Embodied manipulation with past and future morphologies through an open parametric hand design

Area of Science:

• Robotics and Mechanical Engineering
• Biomimicry and Evolutionary Robotics
• Human-Computer Interaction

Background:

• Human-shaped robotic hands offer advanced dexterity but are limited by customization and simulation challenges.
• Understanding the evolution of manipulation and design space is key to optimizing robotic hands.
• Existing robotic hand designs lack accessibility and struggle with real-world interaction fidelity.

Purpose of the Study:

• To introduce a novel, open parametric design for customizable and rapidly fabricated robotic hands.
• To explore the design space of robotic hand morphologies and their emergent behaviors.
• To facilitate research into embodied manipulation and surpass human-level dexterity.

Main Methods:

• Developed an open parametric design integrating simplified customization, fabrication, and control.
• Utilized nonlinear rolling joints, anatomical tendon routing, and low-degree-of-freedom actuation.
• Fabricated three distinct hand designs (human, mirrored human, aye-aye) for comparative analysis.

Main Results:

• Demonstrated rapid production of single-piece 3D-printable hands without compromising dexterous behaviors.
• Evaluated design stability and variable stiffness over two orders of magnitude.
• Manipulation tests showed unique emergent behaviors and varied object handling proficiency across hand designs.

Conclusions:

• Introduced diverse, practical, and open-source robotic hand designs for investigating embodied manipulation.
• Provided a framework for comparing and contrasting different hand morphologies and structural configurations.
• The design facilitates research into dexterous manipulation, potentially surpassing human capabilities.