LEAP Hand: Low-Cost, Efficient, and Anthropomorphic Hand for Robot Learning

Abstract

Fig. 1: (a) LEAP Hand is an anthropomorphic dexterous robot hand designed for robot learning research. It can be assembled in under 4 hours for 2000 USD, is composed of readily available parts, and is robust. (b) to-scale comparison of LEAP Hand and a human hand (c-h) LEAP Hand in different power and precision grasps holding common objects. The hand design and code will be open-sourced to democratize access to hardware for anthropomorhic dexterous manipulation. Videos at https://leap-hand.github.io/Abstract-Dexterous manipulation has been a long-standing challenge in robotics. While machine learning techniques have shown some promise, results have largely been currently limited to simulation. This can be mostly attributed to the lack of suitable hardware. In this paper, we present LEAP Hand, a lowcost dexterous and anthropomorphic hand for machine learning research. In contrast to previous hands, LEAP Hand has a novel kinematic structure that allows maximal dexterity regardless of finger pose. LEAP Hand is low-cost and can be assembled in 4 hours at a cost of 2000 USD from readily available parts. It is capable of consistently exerting large torques over long durations of time. We show that LEAP Hand can be used to perform several manipulation tasks in the real world-from visual teleoperation to learning from passive video data and sim2real. LEAP Hand significantly outperforms its closest competitor Allegro Hand in all our experiments while being 1/8th of the cost. We release the URDF model, 3D CAD files, tuned simulation environment, and a development platform with useful APIs on our website at https://leap-hand.github.io/

Learning dexterity from human hand motion in internet videos

“…We call this system Videodex as it learns dexterity from video. This builds on work published at CoRL 2022 by Shaw et al, (2023b).…”

“…We collect separate demonstrations on the real robot using the 2-finger gripper from xArm UFactory. Separate action priors are trained for the 16 DoF LEAP Hand (Shaw et al, 2023a) and the 2finger gripper.…”

“…We found that the Allegro had higher inaccuracy in control and more hardware failures as compared to LEAP Hand. LEAP Hand outperformed the Allegro Hand 7 À 12% on average in all experiments, thus, we use it for our setup Shaw et al (2023a).…”

A Grasp Taxonomy for People with C5-7 Spinal Cord Injury

“…Egocentric video capture is proposed as a potential alternative solution and a powerful new evaluation tool for clinicians and researchers to evaluate hand and device use in unstructured environments. This has particular promise with the development of new ML tools for grasp/hand detection to reduce the laborious manual video tagging process for data analysis [17], [18]. In fact, this population has already demonstrated its openness to the use of this tool [19].…”