A generalizable artificial intelligence tool for identification and correction of self-supporting structures in additive manufacturing processes

Johnson, Marshall V.; Garanger, Kévin; Hardin, James O.; Berrigan, John D.; Féron, Éric; Kalidindi, Surya R.

doi:10.1016/j.addma.2021.102191

Cited by 12 publications

(11 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The classifier and corrections were built into a closed loop system that worked with (I) DIW and (J) FFF. All images adapted from Johnson et al 257 and materials 257 to prevent costly retraining of ML models for every new process or material, (2) extracting "insights" into the systems behavior that are beyond a human expert like unexpected optimal solutions. In both cases, large quantities of high-quality data are necessary to construct the models without significant investment of human effort into manually extracting the most relevant information through feature engineering.…”

Section: Artificial Intelligence and Machine Learningmentioning

confidence: 99%

Future directions in ceramic additive manufacturing: Fiber reinforcements and artificial intelligence

Rueschhoff,

Baldwin,

Hardin

et al. 2023

J Am Ceram Soc.

Self Cite

View full text Add to dashboard Cite

Research in the field of ceramic additive manufacturing (AM) has been rapidly accelerating, resulting in hundreds of publications and review articles in recent years. While strides have been made in forming near‐net and complex‐shaped ceramic components, challenges remain that inhibit more widespread implementation. In this perspective, we provide a meta‐analysis of recent review articles and highlight a deficiency in two areas of promising future directions to address remaining challenges. The first is incorporation of fiber reinforcements in printed parts to overcome the challenges of poor mechanical performance of monolithic ceramics. Recent work in the area has shown promise incorporating discrete fiber reinforcements as an easier use case given existing equipment limitations, but continuous fibers are needed to reach full toughness potential. Here, we overview some options and future directions bases on success in polymer composites. Second, artificial intelligence (AI) approaches, including machine learning (ML), are suggested in order to accelerate feedstock development and process optimization. While there has been very limited work to date in utilizing AI/ML techniques for ceramic AM, again inspiration and lessons learned are drawn from the polymer AM community.

show abstract

Section: Artificial Intelligence and Machine Learningmentioning

confidence: 99%

Future directions in ceramic additive manufacturing: Fiber reinforcements and artificial intelligence

Rueschhoff,

Baldwin,

Hardin

et al. 2023

J Am Ceram Soc.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Moreover, most existing approaches can only detect a single error modality: poor flow rate 49 , interlayer defects 50 , warp deformation 53 , and large top surface defects 52 , 54 . Often existing methods also require an object to already have been printed successfully to provide comparisons for the error detection 51 , 54 , 55 . This may be especially limiting for custom parts.…”

Section: Introductionmentioning

confidence: 99%

“…For error detection to reach its full potential in reducing 3D printing waste and improving sustainability, cost, and reliability, it must be coupled with error correction. There has been work in detecting and correcting some kinds of errors between subsequent prints of the same object 51 , 55 . However, many prints of that object are required to build the dataset, enabling error correction in that object.…”

Section: Introductionmentioning

confidence: 99%

“…Machine learning and particularly deep learning techniques have achieved state-of-the-art performance across many applications, including vision 47 , by expressing complex representations in terms of other simpler representations 48 . This has led to several exciting recent demonstrations of machine learning in extrusion AM error detection [49][50][51][52][53][54][55][56] . Existing work has only demonstrated error detection in a single part, though, and thus the effectiveness of existing techniques for other parts, particularly parts not seen in the training data, is unknown.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Generalisable 3D printing error detection and correction via multi-head neural networks

Brion

Pattinson

2022

Nat Commun

View full text Add to dashboard Cite

Material extrusion is the most widespread additive manufacturing method but its application in end-use products is limited by vulnerability to errors. Humans can detect errors but cannot provide continuous monitoring or real-time correction. Existing automated approaches are not generalisable across different parts, materials, and printing systems. We train a multi-head neural network using images automatically labelled by deviation from optimal printing parameters. The automation of data acquisition and labelling allows the generation of a large and varied extrusion 3D printing dataset, containing 1.2 million images from 192 different parts labelled with printing parameters. The thus trained neural network, alongside a control loop, enables real-time detection and rapid correction of diverse errors that is effective across many different 2D and 3D geometries, materials, printers, toolpaths, and even extrusion methods. We additionally create visualisations of the network’s predictions to shed light on how it makes decisions.

show abstract

“…Recently, deep learning vision-based approaches have been used to autonomously apply corrections offline after printing for a range of errors. [31][32][33] These methods are very useful for errors that build over time as the material cools, such as cracking and warping, and nonrecoverable failures, for example, poor bridging. Although a significant step forward, these approaches still fail to catch many internal error modalities thanks to externally mounted cameras and result in wasted prints as corrections are applied after the fact.…”

Section: Introductionmentioning

confidence: 99%

Quantitative and Real‐Time Control of 3D Printing Material Flow Through Deep Learning

Brion

Pattinson

2022

Advanced Intelligent Systems

View full text Add to dashboard Cite

3D printing could revolutionize manufacturing through local and on‐demand production while enabling uniquely complex and custom products. However, 3D printing's propensity for production errors prevents autonomous operation and the quality assurance necessary to realize this vision. Human operators cannot continuously monitor or correct errors in real time, while automated approaches predominantly only detect errors. New methodologies correct parameters either offline or with slow response times and poor prediction granularity, limiting their utility. A commonly available 3D printing process metadata is harnessed, alongside the video of the printing process, to build a unique image dataset. Regression models are trained to precisely predict how printing material flow should be altered to correct errors and this should be used to build a fast control loop capable of 3D printing parameter discovery and few‐shot correction. Demonstrations show that the system can learn optimal parameters for unseen complex materials, and achieve rapid error correction on new parts. Similar metadata exists in many manufacturing processes and this approach could enable the adoption of fast data‐driven control systems more widely in manufacturing.

show abstract

A generalizable artificial intelligence tool for identification and correction of self-supporting structures in additive manufacturing processes

Cited by 12 publications

References 38 publications

Future directions in ceramic additive manufacturing: Fiber reinforcements and artificial intelligence

Future directions in ceramic additive manufacturing: Fiber reinforcements and artificial intelligence

Generalisable 3D printing error detection and correction via multi-head neural networks

Quantitative and Real‐Time Control of 3D Printing Material Flow Through Deep Learning

Contact Info

Product

Resources

About