MPGNet: Learning Move-Push-Grasping Synergy for Target-Oriented Grasping in Occluded Scenes

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Li, Dayou, Zhao, Chenkun, Yang, Shuo, Song, Ran, Li, Xiaolei, Zhang, Wei
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909291614568448
author Li, Dayou
Zhao, Chenkun
Yang, Shuo
Song, Ran
Li, Xiaolei
Zhang, Wei
author_facet Li, Dayou
Zhao, Chenkun
Yang, Shuo
Song, Ran
Li, Xiaolei
Zhang, Wei
contents This paper focuses on target-oriented grasping in occluded scenes, where the target object is specified by a binary mask and the goal is to grasp the target object with as few robotic manipulations as possible. Most existing methods rely on a push-grasping synergy to complete this task. To deliver a more powerful target-oriented grasping pipeline, we present MPGNet, a three-branch network for learning a synergy between moving, pushing, and grasping actions. We also propose a multi-stage training strategy to train the MPGNet which contains three policy networks corresponding to the three actions. The effectiveness of our method is demonstrated via both simulated and real-world experiments.
format Preprint
id arxiv_https___arxiv_org_abs_2408_10525
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle MPGNet: Learning Move-Push-Grasping Synergy for Target-Oriented Grasping in Occluded Scenes
Li, Dayou
Zhao, Chenkun
Yang, Shuo
Song, Ran
Li, Xiaolei
Zhang, Wei
Robotics
This paper focuses on target-oriented grasping in occluded scenes, where the target object is specified by a binary mask and the goal is to grasp the target object with as few robotic manipulations as possible. Most existing methods rely on a push-grasping synergy to complete this task. To deliver a more powerful target-oriented grasping pipeline, we present MPGNet, a three-branch network for learning a synergy between moving, pushing, and grasping actions. We also propose a multi-stage training strategy to train the MPGNet which contains three policy networks corresponding to the three actions. The effectiveness of our method is demonstrated via both simulated and real-world experiments.
title MPGNet: Learning Move-Push-Grasping Synergy for Target-Oriented Grasping in Occluded Scenes
topic Robotics
url https://arxiv.org/abs/2408.10525