Hockens-A Hand: Underactuated Robotic Hand

• Hockens-A Hand is an underactuated adaptive robotic hand that integrates offset Hoeckens linkage, double-parallelogram, and four-bar mechanisms to deliver three distinct passive grasping modes.
• It employs passive mechanical intelligence with compliant materials and tactile sensing to enable stable, efficient grasping for parallel pinching, asymmetric scooping, and enveloping tasks.
• Its innovative design reduces actuator count while optimizing grasp force distribution and motion precision, making it ideal for service robotics and spatially constrained applications.

The Hockens-A Hand is an underactuated adaptive robotic hand that achieves high versatility and self-adaptation through a synergy of biomimetic mechanical linkages and compliant materials. Notable for its minimal actuator requirement, it integrates specialized linkage mechanisms—namely, the offset Hoeckens linkage, a double-parallelogram structure, and a four-bar trigger—for three distinct passive grasping modes: parallel pinching, asymmetric scooping, and enveloping grasping. The design harnesses passive mechanical intelligence to facilitate stable, compliant object interaction in constrained or variable environments, and is further enhanced via mesh-textured soft phalanges or multimaterial 3D-printed fingertips with integrated tactile sensing (Guo et al., 15 Oct 2025, Li et al., 2024).

1. Mechanical Architecture

The Hockens-A Hand employs a mechanical architecture entirely based on underactuation, enabling complex hand functionality with either a single linear actuator (in the Hoeckens-linkage variant (Guo et al., 15 Oct 2025)) or two actuators (in the tendon-driven Tactile SoftHand-A variant (Li et al., 2024)). The key structural components include:

Offset Hoeckens Linkage (Vertical Compliance):

• Converts actuator input to nearly linear vertical motion with minimal nonlinearity over the primary grasping range.
• Linkage AB, BC, and BD, with vector closure

$\vec{AB} + \vec{BC} = \vec{AC}$

yields a geometrically constrained output where point D's path deviates by less than 0.0164$\ell$ from linearity across 68.5°–156.6° of input.

Double-Parallelogram Linkage (Fingertip Line Contact):

• Maintains the distal phalanx (DI) vertical during initial closure, facilitating true line contact pinching for flat or thin objects.
• A return spring and vertical stopper hold DI upright in absence of load.

Four-Bar Trigger Linkage (Mode Amplification):

• Links AG (125 mm) and DG (50 mm) comprise a four-bar system, providing amplified phalanx rotation (up to ∼60°, doubling the input swing) and facilitating passive transition between grasping modes via mechanical triggers and stoppers.

Soft Phalanx and Tactile Sensing:

• Soft mesh-textured silicone for enveloping grasping (Guo et al., 15 Oct 2025).
• Multi-material, monolithically 3D-printed TacTip-inspired tactile sensors for feedback and closed-loop adaption in the SoftHand-A variant (Li et al., 2024).

2. Kinematic Derivation and Grasp Mode Optimization

The Hockens-A Hand's motion is analytically described to ensure seamless, robust transitions across three grasping regimes:

Four-Bar Intersection and Trajectory:

• System geometry governed by

$$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$

with optimum AG = 125 mm, DG = 50 mm, delivering $\Delta\theta_{\max}\approx59.8^\circ$.

• Fingertip trajectory $I(\theta_1)$ combines the nearly linear vertical displacement $D = f(\theta_1)$ and rotational mapping $\gamma = g(\theta_1)$; area swept by the fingertip during mode transition computed by the Shoelace formula ($S \approx 153.95$ mm$^2$).

Grasping Modes (Passive, Mechanically Programmed):

• Parallel Pinching: Line contact for regular objects during initial closure.
• Asymmetric Scooping: One finger forms a barrier while the other rotates outward, optimized for thin plates and environmental constraints.
• Symmetric Scooping & Enveloping: Triggering of four-bar and silicone phalanx enables secure grasping of irregular or large objects.

In tendon-driven configurations (SoftHand-A):

• Synergistic joint closure mapped via tendon displacements

$$\theta_m+\theta_p+\theta_d = \frac{\Delta L_d+\Delta L_a}{l_3+l_2+l_1}$$

for coupled closure, transitioning to isolated DIP or PIP actuation as needed.

3. Grasping Force and Power Transmission

The underactuated nature of the Hockens-A Hand leads to a power-based grasp force distribution strategy:

Input–Output Power Analysis:

• Instantaneous input power partitioned as

$\ell$0

where $\ell$1, $\ell$2 are spring-associated, and $\ell$3 the distal phalanx's work.

Normal Force Computation:

$\ell$4

• As primary input $\ell$5 increases, $\ell$6 shifts upward, proportional to actuation power, but falls off with increased distance to the fingertip ($\ell$7 effect).

In antagonist tendon systems:

• Net joint torque per joint $\ell$8:

$\ell$9

• Grasp stabilization exploits synergistic spring-coupled tendon routing and force differentials.

4. Sensing, Control, and Feedback

Integrated Tactile Sensing (SoftHand-A):

• TacTip-inspired module: Black elastomeric skin with white domed markers observed by an internal camera; marker displacement encodes contact location and normal force.
• Normal force $$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$0, where $$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$1 is the average displacement of markers.

Closed-loop and Mirrored Control:

• Gesture mirroring: Vision-based hand pose extraction is mapped to tendon displacements for open-loop teleoperation.
• Tactile feedback: Contact triggers grasp stabilization; slip detection (via marker centroid shift) actuates DIP flexion for adaptive re-gripping; control executed by PID regulation with contact/sensor-based thresholds.

Workflow stages:

1. Gesture synchronization (pre-contact)
2. Grasp stabilization (on contact detection)
3. Adaptive response to slip (contact center displacement $$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$2 triggers DIP flexion)

5. Simulation and Empirical Evaluation

Kinematic and Workspace Validation:

• Hoeckens linkage and four-bar analysis yield deviations of less than 0.5 mm in verticality and $$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$30.0164$$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$4 in output path nonlinearity.
• Fingertip workspace area $$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$5154 mm$$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$6; X–velocity peaks at 7.4 mm/s at $$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$7 s.

Physical Grasping Tests:

Grasp Mode	Test Object	Measured Range / Success Rate
Parallel pinching	ID card, orange	0–122 mm pinch, stable
Asymmetric scooping	0.5 mm PE sheet	>88% success rate
Sym. scooping/silicone	74×110×105 mm tea can	≈90% success; optimal for 60–100 mm diam.
SoftHand-A tests	Hex/tri prism, cylinder	90–100% success; 4–5 fingertips in contact

This suggests that the design reliably adapts to a wide range of object shapes, thicknesses, and environmental constraints (table edge, plate pickup) (Guo et al., 15 Oct 2025, Li et al., 2024).

Performance metrics:

• Grasp success maintained for both thin (plate) and bulky (can) objects, with slip detection latency $$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$80.2 s and gesture-mirroring delay $$\begin{cases} X^2+Y^2=L_{\mathrm{AG}}^2, \ (X-D_x)^2+(Y-D_y)^2=L_{\mathrm{DG}}^2 \end{cases}$$91 s in SoftHand-A.

6. Design Principles, Achievements, and Potential Applications

The Hockens-A Hand demonstrates the efficacy of combining underactuation, mechanical intelligence, and soft materials for robust, multi-modal grasping:

• Design Principles:
  • Single (or dual) actuator with multi-stage, mechanically-triggered grasping morphology.
  • Offset Hoeckens linkage for vertical compliance.
  • Double parallelogram for precise fingertip normal contact.
  • Four-bar trigger for controlled, amplified phalanx rotation and mode switching.
  • Soft or tactile phalanges for enhanced adaptation.
• Achievements:
  • Compact, low-cost human-like dexterity with minimal actuation.
  • Stable grasping of thin, flat, large, and irregularly shaped objects (0.5 mm sheet to 105 mm can).
  • Mechanical self-adaptation for changing environmental constraints (e.g., table-assisted scooping, enveloping).
  • Human-guided gesture mirroring and intelligent slip response in tactile variants.
• Potential Applications:
  • Service robotics (warehousing, domestic assistance).
  • Agricultural picking (compliant harvesting).
  • Safe human–robot interaction (soft, adaptive grasp).
  • Manipulation in spatially constrained environments.

A plausible implication is that integration of passive mechanical intelligence with minimal active control may provide a robust, scalable solution to adaptive grasping in emerging robotic platforms, especially where cost, reliability, and compliance are paramount (Guo et al., 15 Oct 2025, Li et al., 2024).