Ruka-v2 : Tendon Driven Open-Source Dexterous Hand with Wrist and Abduction for Robot Learning
Xinqi (Lucas) Liu 2∗ Ruoxi Hu 1∗ Alejandro Ojeda Olarte 1 Zhuoran Chen 2 Kenny Ma 1 Charles Cheng Ji 1 Lerrel Pinto 1 Raunaq Bhirangi 1 Irmak Guzey 1 · 2026
Read Ruka-v2 : Tendon Driven Open-Source Dexterous Hand with Wrist and Abduction for Robot Learning by Xinqi (Lucas) Liu 2∗ Ruoxi Hu 1∗ Alejandro Ojeda Olarte 1 Zhuoran Chen 2 Kenny Ma 1 Charles Cheng Ji 1 Lerrel Pinto 1 Raunaq Bhirangi 1 Irmak Guzey 1 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-26660.txt.
Introduction
Liu and co-authors present Ruka-v2, an open-source tendon-driven robot hand with a wrist and finger abduction. The first Ruka hand lacked wrist motion and sideways finger spread, two traits human hands use often. Ruka-v2 adds a decoupled two-axis parallel wrist for smooth bending and tilting. Its fingers now abduct and adduct, enabling thin grasps, in-hand rotation, and even calligraphy. In user studies of teleoperated tasks, the new hand cuts completion time by 51.3% and raises success by 21.2% over Ruka.
Catalog record
| CREATOR | Xinqi (Lucas) Liu 2∗ Ruoxi Hu 1∗ Alejandro Ojeda Olarte 1 Zhuoran Chen 2 Kenny Ma 1 Charles Cheng Ji 1 Lerrel Pinto 1 Raunaq Bhirangi 1 Irmak Guzey 1 |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38867 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.26660
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Ruka-v2 : Tendon Driven Open-Source Dexterous Hand with Wrist and Abduction for Robot Learning Xinqi (Lucas) Liu 2∗ Ruoxi Hu 1∗ Alejandro Ojeda Olarte 1 Zhuoran Chen 2 Kenny Ma 1 Charles Cheng Ji 1 Lerrel Pinto 1 Raunaq Bhirangi 1 Irmak Guzey 1 Abstract Abstract: Lack of accessible and dexterous robot hardware has been a significant bottleneck to achieving human-level dexterity in robots. Last year, we released Ruka , a fully open-sourced, tendon-driven humanoid hand with 11 degrees of freedom — 2 per finger and 3 at the thumb - buildable for under $1,300. It was one of the first fully open-sourced humanoid hands, and introduced a novel data-driven approach to finger control that captures tendon dynamics within the control system. Despite these contributions, Ruka lacked two degrees of freedom essential for closely imitating human behavior: wrist mobility and finger adduction/abduction. In this paper, we introduce Ruka-v2 : a fully open-sourced, tendon-driven humanoid hand featuring a decoupled 2-DOF parallel wrist and abduction/adduction at the fingers. The parallel wrist adds smooth, independent flexion/extension and radial/ulnar deviation, enabling manipulation in confined environments such as cabinets. Abduction enables motions such as grasping thin objects, in-hand rotation, and calligraphy. We present the design of Ruka-v2 and evaluate it against Ruka through user studies on teleoperated tasks, finding a 51.3% reduction in completion time and a 21.2% increase in success rate. We further demonstrate its full range of applications for robot learning: bimanual and single-arm teleoperation across 13 dexterous tasks, and autonomous policy learning on 3 tasks. All 3D print files, assembly instructions, controller software, and videos are available at ruka-hand-v2.github.io . \websitedef ruka-hand-v2.github.io \reportnumber \correspondingauthors = [email protected] , [email protected] . ∗ denotes equal contribution. Figure 1 : ( A ) Ruka-v2 has an integrated 2 DOF decoupled parallel wrist, enabling motion in restricted areas, while also introducing finger abduction/adduction to expand dexterity. ( B ) Ruka-v2 has a mount attachment that is located at the side of wrist rather than the bottom, which makes it easier to attach them to table-top manipulators. ( C ) We designed and developed an attachable magnetic encoder to better detect the joint angle accuracy, removing the requirement for expensive motion capture gloves. ( D ) We equipped Ruka-v2 with e-flesh fingertips [ 22 ] body form factor for softer and compliant grip. This enables optional equipment of tactile sensors on the hand as well. Everything is open-source and accessible at our website: ruka-hand-v2.github.io 1 Introduction Humans complete many tasks in their daily lives effortlessly. While intelligence plays a significant role in this, so does the capability of the human hand. Human hands are compact, possess more than 20 degrees of freedom (DOF), are equipped with a 3 DOF wrist, feature distributed tactile sensing, and have a compliant, soft-bodied form [ 12 , 13 , 2 , 14 ] . Inspired by these properties, many frameworks in the field have sought to replicate this dexterity in robotic systems. While the community has made significant progress on the algorithmic front [ 8 , 36 , 7 , 3 , 10 , 5 ] , hardware remains a major bottleneck. This gap has pushed research labs and commercial companies alike to improve robotic hand designs with the goal of matching human-level dexterity [ 33 , 27 , 26 , 16 , 34 , 4 , 37 , 18 , 24 , 32 ] . The most simplistic hand design is to attach an actuator directly to each finger segment, creating an exact mapping between joint values and motors. The Allegro [ 1 ] and open-sourced LEAP [ 28 ] hands are great examples of this approach. While both have been widely used in the community, they are typically bulky, heavy, and require significant algorithmic effort to close the human-to-robot embodiment gap [ 6 , 20 , 19 , 25 ] in order to complete dexterous autonomous tasks. Recently, much smaller direct-driven hands have been introduced that use smaller actuators attached to the fingers, maintaining a form factor similar to a human hand [ 33 , 27 ] . While these hands offer great dexterity and accuracy, they either overheat fairly quickly (within 30 minutes of use) [ 33 ] or are priced very high — around $50,000 [ 27 ] — limiting their accessibility. To make these hands more compact, there has been a significant amount of past work focusing on tendon-driven hand design [ 37 , 35 , 15 , 18 , 32 , 24 , 4 , 26 ] . This design typically moves all actuators outside of the hand, usually toward the wrist, and connects them to the hand joints via a rope-like mechanism that imitates a "tendon." The first commercial example of this is the Shadow Hand [ 26 ] , which demonstrated great capabilities over the years, but these hands are very expensive — priced at more than $100,000 — and are difficult to repair due to their complicated design. To make this more accessible, a growing number of tendon-driven hands have emerged from research institutions. Some of these hands are limited in dexterity, difficult to use, or offer a lower number of DOF [ 30 , 15 ] ; others, while dexterous, are either not fully open-source — preventing modification to printable files [ 4 ] — or lack sufficient documentation, causing difficulties in reproduction [ 29 , 35 , 18 , 32 , 24 ] . Notably, despite humans relying heavily on wrist motion to accomplish most daily tasks, the majority of these designs focus on the hand itself instead of the wrist. Building on our prior work Ruka , we introduce Ruka-v2 : a humanoid, tendon-driven hand featuring a decoupled parallel 2 DOF wrist and abduction/adduction capabilities across the fingers. We open-source all 3D printable files alongside documentation supported by instructional videos, teleoperate Ruka-v2 to complete 10 single arm and 3 bimanual dexterous tasks including writing, and demonstrate learned autonomous policies on 3 tasks, showcasing its utility for robot learning. We also conduct user experience experiments comparing Ruka-v2 to Ruka by asking users to teleoperate 3 different tasks while recording their completion times and success rates. We list our contributions as follows: 1. We demonstrate a 51.3% reduction in completion time and a 21.2% increase in success rate over Ruka on user experience, highlighting the need for capable wrist actuation and abduction/adduction. We further elaborate on these differences in Fig. 15 . (Section 4.5 ) 2. We elaborate on a set of design principles that make hands easier to use in robot learning, such as wrist attachments and soft fingertips. 3. We design attachable magnetic sensors per joint, that can be fitted to any Ruka-v2 (and Ruka as well) to calibrate and detect joint angle errors across all tendon-driven joints. These attachments are also open-sourced. We evaluate the accuracy of our control mechanism of Ruka-v2 in Section 4.3 and describe the sensor mechanism in detail in Section 2.5 . Collectively, we believe that Ruka-v2 will serve as a valuable tool for dexterous manipulation research, both for algorithmic development and, with modifiable CAD files, for hardware research as well. We provide a thorough comparison of Ruka-v2 to other listed robot hardware in Table