IRVLUTD / gqcnn

[DexNet] Python module for GQ-CNN training and deployment with ROS integration.

Note: Python 2.x support has officially been dropped.

Berkeley AUTOLAB's GQCNN Package

Package Overview

The gqcnn Python package is for training and analysis of Grasp Quality Convolutional Neural Networks (GQ-CNNs). It is part of the ongoing Dexterity-Network (Dex-Net) project created and maintained by the AUTOLAB at UC Berkeley.

Installation and Usage

Please see the docs for installation and usage instructions.

Citation

If you use any part of this code in a publication, please cite the appropriate Dex-Net publication.

SceneReplica experiments

Please see ros_nodes/scenereplica_ros_experiment.py for running a publisher/subscriber loop.

• Run the DexNet grasp planning service as described in the documentation
• Run the scenereplica_ros_experiment.py script.
  • Assumes that a top-down depth and mask image is being published
  • Plans the graps and publishes the (grasp center pixel, angle) as Point msg: (u, v, theta)
  • Its a crude workaround to publish the grasp, feel free to implement a custom msg for this purpose and improve the code!