Design an anthropomorphic dexterous hand for expressive piano performance

Area of Science:

• Robotics
• Biomechanics
• Music Technology

Background:

• Expressive piano performance requires high-speed, forceful, and coordinated robotic actions, which current systems struggle to replicate.
• Existing robotic hands lack the dexterity, speed, and force output necessary for human-level piano playing.
• This research introduces a data-driven, bio-inspired dexterous robotic hand engineered for high-fidelity piano performance.

Purpose of the Study:

• To develop a robotic hand capable of replicating the complex demands of expressive piano performance.
• To bridge the gap between human motor control in music and robotic manipulation capabilities.
• To create a hardware foundation for autonomous musical performance.

Main Methods:

• Extracted kinematic primitives and joint coupling patterns from professional pianist motion capture data.
• Integrated human motion priors into the robotic hand's mechanical design via morphological coupling and actuator allocation.
• Selected actuators based on biomechanical data for human peak velocities and force profiles to ensure adequate performance bandwidth.

Main Results:

• The robotic hand demonstrated human-like joint coordination and achieved peak joint velocities of 53.88 rad/s.
• Sufficient fingertip force was generated for authentic piano interaction.
• Successfully performed a Grade 7 piano piece, 'Croatian Rhapsody', showcasing its expressive capabilities.

Conclusions:

• Established a pathway from human motion statistics to embodied robotic intelligence for musical performance.
• Provided a high-performance hardware solution for autonomous musical expression.
• The developed robotic hand advances the field of human-robot interaction in artistic domains.