Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CRAFT: A Tendon-Driven Hand with Hybrid Hard-Soft Compliance

Leo Lin, Shivansh Patel, Jay Moon, Svetlana Lazebnik, Unnat Jain

Abstract

We introduce CRAFT hand, a tendon-driven anthropomorphic hand with hybrid hard-soft compliance for contact-rich manipulation. The design is based on a simple idea: contact is not uniform across the hand. Impacts concentrate at joints, while links carry most of the load. CRAFT places soft material at joints and keeps links rigid, and uses rollingcontact joint surfaces to keep flexion on repeatable motion paths. Fifteen motors mounted on the fingers drive the hand through tendons, keeping the form factor compact and the fingers light. In structural tests, CRAFT improves strength and endurance while maintaining comparable repeatability. In teleoperation, CRAFT improves handling of fragile and low-friction items, and the hand covers 33/33 grasps in the Feix taxonomy. The full design costs under $600 and will be released open-source with visionbased teleoperation and simulation integration. Project page: http://craft-hand.github.io/

CRAFT: A Tendon-Driven Hand with Hybrid Hard-Soft Compliance

Abstract

We introduce CRAFT hand, a tendon-driven anthropomorphic hand with hybrid hard-soft compliance for contact-rich manipulation. The design is based on a simple idea: contact is not uniform across the hand. Impacts concentrate at joints, while links carry most of the load. CRAFT places soft material at joints and keeps links rigid, and uses rollingcontact joint surfaces to keep flexion on repeatable motion paths. Fifteen motors mounted on the fingers drive the hand through tendons, keeping the form factor compact and the fingers light. In structural tests, CRAFT improves strength and endurance while maintaining comparable repeatability. In teleoperation, CRAFT improves handling of fragile and low-friction items, and the hand covers 33/33 grasps in the Feix taxonomy. The full design costs under $600 and will be released open-source with visionbased teleoperation and simulation integration. Project page: http://craft-hand.github.io/
Paper Structure (12 sections, 13 figures, 1 table)

This paper contains 12 sections, 13 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Kinematic Design
  4. Teleoperation and Simulation Support
  5. Experiments
  6. Structural Tests
  7. Teleoperation Tests
  8. Grasp Taxonomy Assessment
  9. Whole Arm Teleopration
  10. Structural Comparison to Representative Robot Hands
  11. Conclusion
  12. Acknowledgements

Figures (13)

  • Figure 1: CRAFT Hand. A) Full hand view showing the compact, anthropomorphic form factor with forearm-mounted actuators. B) CRAFT holding chopsticks to show complex dexterity. C) CRAFT holding a tennis ball, demonstrating its human-like size. D) One finger composed of hard PLA links and soft TPU joints. E) CRAFT conforming to a surface under contact, illustrating passive compliance at the joints.
  • Figure 2: Joints Structure. Black denotes the rigid components, while blue denotes the compliant (TPU) components. The index finger and thumb are highlighted to illustrate kinematics, with purple components indicating joint axes. Fingers feature a 2-DoF MCP and coupled PIP/DIP joints; the thumb mirrors this with a 2-DoF CMC and coupled MP/IP joints. PIP/DIP and MP/IP employ rolling contacts, while MCP/CMC use the cylindrical snap-fit interface.
  • Figure 3: Finger Structure. Dark gray denotes the hard (PLA) components and light gray denotes the soft (TPU) components. (A) PIP/DIP Tendon. (B) Bidirectional Linkage. (C) Elastic bands. (D) MCP Flexion/Extension Tendons. (E) Abduct/Adduct Tendons.
  • Figure 4: Whole-Arm Teleoperation.(A) An operator stands in front of a single RGB camera (highlighted with a dashed box). (B) FrankMocap estimates the 3D pose of the operator's wrist relative to their torso, used for whole-arm retargeting. (C) HaMeR estimates the operator's hand pose for finger retargeting. (D) The estimated poses are mapped to CRAFT, mounted as the end-effector of a robotic arm.
  • Figure 5: CRAFT in MuJoCo Simulation. CRAFT grasping a sphere, with rolling contact joints modeled as equality constraints and joints treated as revolute constraints.
  • ...and 8 more figures