of keys. Therefore, the cooperation ability of the robotic arms and hands is essential for the smooth performance of piano humanoid robot. In this paper, we proposed an intelligent piano playing algorithm for the piano humanoid robot with a visual location module and a scoring-based action scheduling module. With the visual location module, the robot can realize the autonomous location of piano keys whether in preparation or in the process of playing the piano, so that the robot can be robust to deal with any sudden state, such as collision or missing the piano. With the relative position between the piano key and the robot as input, the action scheduling module can transform the musical score to a series of optimum combined actions of robotic arms and hands to increase the fluency of performance. Finally, the intelligent piano playing algorithm has been successfully applied on the piano humanoid robot independently developed by our team.