Design of a hybrid tendon-linkage finger with coupled and adaptive motion

Hyojae Kang, Giup Seo, Jae-Young Lee, Yeong Jae Park, Jongwoo Park, Hyunmin Do, So-Hee Park, Dongil Park, Jeongdo AhnarXiv 2026 · arXiv

TL;DR

Researchers have developed a hybrid robotic finger that achieves four degrees of freedom (DoF) using only three actuators by cleverly combining tendon drives with a spring-loaded four-bar linkage. This design allows the finger to maintain precise control during free movement while automatically adapting its shape when it encounters objects, all without the bulk of additional motors.

What Problem Does This Paper Address?

In the quest for human-like robotic dexterity, engineers face a persistent trade-off: more joints (degrees of freedom) allow for better grasping, but each joint typically requires its own motor and transmission. This leads to bulky, heavy fingers that are difficult to control and maintain.

One solution is underactuation, where fewer motors drive more joints. However, purely underactuated systems often lack the "coupled" coordination needed for natural movement or the "adaptive" flexibility required to wrap around irregularly shaped objects. Furthermore, routing tendons through multiple joints often introduces friction and transmission losses. The authors tackle this by proposing a hybrid architecture that separates active remote actuation, nominal distal coupling, and contact-dependent adaptation into distinct mechanical elements.

Methods

The team from the Korea Institute of Machinery & Materials (KIMM) designed a finger with a unique transmission hierarchy:

  • MCP Joint (Base): The abduction/adduction (side-to-side) motion is driven by a bevel gear, while flexion/extension is driven by a bidirectional tendon.
  • PIP Joint (Middle): Driven by a separate bidirectional tendon. Notably, the routing of this tendon through the MCP region creates a mechanical coupling where moving the base joint naturally influences the middle joint.
  • DIP Joint (Tip): This joint has no motor. Instead, it is connected to the PIP joint via a stopper-constrained four-bar linkage equipped with a torsion spring.
Figure 1: Overall structure of the proposed robotic finger: (a) specification, (b) joint and phalanx configuration, and (c) internal transmission components with detailed views.

In "free-space" (no contact), the mechanical stopper ensures the tip moves in a fixed, predictable relationship with the middle joint. When the finger hits an object, the torsion spring allows the tip to deviate from this fixed path, "wrapping" around the object.

Datasets

This paper does not use traditional machine learning datasets. Instead, the authors validated their design through:

  • Kinematic Modeling: Developing mathematical models for the coupled and adaptive motions.
  • ArUco Marker Tracking: Using computer vision to measure real-world joint angles during 187 different motion tests.
  • Force Testing: Measuring peak fingertip force using a dedicated load cell.
  • Grasping Trials: Testing the finger against seven different objects (cylinders of varying diameters, cuboids, and irregular shapes).

Key Insights and Ideas

  • The "Stopper" Strategy: By using a mechanical stopper in the four-bar linkage, the researchers achieved the best of both worlds: the predictability of a rigid linkage during approach and the flexibility of a compliant mechanism during the grasp.
  • Intentional Parasitic Coupling: Rather than trying to eliminate the interference between the MCP and PIP tendons, the authors modeled it as a "mechanically induced contribution." This simplifies the control logic by treating the coupling as a known geometric property.
  • Remote Actuation: By placing all three motors in the "palm" housing rather than the finger links, the design achieves a slim profile (decreasing from ~18mm to ~13.6mm at the tip), which is crucial for multi-fingered hand integration.
The kinematic chain showing how tendon displacements translate to joint rotations.

Key Results

  • Precision: The measured coupled motion closely followed theoretical predictions, with a Mean Absolute Error (MAE) of only 2.54°.
  • Adaptability: The finger successfully demonstrated distinct contact postures for objects ranging from a tiny 7.9 mm cylinder to a large 90.2 mm cylinder and various cuboids.
  • Strength: Despite its slim, underactuated design, the finger produced a mean peak fingertip force of 4.56 N.
  • Workspace: The hybrid design maintains a robust reachable workspace, as shown in the kinematic analysis.
Figure 4: Reachable fingertip workspace during uncon- strained motion.

Limitations and Future Work

The authors identified a few hurdles for the next iteration:

  • Mechanical Interference: At high flexion angles (above 70°), the internal linkage components sometimes bumped into each other, preventing the finger from reaching its full theoretical adaptive range.
  • Backdrivability: While the PIP joint could be moved manually (backdrivable), the MCP joint was not, which limits the finger's ability to safely absorb impacts at the base.
  • Friction: The complex tendon routing paths introduce friction that reduces overall efficiency.

Future work will focus on simplifying tendon paths to improve backdrivability, revising the distal linkage to prevent internal interference, and eventually scaling the design to create a full multi-fingered robotic hand with an opposable thumb.

Summary updated

Abstract

Dexterous robotic fingers require coordinated joint motion and adaptability to object contact, while independently actuating multiple joints increases the complexity of the transmission mechanism. This paper presents a hybrid tendon--linkage robotic finger with four mechanical degrees of freedom driven by three actuators. The MCP abduction/adduction joint is driven through a bevel gear, while MCP and PIP flexion/extension are driven by separate bidirectional tendon transmissions. The DIP joint is not independently actuated but is coupled to the PIP joint through a four-bar linkage with a torsional spring and mechanical stopper. The tendon routing also introduces mechanical coupling between the MCP and PIP joints. Kinematic analysis was conducted to describe the coupled and adaptive motions, and experiments were performed to evaluate the joint motion characteristics. ArUco-based measurements showed that the coupled motion closely matched the theoretical prediction, whereas the adaptive motion was partially limited by interference between linkage components. The fabricated finger demonstrated different contact postures with cylindrical, cuboid, and irregularly shaped objects, and the mean peak fingertip force was 4.56 N. The proposed mechanism provides both coupled motion and contact-dependent adaptation without requiring an additional actuator for the DIP joint.
BibTeX
@misc{kang2026design,
  author = {Hyojae Kang and Giup Seo and Jae-Young Lee and Yeong Jae Park and Jongwoo Park and Hyunmin Do and So-Hee Park and Dongil Park and Jeongdo Ahn},
  title = {{Design of a hybrid tendon-linkage finger with coupled and adaptive motion}},
  year = {2026},
  eprint = {2611.11885},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.11885},
}
Figure