NYU-robot-learning / DIME-Controllers

Repository containing the base controller for DIME

DIME Controller

This repository contains the installer and python wrapper for controlling an Allegro Hand attached to a Kinova JACO Arm. It is part of the official implementation of DIME. The base repo containing all the information can be found at DIME - Models.

This repository contains symlinks connecting to the following repositories:

1. Allegro Hand Controller - DIME
2. Kinova Arm Controller - DIME

Setup

We would advice you to install the Kinova SDK and then proceed with running the setup.sh file. The setup.sh installs all the required dependencies.

Launching the controllers

You have to launch the roslaunch files for both the Allegro Hand and Kinova Arm separately using the following commands:

• For launching the Kinova Arm controller (this includes the gravity compensation module for the Allegro Hand):

  roslaunch kinova_arm kinova_robot.launch
  

• For launching the Allegro Hand controller with/without the visualizer:

  roslaunch allegro_hand allegro_hand.launch VISUALIZE:=<true|false>
  

Using the python wrapper

You can import the python wrapper directly since we include this directory in PYTHONPATH (the setup.sh file takes care of this). You can run the following commands to move the robot:

from move_dexarm import DexArmControl

arm_controller = DexArmControl()
move_robot(allegro_angles, kinova_angles)

Citation

If you use this repo in your research, please consider citing the paper as follows:

@article{arunachalam2022dime,
  title={Dexterous Imitation Made Easy: A Learning-Based Framework for Efficient Dexterous Manipulation},
  author={Sridhar Pandian Arunachalam and Sneha Silwal and Ben Evans and Lerrel Pinto},
  journal={arXiv preprint arXiv:2203.13251},
  year={2022}
}