Addressing the Sim2Real Gap in Robotic 3-D Object Classification

Weibel, JB; Patten, T; Vincze, M

Addressing the Sim2Real Gap in Robotic 3-D Object Classification

Weibel, JB Patten, T

Vincze, M

Permalink

Publisher:: IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
Publication Type:: Journal Article
Citation:: IEEE Robotics and Automation Letters, 2020, 5, (2), pp. 407-413
Issue Date:: 2020-04-01

Closed Access

	Filename	Description	Size
	08931599.pdf	Published version	1.93 MB	Adobe PDF	View/Open

Copyright Clearance Process

Recently Added
In Progress
Closed Access

This item is closed access and not available.

Full metadata record

Field	Value	Language
dc.contributor.author	Weibel, JB
dc.contributor.author	Patten, T https://orcid.org/0000-0003-1139-9451
dc.contributor.author	Vincze, M
dc.date.accessioned	2021-02-12T20:45:04Z
dc.date.available	2021-02-12T20:45:04Z
dc.date.issued	2020-04-01
dc.identifier.citation	IEEE Robotics and Automation Letters, 2020, 5, (2), pp. 407-413
dc.identifier.issn	2377-3766
dc.identifier.issn	2377-3766
dc.identifier.uri	http://hdl.handle.net/10453/146051
dc.description.abstract	© 2016 IEEE. Object classification with 3D data is an essential component of any scene understanding method. It has gained significant interest in a variety of communities, most notably in robotics and computer graphics. While the advent of deep learning has progressed the field of 3D object classification, most work using this data type are solely evaluated on CAD model datasets. Consequently, current work does not address the discrepancies existing between real and artificial data. In this work, we examine this gap in an indoor service robotic context by specifically addressing the problem of classification when transferring from artificial CAD models to real reconstructed objects. This is performed by training on ModelNet (CAD models) and evaluating on ScanNet (objects extracted from reconstructed rooms). We show that standard methods do not perform well in this task. We thus introduce a method that carefully samples object parts that are reproducible under various transformations and hence robust. Using graph convolution to classify the composed graph of parts, our method improves upon the baseline. Code is publicly available at https://rgit.acin.tuwien.ac.at/jb.weibel/cad2real_object_clf.
dc.language	English
dc.publisher	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
dc.relation.ispartof	IEEE Robotics and Automation Letters
dc.relation.isbasedon	10.1109/LRA.2019.2959497
dc.rights	© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.	en_US
dc.rights	info:eu-repo/semantics/closedAccess
dc.subject	0913 Mechanical Engineering
dc.title	Addressing the Sim2Real Gap in Robotic 3-D Object Classification
dc.type	Journal Article
utslib.citation.volume	5
utslib.for	0913 Mechanical Engineering
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology/School of Mechanical and Mechatronic Engineering
pubs.organisational-group	/University of Technology Sydney
utslib.copyright.status	closed_access	*
dc.date.updated	2021-02-12T20:44:45Z
pubs.issue	2
pubs.publication-status	Published
pubs.volume	5
utslib.citation.issue	2

Abstract:

© 2016 IEEE. Object classification with 3D data is an essential component of any scene understanding method. It has gained significant interest in a variety of communities, most notably in robotics and computer graphics. While the advent of deep learning has progressed the field of 3D object classification, most work using this data type are solely evaluated on CAD model datasets. Consequently, current work does not address the discrepancies existing between real and artificial data. In this work, we examine this gap in an indoor service robotic context by specifically addressing the problem of classification when transferring from artificial CAD models to real reconstructed objects. This is performed by training on ModelNet (CAD models) and evaluating on ScanNet (objects extracted from reconstructed rooms). We show that standard methods do not perform well in this task. We thus introduce a method that carefully samples object parts that are reproducible under various transformations and hence robust. Using graph convolution to classify the composed graph of parts, our method improves upon the baseline. Code is publicly available at https://rgit.acin.tuwien.ac.at/jb.weibel/cad2real_object_clf.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/146051