[Paper Review] Design and Kinematic Optimization of a Novel Underactuated Robotic Hand Exoskeleton

Mine Saraç, Massimiliano Solazzi+3 more|arXiv (Cornell University)|May 7, 2020

Stroke Rehabilitation and Recovery16 references37 citations

TL;DR

The paper proposes a novel underactuated, linkage-based hand exoskeleton with a single linear actuator per finger, analyzes its kinematics and statics, and optimizes link lengths to maximize force transmission and grasp adaptability.

ABSTRACT

This study presents the design and the kinematic optimization of a novel, underactuated, linkage-based robotic hand exoskeleton to assist users in performing grasping tasks. The device has been designed to apply only normal forces to the finger phalanges during flexion/extension of the fingers, while providing automatic adaptability for different finger sizes. Thus, the easiness of the attachment to the user's fingers and better comfort have been ensured. The analyses of the device kinematic pose, statics, and stability of grasp have been performed. These analyses have been used to optimize the link lengths of the mechanism, ensuring that a reasonable range of motion is satisfied while maximizing the force transmission on the finger joints. Finally, the usability of a prototype with multiple fingers has been tested during grasping tasks with different objects.

Motivation & Objective

• Introduce a planar, underactuated hand exoskeleton that applies only normal forces to finger phalanges.
• Ensure automatic adaptability to different finger sizes and grasp shapes without mechanical adjustments.
• Analyze pose, statics, and grasp stability to optimize link lengths for high force transmission.
• Demonstrate feasibility with multi-finger prototype and assess grasping with various objects.

Proposed method

• Model the finger as a planar open-chain excluding thumb ab/adduction.
• Attach passive sliders to finger phalanges and drive each finger component with a single linear actuator.
• Perform pose analysis using four independent vector loops to obtain a unique configuration from finger pose and actuator displacement.
• Derive a differential kinematics Jacobian to relate actuator and measured velocities to finger joint velocities.
• Conduct static grasp analysis via inverse Jacobian transpose to relate actuator force to finger joint torques.
• Optimize link lengths through a constrained exhaustive search with sensitivity analysis to maximize a combined torque measure while respecting joint ranges and safety ratios.

Experimental results

Research questions

• RQ1 How can an underactuated linkage-based hand exoskeleton achieve stable, perpendicular force transmission to finger phalanges during grasping?
• RQ2 What link-length configuration maximizes force transmission while preserving natural joint ranges for different finger sizes?
• RQ3 Can a single actuator per finger suffice for natural grasping across objects with varying shapes and sizes?
• RQ4 What are the kinematic and static constraints necessary to ensure safe and effective grasping with the proposed device.

Key findings

• The optimized multi-finger exoskeleton enables realistic finger joint ranges: MCP 0–80°, PIP 0–90° for the index finger.
• A sensitivity analysis identifies six key link lengths (EJ, CI, EF, ED, CD, BC) as most influential for performance.
• Exhaustive search with constraints increases the performance index p by about 50% over initial configurations.
• Prototype index-finger implementation achieved maximum flexion/extension configurations via 3D printing.
• The design accommodates small, medium, and large hand sizes, indicating broad usability.