PianoPal: A Robotic Multimedia System for Interactive Piano Instruction Based on Q-Learning and Real-Time Feedback
摘要
Traditional piano learning often requires significant equipment and skill, which can be intimidating for novices. This paper introduces PianoPal, a multimedia piano instruction system based on robotic interaction, designed for novices. The system includes three modules: intelligent performance, piano interaction, and real-time feedback. To test PianoPal’s effectiveness, we conducted two experiments. The first evaluated the fingering planning function using Q-Learning, showing it outperformed alternatives across all metrics. The second experiment involved 24 novice users, with pre- and post-experience questionnaires and interviews indicating that PianoPal improved one- and two-handed performances and was effective in the short term. Overall, PianoPal proves effective in aiding novice piano learning and has been successfully integrated into our robotic piano device, laying a solid foundation for future piano tutoring robot developments.