Optimizing Arm Movement in Cartesian- Type Piano-Playing Robot with Dynamic Programming

Fingering plays a crucial role in piano performance. Accordingly, when using a robot to play the piano, appropriate fingering and arm movement is required. Research on piano-playing using various types of robots has been conducted. However, most studies have focused on robots with structures similar to humans. As a result, the same fingering algorithm cannot be applied to robots with different structures. In previous research, a robot with a Cartesian robot structure, where one finger is assigned to each key to ensure the same key point, was used to associate sound volume with music notation. In that study, arm movement focused solely on preventing collisions, which led to unnecessary arm movements and degraded the performance. Therefore, in this study, we propose a method to optimize the robot's performance as an extension of previous research. The proposed method uses dynamic programming to calculate the cost for all possible arm positions and find the optimal path. Additionally, to prevent performance degradation, arm movement is avoided for short-duration notes. The proposed method compares the results based on the hand's playable range, and performance improvement is verified through experiments by comparing the performance of the new method and the previous algorithm on the same musical score.