Vai al contenuto
Carrello
0 elementi

Sci-Articles-Flexible-TPU-Filaflex

RUKA: Rethinking the Design of Humanoid Hands with Learning

di Ignacio García 23 Apr 2025

RUKA: A Humanoid Robotic Hand Empowered by 3D Printing and Flexible Materials

RUKA is a compact, low-cost, open-source robotic hand that combines 3D-printed components and learning-based control to achieve human-like dexterity. With soft pads made of Filaflex Foamy and a tendon-driven design, RUKA excels in teleoperation, grasping performance, and accessibility, setting a new benchmark for affordable robotics.

 

Human-Centered Design: Functionality and Accessibility

The RUKA hand was designed to closely mimic the shape and size of a human hand using off-the-shelf components and 3D-printed parts. Its soft pads—made from Filaflex Foamy TPU—enhance grip and protect against impacts. These pads are essential for handling real-world objects in human environments.

Impressively, RUKA can replicate 29 out of 33 grasps from the GRASP taxonomy, including power and precision grips. Its anthropomorphic dimensions also allow seamless integration with motion-capture gloves like MANUS.


Filaflex Foamy pads used on RUKA’s fingers and palm for better grip and impact absorption.

Flexible Materials Meet Robotic Power

RUKA’s soft components—crafted with Filaflex Foamy—make it capable of grasping a wide range of objects with stability and adaptability. In strength tests, RUKA achieved:

  • Pinch force: 2.74 N
  • Payload: 6.0 kg
  • Slip force (DIP/PIP): 33.02 N
  • Slip force (MCP): 16.15 N

These numbers surpass other commercial hands like LEAP or Allegro, highlighting the performance edge of using flexible, 3D-printed materials.



Learning-based controller architecture using MANUS gloves to predict motor positions from fingertip data.

Smarter Control Through Learning

Controlling tendon-driven hands is complex. RUKA tackles this by using deep learning models trained on data collected via MANUS gloves. These models map fingertip or joint positions to motor commands using LSTM neural networks.

This setup enables high-frequency teleoperation and precise imitation of human hand movements without internal joint sensors—significantly lowering costs and mechanical complexity.

Open, Modular, and Maker-Friendly

RUKA is fully open-source: from CAD files in OnShape to control code and a MuJoCo simulation model. Its 3D-printed construction allows quick repairs and customization. The total assembly time is around 7 hours, and most repairs take less than 20 minutes.

Multiple configurations are available—from a $500 lightweight version to a $1300 high-torque version. This makes RUKA a flexible platform for makers, researchers, and educators.

Conclusion: A Robotic Hand Built on Flexibility

RUKA shows what’s possible when flexible materials like Filaflex Foamy meet open-source robotics and smart control systems. It brings human-like dexterity into reach—literally—for developers and researchers looking to build the next generation of humanoid interaction.

Source: “RUKA: Rethinking the Design of Humanoid Hands with Learning” by Zorin, Guzey, Yan, Iyer, Kondrich, Bhattasali, Pinto. NYU Robotics. Read the paper.

Articolo precedente
Prossimo articolo

Grazie per esserti iscritto!

Questa email è stata registrata!

Acquista il look

Scegli le opzioni

Recreus
Modifica opzione
Notifica Disponibilità in Magazzino

Scegli le opzioni

this is just a warning
Login