Rotograb: Combining Biomimetic Hands with Industrial Grippers using a Rotating Thumb

Arnaud Bersier, Matteo Leonforte, Alessio Vanetta, Sarah Lia Andrea Wotke, Andrea Nappi, Yifan Zhou, Sebastiano Oliani, Alexander M. Kübler, Robert K. Katzschmann

TL;DR

This paper introduces Rotograb, a novel robotic hand that merges the dexterity of biomimetic hands with the efficiency of industrial grippers, featuring a unique rotating thumb. This design significantly expands the hand's workspace and enables sophisticated in-hand manipulation, bridging a critical gap in robotic capabilities.

What Problem Does This Paper Address?

The field of robotics faces a trade-off: industrial grippers are efficient for simple pick-and-place tasks but lack dexterity, while biomimetic robotic hands offer high dexterity but are often complex, fragile, and expensive. This creates a capability gap for applications requiring both precision manipulation and robust grasping. The authors aim to bridge this gap by developing a robotic hand that combines the strengths of both approaches, offering versatility for a wide range of tasks.

Methods

The core innovation of Rotograb is its rotating thumb mechanism. This allows the thumb to switch between different configurations, enabling both dexterous "human-like" grasps and powerful "industrial-like" grasps. The hand also features a novel rolling-contact joint with a cutout designed to simplify finger kinematics and tendon routing, minimizing interference and improving control.

For control, the researchers implemented two approaches:

  • Teleoperation: Using a depth camera (OAK-D Pro) and a hand-tracking model (Mediapipe), the system maps human hand movements to the robotic gripper for intuitive control.
  • Reinforcement Learning (RL): For in-hand object manipulation, specifically rotating a ball, they employed a Proximal Policy Optimization (PPO) algorithm within the Isaac Gym simulator. This allowed for rapid training of dexterous manipulation policies.

Datasets

  • YCB Object and Model Set [14]: This dataset was used for evaluating the grasping capabilities of Rotograb with a variety of objects.

Key Insights and Ideas

  • The Rotating Thumb is a Game-Changer: The ability to reconfigure the thumb's position is not just about aesthetics; it fundamentally alters the hand's grasping capabilities, allowing it to adapt to diverse object shapes and task requirements.
  • Simplified Kinematics Unlock Dexterity: The novel joint design with a cutout effectively simplifies the complex tendon routing, making control more manageable without sacrificing significant robustness. This is a clever engineering solution to a common robotics challenge.
  • Bridging the Sim-to-Real Gap: The use of domain randomization in the RL training process is crucial for transferring policies learned in simulation to the physical robot, a persistent challenge in robotics.
  • Ambidexterity Achieved: The rotating thumb, combined with the hand's overall design, allows for ambidextrous grasping, a valuable trait for robots operating in human-centric environments.

Key Results

  • Expanded Workspace: The rotating thumb significantly increases the reachable workspace compared to fixed-thumb designs.
  • Versatile Grasping: Rotograb successfully grasped a wide range of objects from the YCB dataset, demonstrating proficiency in both precision and power grasps.
  • In-Hand Manipulation: The hand successfully performed in-hand object rotation tasks, including rolling a ball at approximately 5.45 rounds per minute and re-orienting a Rubik's cube at about 5.62 rounds per minute.
  • Cost-Effective: The use of 3D printing and readily available components resulted in a material cost of approximately 1100€ for the core robotic hand.

Limitations and Future Work

  • Limited Finger Adduction/Abduction: The authors acknowledge that adding adduction and abduction to the fingers would further improve the workspace and manipulation capabilities.
  • Mapping Challenges: Mapping human hand movements to the robotic hand's DoFs remains challenging due to differences in shape and size, requiring further research.
  • Robustness and Slenderer Fingers: Future work aims to increase the hand's robustness and develop more slender fingers with improved dexterity.
  • Enhanced Control and Teleoperation: Further improvements in control strategies and teleoperation mapping are planned.
  • Quantitative Manipulation Experiments: The authors intend to conduct more quantitative experiments to analyze and improve the hand's manipulation skills.

Summary updated

Abstract

The development of robotic grippers and hands for automation aims to emulate human dexterity without sacrificing the efficiency of industrial grippers. This study introduces Rotograb, a tendon-actuated robotic hand featuring a novel rotating thumb. The aim is to combine the dexterity of human hands with the efficiency of industrial grippers. The rotating thumb enlarges the workspace and allows in-hand manipulation. A novel joint design minimizes movement interference and simplifies kinematics, using a cutout for tendon routing. We integrate teleoperation, using a depth camera for real-time tracking and autonomous manipulation powered by reinforcement learning with proximal policy optimization. Experimental evaluations demonstrate that Rotograb’s rotating thumb greatly improves both operational versatility and workspace. It can handle various grasping and manipulation tasks with objects from the YCB dataset, with particularly good results when rotating objects within its grasp. Rotograb represents a notable step towards bridging the capability gap between human hands and industrial grippers. The tendon-routing and thumb-rotating mechanisms allow for a new level of control and dexterity. Integrating teleoperation and autonomous learning underscores Rotograb’s adaptability and sophistication, promising substantial advancements in both robotics research and practical applications.
BibTeX
@inproceedings{bersier2024rotograb,
  author = {Arnaud Bersier and Matteo Leonforte and Alessio Vanetta and Sarah Lia Andrea Wotke and Andrea Nappi and Yifan Zhou and Sebastiano Oliani and Alexander M. Kübler and Robert K. Katzschmann},
  title = {{Rotograb: Combining Biomimetic Hands with Industrial Grippers using a Rotating Thumb}},
  booktitle = {IEEE/RSJ International Conference on Intelligent Robots and Systems},
  year = {2024},
  url = {https://openreview.net/forum?id=pI2ZoJXon4},
}
Figure