In-line 3D print failure detection using computer vision

Rasmus Ahrenkiel Lyngby; Jakob Wilm; Eyþór Rúnar Eiríksson; Jannik Boll Nielsen; Janus Nørtoft Jensen; Henrik Aanæs; David Bue Pedersen

In-line 3D print failure detection using computer vision

Rasmus Ahrenkiel Lyngby, Jakob Wilm, Eyþór Rúnar Eiríksson, Jannik Boll Nielsen, Janus Nørtoft Jensen, Henrik Aanæs, David Bue Pedersen

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

1 Downloads (Pure)

Abstract

Here we present our findings on a novel real-time vision system that allows for automatic detection of failure conditions that are considered outside of nominal operation. These failure modes include warping, build plate delamination and extrusion failure. Our system consists of a calibrated camera whose position and orientation is known in the machine coordinate system. We simulate what the object under print should look like for any given moment in time. This is compared to a segmentation of the current print, and statistical detection of significant deviation. We demonstrate that this methodology precisely and unambiguously detects the time point of print failure.

Original language	English
Title of host publication	Dimensional Accuracy and Surface Finish in Additive Manufacturing
Number of pages	4
Publication date	2017
Publication status	Published - 2017
Event	euspen and ASPE Special Interest Group Meeting: Additive Manufacturing: Dimensional Accuracy and Surface Finish from Additive Manufacturing - Katholieke Universiteit Leuven, Leuven, Belgium Duration: 10 Oct 2017 → 11 Oct 2017

Conference

Conference	euspen and ASPE Special Interest Group Meeting: Additive Manufacturing
Location	Katholieke Universiteit Leuven
Country/Territory	Belgium
City	Leuven
Period	10/10/2017 → 11/10/2017

Keywords

Failure detection
Computer vision
Fused deposition modeling (FDM)

Cite this

@inproceedings{37c67c08523e43d7a93ac48e81b81667,

title = "In-line 3D print failure detection using computer vision",

abstract = "Here we present our findings on a novel real-time vision system that allows for automatic detection of failure conditions that are considered outside of nominal operation. These failure modes include warping, build plate delamination and extrusion failure. Our system consists of a calibrated camera whose position and orientation is known in the machine coordinate system. We simulate what the object under print should look like for any given moment in time. This is compared to a segmentation of the current print, and statistical detection of significant deviation. We demonstrate that this methodology precisely and unambiguously detects the time point of print failure.",

keywords = "Failure detection, Computer vision, Fused deposition modeling (FDM)",

author = "Lyngby, {Rasmus Ahrenkiel} and Jakob Wilm and Eir{\'i}ksson, {Ey{\th}{\'o}r R{\'u}nar} and Nielsen, {Jannik Boll} and Jensen, {Janus N{\o}rtoft} and Henrik Aan{\ae}s and Pedersen, {David Bue}",

year = "2017",

language = "English",

booktitle = "Dimensional Accuracy and Surface Finish in Additive Manufacturing",

note = "euspen and ASPE Special Interest Group Meeting: Additive Manufacturing : Dimensional Accuracy and Surface Finish from Additive Manufacturing ; Conference date: 10-10-2017 Through 11-10-2017",

}

In-line 3D print failure detection using computer vision. / Lyngby, Rasmus Ahrenkiel; Wilm, Jakob; Eiríksson, Eyþór Rúnar; Nielsen, Jannik Boll; Jensen, Janus Nørtoft; Aanæs, Henrik; Pedersen, David Bue.

Dimensional Accuracy and Surface Finish in Additive Manufacturing. 2017.

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

TY - GEN

T1 - In-line 3D print failure detection using computer vision

AU - Lyngby, Rasmus Ahrenkiel

AU - Wilm, Jakob

AU - Eiríksson, Eyþór Rúnar

AU - Nielsen, Jannik Boll

AU - Jensen, Janus Nørtoft

AU - Aanæs, Henrik

AU - Pedersen, David Bue

PY - 2017

Y1 - 2017

N2 - Here we present our findings on a novel real-time vision system that allows for automatic detection of failure conditions that are considered outside of nominal operation. These failure modes include warping, build plate delamination and extrusion failure. Our system consists of a calibrated camera whose position and orientation is known in the machine coordinate system. We simulate what the object under print should look like for any given moment in time. This is compared to a segmentation of the current print, and statistical detection of significant deviation. We demonstrate that this methodology precisely and unambiguously detects the time point of print failure.

AB - Here we present our findings on a novel real-time vision system that allows for automatic detection of failure conditions that are considered outside of nominal operation. These failure modes include warping, build plate delamination and extrusion failure. Our system consists of a calibrated camera whose position and orientation is known in the machine coordinate system. We simulate what the object under print should look like for any given moment in time. This is compared to a segmentation of the current print, and statistical detection of significant deviation. We demonstrate that this methodology precisely and unambiguously detects the time point of print failure.

KW - Failure detection

KW - Computer vision

KW - Fused deposition modeling (FDM)

M3 - Article in proceedings

BT - Dimensional Accuracy and Surface Finish in Additive Manufacturing

T2 - euspen and ASPE Special Interest Group Meeting: Additive Manufacturing

Y2 - 10 October 2017 through 11 October 2017

ER -

In-line 3D print failure detection using computer vision

Abstract

Conference

Keywords

Fingerprint

Cite this