CMG-Net: An End-to-End Contact-Based Multi-Finger Dexterous Grasping Network

TL;DR

This paper introduces CMG-Net, an end-to-end neural network that predicts multi-finger grasp poses using a contact-based representation, enabling efficient grasping of unknown objects in cluttered environments with high accuracy.

Contribution

The paper proposes a novel contact-based grasping representation and an end-to-end network, CMG-Net, that improves grasp prediction efficiency and accuracy over previous methods.

Findings

Outperforms state-of-the-art methods for three-finger robotic hands.

Synthetic training data generalizes well to real robot grasping.

Contact-based representation reduces prediction complexity.

Abstract

In this paper, we propose a novel representation for grasping using contacts between multi-finger robotic hands and objects to be manipulated. This representation significantly reduces the prediction dimensions and accelerates the learning process. We present an effective end-to-end network, CMG-Net, for grasping unknown objects in a cluttered environment by efficiently predicting multi-finger grasp poses and hand configurations from a single-shot point cloud. Moreover, we create a synthetic grasp dataset that consists of five thousand cluttered scenes, 80 object categories, and 20 million annotations. We perform a comprehensive empirical study and demonstrate the effectiveness of our grasping representation and CMG-Net. Our work significantly outperforms the state-of-the-art for three-finger robotic hands. We also demonstrate that the model trained using synthetic data performs very well for real robots.