Hierarchical Optimal Motion Planning for Piano Playing Robots With Dexterous Fingers

Abstract

Humanoid robots with dexterous operation ability have been widely used in the field of human daily entertainment. However, for music performance objectives, it is still a challenging issue to achieve efficient trajectory planning for piano playing robots with multiple degrees of freedom fingers, which can be as dexterous as a human hand and adapt to different musical styles. Based on the self-built humanoid piano playing robot, this paper proposes a continuous motion planning method from piano score to robot finger end movements. First, a hierarchical motion planning framework combining global discrete actions with local continuous actions is proposed, which simultaneously realizes finger assignment based on piano scores and continuous action for the entire robotic arm. Moreover, on the basis of the Gaussian pseudospectral method, an energy optimal motion planning method and a minimal oscillation motion planning scheme are proposed for fingers and arms, respectively, which can increase the continuity of actions, optimize working efficiency, and improve stability during piano performance.