Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Designing Anthropomorphic Soft Hands through Interaction

Pragna Mannam, Kenneth Shaw, Dominik Bauer, Jean Oh, Deepak Pathak, Nancy Pollard

TL;DR

This work addresses the challenge of designing highly dexterous soft hands by bridging rapid fabrication and real-world evaluation through teleoperation. The authors prototyped a $16$-DoF tendon-driven soft hand (DASH) via 3D printing and iteratively refined its geometry over five iterations using teleoperated tasks. They evaluate on a suite of $30$ manipulation tasks, totaling $900$ demonstrations, and show that the final DASH ($v5$) solves $19$ of the $30$ tasks, outperforming the Allegro baseline ($7$/30). They open-source the CAD models and the teleoperation dataset to enable community use and benchmarking.

Abstract

Modeling and simulating soft robot hands can aid in design iteration for complex and high degree-of-freedom (DoF) morphologies. This can be further supplemented by iterating on the design based on its performance in real world manipulation tasks. However, iterating in the real world requires an approach that allows us to test new designs quickly at low costs. In this paper, we leverage rapid prototyping of the hand using 3D-printing, and utilize teleoperation to evaluate the hand in real world manipulation tasks. Using this method, we design a 3D-printed 16-DoF dexterous anthropomorphic soft hand (DASH) and iteratively improve its design over five iterations. Rapid prototyping techniques such as 3D-printing allow us to directly evaluate the fabricated hand without modeling it in simulation. We show that the design improves over five design iterations through evaluating the hand's performance in 30 real-world teleoperated manipulation tasks. Testing over 900 demonstrations shows that our final version of DASH can solve 19 of the 30 tasks compared to Allegro, a popular rigid hand in the market, which can only solve 7 tasks. We open-source our CAD models as well as the teleoperated dataset for further study.

Designing Anthropomorphic Soft Hands through Interaction

TL;DR

This work addresses the challenge of designing highly dexterous soft hands by bridging rapid fabrication and real-world evaluation through teleoperation. The authors prototyped a -DoF tendon-driven soft hand (DASH) via 3D printing and iteratively refined its geometry over five iterations using teleoperated tasks. They evaluate on a suite of manipulation tasks, totaling demonstrations, and show that the final DASH () solves of the tasks, outperforming the Allegro baseline (/30). They open-source the CAD models and the teleoperation dataset to enable community use and benchmarking.

Abstract

Modeling and simulating soft robot hands can aid in design iteration for complex and high degree-of-freedom (DoF) morphologies. This can be further supplemented by iterating on the design based on its performance in real world manipulation tasks. However, iterating in the real world requires an approach that allows us to test new designs quickly at low costs. In this paper, we leverage rapid prototyping of the hand using 3D-printing, and utilize teleoperation to evaluate the hand in real world manipulation tasks. Using this method, we design a 3D-printed 16-DoF dexterous anthropomorphic soft hand (DASH) and iteratively improve its design over five iterations. Rapid prototyping techniques such as 3D-printing allow us to directly evaluate the fabricated hand without modeling it in simulation. We show that the design improves over five design iterations through evaluating the hand's performance in 30 real-world teleoperated manipulation tasks. Testing over 900 demonstrations shows that our final version of DASH can solve 19 of the 30 tasks compared to Allegro, a popular rigid hand in the market, which can only solve 7 tasks. We open-source our CAD models as well as the teleoperated dataset for further study.
Paper Structure (30 sections, 2 equations, 7 figures, 3 tables)

This paper contains 30 sections, 2 equations, 7 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Experiment Setup
  4. Robot Hand design
  5. Finger Joints
  6. Tendons
  7. Fabrication using 3D-printing
  8. Hand Evaluation using Teleoperation
  9. Learning Kinematics
  10. Teleoperation System
  11. Manipulation Tasks
  12. DASH Iterative Design Studies
  13. Iteration v1
  14. Design & Fabrication
  15. Evaluation
  16. ...and 15 more sections

Figures (7)

  • Figure 1: Manipulation task performance over five iterations of DASH designed through rapid prototyping and real-world evaluation on tasks alongside task performance of our baseline hand Allegro.
  • Figure 2: (a) Our soft robotic hand design process involving rapid prototyping and real-world evaluation (b) CAD models and differences across DASH iterations v1 through v5, as explained in Section \ref{['sec:DASHdesignandexpts']}.
  • Figure 3: (a) Assembly of DASH-v3 (top to bottom) including fingers, palm, top plate, motors, bottom plate, xArm6 mount. (b) Calibration procedure to map motor angles to finger joint positions, where tendon 0 actuates MCP side-to-side, tendon 1 actuates MCP forward folding motion, and tendon 2 curls the finger controlling both PIP and DIP joints.
  • Figure 4: Manus Meta Quantum Metagloves used for tracking the hand for teleoperating the robot arm and DASH.
  • Figure 5: Subset of tasks with different performance success across v1 to v5 on specific tasks used to inform design iteration. The top row of inset images shows representative tasks of successful tasks for each hand.
  • ...and 2 more figures