Customizable 6 Degrees of Freedom Grasping Dataset and an Interactive Training Method for Graph Convolutional Network
This repository contains the code and documentation for our research on Customizable 6 Degrees of Freedom Grasping Dataset and an Interactive Training Method for Graph Convolutional Network.
-
Data generator: Propose an innovative method for generating robotic grasping datasets which simulates real-life robotic gripper actions in a virtual setting instead of relying on manual annotations.
-
End-to-end graph convolution network: Design an end-to-end graph convolution grasping prediction network, which can directly predict the optimal grasping points and orientations from partial point cloud.
-
Interactive training method: Introduce an interactive training method which reforms training style of deep learning models by allowing real-time adjustments to both the training data and model's training hyperparameters.
A video demonstrating our work can be found at the link below:
https://drive.google.com/file/d/1RFnYAczci9f38wJJlxjA2PbEv1gkQ8Db/view?usp=sharing
https://drive.google.com/file/d/1sqJhQqb1Dql0dZZQCjCVdGd6uNoFpVuj/view?usp=sharing
If you need more detailed code, please feel free to contact us at nwh1093412390@sjtu.edu.cn
This project is licensed under the MIT License. See the LICENSE file for more details.