3DSGrasp: 3D Shape-Completion for Robotic Grasp

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Mohammadi, Seyed S., Duarte, Nuno F., Dimou, Dimitris, Wang, Yiming, Taiana, Matteo, Morerio, Pietro, Dehban, Atabak, Moreno, Plinio, Bernardino, Alexandre, Del Bue, Alessio, Santos-Victor, Jose
Format: Preprint
Veröffentlicht: 2023
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918244830412800
author Mohammadi, Seyed S.
Duarte, Nuno F.
Dimou, Dimitris
Wang, Yiming
Taiana, Matteo
Morerio, Pietro
Dehban, Atabak
Moreno, Plinio
Bernardino, Alexandre
Del Bue, Alessio
Santos-Victor, Jose
author_facet Mohammadi, Seyed S.
Duarte, Nuno F.
Dimou, Dimitris
Wang, Yiming
Taiana, Matteo
Morerio, Pietro
Dehban, Atabak
Moreno, Plinio
Bernardino, Alexandre
Del Bue, Alessio
Santos-Victor, Jose
contents Real-world robotic grasping can be done robustly if a complete 3D Point Cloud Data (PCD) of an object is available. However, in practice, PCDs are often incomplete when objects are viewed from few and sparse viewpoints before the grasping action, leading to the generation of wrong or inaccurate grasp poses. We propose a novel grasping strategy, named 3DSGrasp, that predicts the missing geometry from the partial PCD to produce reliable grasp poses. Our proposed PCD completion network is a Transformer-based encoder-decoder network with an Offset-Attention layer. Our network is inherently invariant to the object pose and point's permutation, which generates PCDs that are geometrically consistent and completed properly. Experiments on a wide range of partial PCD show that 3DSGrasp outperforms the best state-of-the-art method on PCD completion tasks and largely improves the grasping success rate in real-world scenarios. The code and dataset will be made available upon acceptance.
format Preprint
id arxiv_https___arxiv_org_abs_2301_00866
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle 3DSGrasp: 3D Shape-Completion for Robotic Grasp
Mohammadi, Seyed S.
Duarte, Nuno F.
Dimou, Dimitris
Wang, Yiming
Taiana, Matteo
Morerio, Pietro
Dehban, Atabak
Moreno, Plinio
Bernardino, Alexandre
Del Bue, Alessio
Santos-Victor, Jose
Robotics
Artificial Intelligence
Real-world robotic grasping can be done robustly if a complete 3D Point Cloud Data (PCD) of an object is available. However, in practice, PCDs are often incomplete when objects are viewed from few and sparse viewpoints before the grasping action, leading to the generation of wrong or inaccurate grasp poses. We propose a novel grasping strategy, named 3DSGrasp, that predicts the missing geometry from the partial PCD to produce reliable grasp poses. Our proposed PCD completion network is a Transformer-based encoder-decoder network with an Offset-Attention layer. Our network is inherently invariant to the object pose and point's permutation, which generates PCDs that are geometrically consistent and completed properly. Experiments on a wide range of partial PCD show that 3DSGrasp outperforms the best state-of-the-art method on PCD completion tasks and largely improves the grasping success rate in real-world scenarios. The code and dataset will be made available upon acceptance.
title 3DSGrasp: 3D Shape-Completion for Robotic Grasp
topic Robotics
Artificial Intelligence
url https://arxiv.org/abs/2301.00866