Title: Accelerating In-Process Automatic Inspection Engineering: Removing Inhibitors to Rapid Machine Learning-Based Application Development

Authors: Scott Blake, Amir Atai

DOI: 10.33599/nasampe/s.21.0472

Abstract: Although in-process automatic inspection has made inroads in the fabrication of low-volume flight-critical composite components (<1,000 per year), novel enabling technologies are needed to meet the needs of new high-volume structures. For such applications, an automatic inspection technology must be able to verify (and/or flag for nonconformance) a unique set of features and potential anomalies arising from the particulars of each structure’s design and production process. Therefore, each component requires its own inspection engineering employing: (1) The means and methodology to capture a database of calibrated images that display all the features and anomalies that the automatic inspection system must detect and/or measure; and (2) Application of artificial intelligence/machine learning (AI/ML) to develop the algorithms that analyze the captured images. Critically important to widespread use, this inspection system cannot rely on complex AI/ML coding and optimization for each application, which would be time- and cost-prohibitive. The remedy is a low-code AI/ML workflow engine that will democratize inspection application development. This paper will outline the requirements for automatic inspection of high-volume composite components and describe the enabling technologies of (1) image database generation using a machine vision system and (2) democratized AI/ML-based algorithm development.

References: 1. Rudberg, T., Nielson, J., Henscheid, M., and Cemenska, J., “Improving AFP Cell Performance,” SAE Int. J. Aerosp. 7(2):2014, DOI:10.4271/2014-01-2272. 2. Halbritter, A., and Harper, R., “Big Parts Demand Big Changes to the Fiber Placement Status Quo,” presented at SME Composites Manufacturing, Mesa, AZ, March 2012. 3. Cemenska, J., Rudberg, T., Henscheid, M., et al., “AFP Automated Inspection System Performance and Expectations,” SAE Technical Paper 2017-01-2150, 2017, DOI:10.4271/2017-01-2150. 4. “Artificial Intelligence (AI) vs. Machine Learning vs Deep Learning.” 25 Jan 2019: https://skymind.ai/wiki/ai-vs-machine-learning-vs-deep-learning. 5. Blake, S., “Elements and Mechanisms for Applying Artificial Intelligence to Composites Fabrication,” SAMPE Technical Paper TP19-1475, 2019, DOI:10.33599/nasampe/s.19.1435 6. Garcia-Garcia, A., Orts-Escolano, S. Oprea, S., Villena-Martinez, V., Martinez-Gonzalez, P. and Garcia-Rodriguez, J., “A Survey on Deep Learning Techniques for Image and Video Semantic Segmentation,” Applied Soft Computing 70:2018, DOI: 10.1016/j.asoc.2018.05.018. 7. Scime, L., Siddel, D., Baird, S., and Paquit, V., “Layer-Wise Anomaly Detection and Classification for Powder Bed Additive Manufacturing Processes: A Machine-Agnostic Algorithm for Real-Time Pixel-Wise Semantic Segmentation,” Additive Manufacturing 36 (2020), DOI:10.1016/j.addma.2020.101453. 8. Paquit, Vincent, Interview by Scott Blake and Karen Mason, August 18, 2020.

Conference: SAMPE NEXUS 2021

Publication Date: 2021/06/29

SKU: TP21-0000000472

Pages: 12

Price: FREE