Ruka-v2: Tendon Driven Open-Source Dexterous Hand with Wrist and Abduction for Robot Learning

Xinqi Liu, Ruoxi Hu, Alejandro Ojeda Olarte, Zhuoran Chen, Kenny Ma, Charles Cheng Ji, Lerrel Pinto, Raunaq Bhirangi, Irmak Guzey

Keywords: Dexterous manipulation, humanoid hand, tendon- driven, robot learning, teleoperation

TL;DR: Ruka-v2 is a fully open-sourced, tendon-driven humanoid hand that overcomes a hardware bottleneck in achieving human-like dexterity by adding a 2-DOF parallel wrist and finger abduction all while maintaining a total material cost under $2,000.

Abstract: Lack of accessible and dexterous robot hardware has been a significant bottleneck to achieving human-level dexterity in robots. Previously, we released RUKA, a fully open-sourced, tendon-driven humanoid hand with 11 degrees of freedom, buildable for under $1,300. Despite its contributions, RUKA lacked wrist mobility and finger adduction/abduction—two degrees of freedom essential for imitating human behavior. 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 enables manipulation in confined environments such as cabinets, while abduction enables motions such as grasping thin objects and calligraphy. We evaluate RUKA-v2 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 teleoperation across 13 dexterous tasks and autonomous policy learning on 3 tasks. All design files, assembly instructions, and software are open-sourced.

Submission Number: 16