Technical Challenges in Prosthetic Hands

• Electrical prosthetic hands must be:

  • Lightweight (<500g) to decrease user burden
  • Human-shaped for cosmetic purposes
  • Have motors inside for protection against damage and dirt

• Current limitations:

  • Commercial prosthetic hands achieve only static grasp postures
  • Existing in-hand manipulation approaches either:
    • Become too heavy (>1315g) due to numerous motors
    • Require complex mechanisms with external motor placement

• Daily activities requiring in-hand manipulation:

  • Reorienting objects before insertion (cards, USB devices)
  • Operating rotational tools (screwdrivers)
  • Adjusting object orientation for proper use (stamps)

Single-Axis Thumb Design

The key insight of our approach comes from research on human thumb movement. Despite the human thumb's anatomical capability for multi-axis movement, studies have shown that it primarily rotates around a single axis at the carpometacarpal (CM) joint during daily grasping tasks, particularly during fingertip-based manipulation.

Our innovation consists of two components:

1. Biological inspiration:

   • We implement a single-axis thumb design at the CM joint that mimics the primary rotational pattern observed in human thumbs
   • This simplification enables precision–lateral manipulation (transition between precision and lateral grasps) while reducing mechanical complexity

2. Strategic actuation distribution:

   • One motor for the thumb's rotational movement
   • Independent motors for index and middle fingers for precise control
   • Shared motor for ring and little fingers to reduce weight while maintaining power grasp capability

Computational Optimization for Thumb Positioning

While a single-axis design reduces complexity, the positioning of this axis is critical for successful manipulation. We developed an optimization approach to determine the ideal position:

Optimization Goal: Maximize the range of object widths that can be manipulated during precision-lateral transitions

Process:

1. Identify valid thumb configurations from the search space $\Omega$ based on the distance between the fingertips of the thumb and index finger that enable basic grasping postures

   • Add these configurations to the set of valid configurations $\Omega_{\mathrm{valid}}$

2. Calculate manipulation range $W(\bm \omega)$ for each candidate thumb configuration $\bm \omega$ in $\Omega_{\mathrm{valid}}$

   • i. Fix the index finger position and move the thumb to manipulate objects

   • ii. For each index finger position $i$, calculate the range of manipulatable object widths:

     $$W_i(\bm \omega) = [d_{\max}(i, \bm{\omega}) - (r_{\mathrm T} + r_{\mathrm I}), d_{\min}(i, \bm{\omega}) - (r_{\mathrm T} + r_{\mathrm I}) + 2\delta_m]$$

     where $d_{\max}(i,\bm{\omega})$ and $d_{\min}(i,\bm{\omega})$ are the maximum and minimum distances during manipulation, $\delta_m$ is the allowable fingertip deformation, and $r_T$ and $r_I$ are the radii of the thumb and index finger.

   • iii. Calculate the intersection of all $W_i(\bm{\omega})$ across the range of index finger positions $I$:

     $$W(\bm{\omega}) = \bigcap_{i \in I} W_i(\bm{\omega})$$

3. Extract the optimal thumb configuration $\bm{\omega}_{\text{opt}}$ from $\Omega_{\text{valid}}$ that enables the widest range of object manipulation:

   $$\bm{\omega}_{\text{opt}} = \mathop{\rm arg~max}\limits_{\bm{\omega} \in \Omega_{\text{valid}}} |W(\bm{\omega})|$$

Through this optimization approach, we explored 19,200,000 possible axis positions to find the configuration that maximizes manipulation capability while ensuring all basic grasp types remain possible.

PLEXUS Hand Design and Mechanisms

PLEXUS Hand: Lightweight (311g) with four internal motors

Four-motor actuation system:

• Thumb mechanism:

  • Single LAS16-023D linear motor
  • Piston-crank mechanism for linear-to-rotary motion conversion

• Index and middle fingers:

  • Individual LAS10-023D actuators
  • Four-bar linkage mechanisms for natural curling movements

• Ring and little fingers:

  • Shared single LAS10-023D actuator
  • Differential mechanism for adaptive grasping

(a) Four-bar linkage mechanism for finger curling (b) Differential mechanism for ring-little fingers

Basic Hand Postures

PLEXUS Hand can perform all five basic postures required for prosthetic applications:

• Power grasp: For holding larger objects and tools
• Precision grasp: For picking up small objects with fingertips
• Tripod grasp: For holding objects between thumb, index, and middle fingers
• Lateral grasp: For holding flat objects like keys
• Index pointing: For pressing buttons or pointing

Quantitative Evaluation with Various Objects

We evaluated the performance of PLEXUS Hand across:

• Primitive objects: Square prisms and cylinders (5-30mm)
• Common objects: Cards, pens, seals, spoons, screwdrivers

Success Rates for Precision-Lateral Manipulation

Real-World Tasks Demonstration

PLEXUS Hand successfully performed tasks requiring precision-lateral manipulation:

Key Advantages

• Rotational manipulation enables users to perform everyday tasks without compensatory movements
• Lightweight design (311g) reduces user fatigue
• All motors inside increases durability and improves cosmetic appearance
• Simple mechanism ensures reliability and maintainability

Summary and Future Work

• PLEXUS Hand achieves:

  • In-hand manipulation with only 4 motors
  • Lightweight design (311g)
  • Five basic hand postures
  • High success rates in precision-lateral manipulation tasks

• Future improvements:

  • Adaptive control to enhance object handling with varying shapes
  • Finger shape optimization for better contact with cylindrical objects
  • Flexible palm development for more stable object manipulation

• Potential impact:

  • Increased number of achievable activities of daily living
  • Reduced burden on users by eliminating need for compensatory movements
  • Enhanced quality of life for people with upper limb deficiencies

Acknowledgment: This work was supported by JSPS KAKENHI Grant Number JP24KJ0248.