Grasp detection research focuses at the moment on finnding neural networks that given a RGB-D image or point cloud, yield a parametric grasp description that can be used to firmly grip target objects. There is a need for these models to be small and efficient, such that they can be used in embedded hardware. Furthermore these models tend to only work for top-down views, which highly restrict the ways objects can be grasped. In this work, we focus on improving an existing shallow network, GG-CNN, and propose a new orthographic pipeline to enable the use of these models independently of the orientation of the camera.

After clonning this repository run git submodule init && git submodule update to download the ggcnn dependency.

Next install the dependencies and setup a virtualenv using make {cpu|gpu} depending on whether you have a dedicated graphics cardcompatible with tensorflow or not.

This codebase was developed and tested on python2.7

Available at link

Please adequately refer to the papers any time this code is being used. If you do publish a paper where MORE helped your research, we encourage you to cite the following paper in your publications:

@article{munozextending,
  title={Extending GG-CNN through Automated Model Space Exploration using Knowledge Transfer},
  author={Mu{\~n}oz, Mario R{\'\i}os and Schomaker, Lambert and Kasaei, S Hamidreza}
}

Mario Ríos Muñoz and Hamidreza Kasaei

Work done while at RUG.