Learning-Based Underwater Autonomous Grasping via 3D Point Cloud

Cong Wang; Qifeng Zhang; Shuo Li; Xiaohui Wang; David Lane; Yvan Petillot; Sen Wang

doi:10.23919/OCEANS44145.2021.9706035

Learning-Based Underwater Autonomous Grasping via 3D Point Cloud

Cong Wang, Qifeng Zhang, Shuo Li, Xiaohui Wang, David Lane, Yvan Petillot, Sen Wang

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

2 Citations (Scopus)

Abstract

Underwater autonomous grasping is a challenging task for robotic research. In this paper, we propose a learning-based underwater grasping method using 3D point cloud generated from an underwater stereo camera. First, we use Pinax-model for accurate refraction correction of a stereo camera in a flat-pane housing. Second, dense point cloud of the target is generated using the calibrated stereo images. An improved Grasp Pose Detection (GPD) method is then developed to generate the candidate grasping poses and select the best one based on kinematic constraints. Finally, an optimal trajectory is planned to finish the grasping task. Experiments in a water tank have proved the effectiveness of our method.

Original language	English
Title of host publication	OCEANS 2021
Subtitle of host publication	San Diego - Porto
Publisher	IEEE
ISBN (Electronic)	9780692935590
DOIs	https://doi.org/10.23919/OCEANS44145.2021.9706035
Publication status	Published - 15 Feb 2022
Event	OCEANS 2021: San Diego - Porto - San Diego, United States Duration: 20 Sept 2021 → 23 Sept 2021

Conference

Conference	OCEANS 2021: San Diego - Porto
Country/Territory	United States
City	San Diego
Period	20/09/21 → 23/09/21

Keywords

3D point cloud
Grasp Pose Detection (GPD)
Stereo camera
Underwater grasping

ASJC Scopus subject areas

Oceanography

Access to Document

10.23919/OCEANS44145.2021.9706035

Cite this

@inproceedings{9438ae9c695645b6bf998ad95fb875fd,

title = "Learning-Based Underwater Autonomous Grasping via 3D Point Cloud",

abstract = "Underwater autonomous grasping is a challenging task for robotic research. In this paper, we propose a learning-based underwater grasping method using 3D point cloud generated from an underwater stereo camera. First, we use Pinax-model for accurate refraction correction of a stereo camera in a flat-pane housing. Second, dense point cloud of the target is generated using the calibrated stereo images. An improved Grasp Pose Detection (GPD) method is then developed to generate the candidate grasping poses and select the best one based on kinematic constraints. Finally, an optimal trajectory is planned to finish the grasping task. Experiments in a water tank have proved the effectiveness of our method. ",

keywords = "3D point cloud, Grasp Pose Detection (GPD), Stereo camera, Underwater grasping",

author = "Cong Wang and Qifeng Zhang and Shuo Li and Xiaohui Wang and David Lane and Yvan Petillot and Sen Wang",

note = "Funding Information: ACKNOWLEDGMENT This work was supported by the National Key Research and Development Program of China under grant number 2016YFC0300401. The authors would like to thank Joshua Roe and Leonard McLean for their technical support on data collection and trials. Publisher Copyright: {\textcopyright} 2021 MTS.; OCEANS 2021: San Diego - Porto ; Conference date: 20-09-2021 Through 23-09-2021",

year = "2022",

month = feb,

day = "15",

doi = "10.23919/OCEANS44145.2021.9706035",

language = "English",

booktitle = "OCEANS 2021",

publisher = "IEEE",

address = "United States",

}

TY - GEN

T1 - Learning-Based Underwater Autonomous Grasping via 3D Point Cloud

AU - Wang, Cong

AU - Zhang, Qifeng

AU - Li, Shuo

AU - Wang, Xiaohui

AU - Lane, David

AU - Petillot, Yvan

AU - Wang, Sen

N1 - Funding Information: ACKNOWLEDGMENT This work was supported by the National Key Research and Development Program of China under grant number 2016YFC0300401. The authors would like to thank Joshua Roe and Leonard McLean for their technical support on data collection and trials. Publisher Copyright: © 2021 MTS.

PY - 2022/2/15

Y1 - 2022/2/15

N2 - Underwater autonomous grasping is a challenging task for robotic research. In this paper, we propose a learning-based underwater grasping method using 3D point cloud generated from an underwater stereo camera. First, we use Pinax-model for accurate refraction correction of a stereo camera in a flat-pane housing. Second, dense point cloud of the target is generated using the calibrated stereo images. An improved Grasp Pose Detection (GPD) method is then developed to generate the candidate grasping poses and select the best one based on kinematic constraints. Finally, an optimal trajectory is planned to finish the grasping task. Experiments in a water tank have proved the effectiveness of our method.

AB - Underwater autonomous grasping is a challenging task for robotic research. In this paper, we propose a learning-based underwater grasping method using 3D point cloud generated from an underwater stereo camera. First, we use Pinax-model for accurate refraction correction of a stereo camera in a flat-pane housing. Second, dense point cloud of the target is generated using the calibrated stereo images. An improved Grasp Pose Detection (GPD) method is then developed to generate the candidate grasping poses and select the best one based on kinematic constraints. Finally, an optimal trajectory is planned to finish the grasping task. Experiments in a water tank have proved the effectiveness of our method.

KW - 3D point cloud

KW - Grasp Pose Detection (GPD)

KW - Stereo camera

KW - Underwater grasping

UR - http://www.scopus.com/inward/record.url?scp=85125903221&partnerID=8YFLogxK

U2 - 10.23919/OCEANS44145.2021.9706035

DO - 10.23919/OCEANS44145.2021.9706035

M3 - Conference contribution

AN - SCOPUS:85125903221

BT - OCEANS 2021

PB - IEEE

T2 - OCEANS 2021: San Diego - Porto

Y2 - 20 September 2021 through 23 September 2021

ER -

Learning-Based Underwater Autonomous Grasping via 3D Point Cloud

Abstract

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this