Efficient and Accurate Candidate Generation for Grasp Pose Detection in SE(3)

TL;DR

This paper introduces a novel method for generating grasp candidates in 3D space that improves efficiency and accuracy over existing baselines, advancing robotic grasp detection in unstructured environments.

Contribution

A new grasp candidate generation approach for SE(3) that outperforms existing 3D grasp detection methods in efficiency and accuracy.

Findings

Significantly better performance than major 3D grasp detection baselines.

Efficient identification of high-quality grasp candidates.

Improved grasp detection in unstructured environments.

Abstract

Grasp detection of novel objects in unstructured environments is a key capability in robotic manipulation. For 2D grasp detection problems where grasps are assumed to lie in the plane, it is common to design a fully convolutional neural network that predicts grasps over an entire image in one step. However, this is not possible for grasp pose detection where grasp poses are assumed to exist in SE(3). In this case, it is common to approach the problem in two steps: grasp candidate generation and candidate classification. Since grasp candidate classification is typically expensive, the problem becomes one of efficiently identifying high quality candidate grasps. This paper proposes a new grasp candidate generation method that significantly outperforms major 3D grasp detection baselines. Supplementary material is available at https://atenpas.github.io/psn/.