DELTAHANDS is a synergistic dexterous hand framework that utilizes soft Delta robots. Delta hands are low-cost, easy to build and control, and highly configurable regarding design parameters. We characterize the Delta robot's kinematics accuracy, force profile, and workspace range which can guide the design and configuration of our robotic hands. We present two different actuation synergies to reduce the control difficulty and energy consumption while keeping the hands' dexterity. And we show the capability of using the hand framework to further explore various synergistic configurations. We demonstrate diverse object grasping with hand prototypes and quantitative evaluations in the simulator. Furthermore, we teleoperate a Delta hand with a Franka robot arm to complete three dexterous manipulation tasks: cloth folding, cap opening, and cable arrangement.

The structure of Delta hands is modularized to fingertips, fingers, hand frame, and actuators from top to down. This greatly simplifies hand manufacturing and re-configurations by simply modifying or replacing certain components. 

We provide both 

1) the Arduino codes including low-level PID control and communication between the actuators and the micro-controller [link]

2) the Python codes including Delta hands' kinematics and communication between the micro-controller and control pc [link]

We provide

1) A URDF generator to generate Delta hand configurations [link]

2) A simulation environment based on PyBullet for manipulation [link]

We demonstrate grasping with both simulated and real Delta hands on various objects.

Maximum before-slipping pulling forces on objects with different size, shape, and surface frictions. 

We demonstrate finger-gaiting with a 9-actuator Delta hand on two objects.

We demonstrate teleoperation on three manipulation tasks: cloth folding, cap opening, and cable arrangement.

@article{si2023deltahands,

  title={DELTAHANDS: A Synergistic Dexterous Hand Framework Based on Delta Robots},

  author={Si, Zilin and Zhang, Kevin and Kroemer, Oliver and Temel, F Zeynep},

  journal={arXiv preprint arXiv:2310.05266},

  year={2023}

}