|▲ AiV / CEO Minsoo Sung|
The fight of the century between AlphaGo and Sedol Lee in March 2016 opened the new era of artificial intelligence. Broadcast in Korean, English, Japanese and Chinese, AlphaGo beat Lee by 4 to 1. In the year 2022, AI has permeated into our everyday life loaded in many cutting edge gadgets and equipment, whether we are aware of it or not, from which we enjoy benefits of convenience at home or business. At home we control some of the house operation such as adjusting temperature, cleaning and asking today’s weather with automated home appliances. At work we no longer need to inspect defects of a product with our own eyes as, for example, a vision inspection solution does it instead.
AiV is a South Korean deep learning computer vision company. Boasting world class deep learning vision technology and mechatronics convergence, AiV is creating a fresh ripple in the manufacturing industry in South Korea. The company especially has accumulated technique needed in defects inspection of products which normally is the last phase of manufacturing. Experts at AiV have professional knowledge and years of experience in the field. Thanks to this excellent human resource, the company was able to develop its own deep learning computer vision technology within 2 years of its startup. Taking the momentum, they pushed their ability a step ahead and succeeded development of a software platform that facilitates introduction of new AI nerve networks.
Follows are a general procedure of manufacturing a product: original material - processing - assembly - inspection. The inspection normally divides into two areas: measurement and surface. According to Minsoo Sung, CEO of AiV, most manufacturers use naked eye or depending on MV (Traditional Machine Vision). The former lacks accuracy but costs more as it needs sufficient labors. The latter also has a problem to inspect atypical defects of products accurately. Both requires defect classification as the results and accuracy varies person by person and case by case otherwise. AiV’s deep learning computer vision solutions in this respect is but good news.
Epoch Breaking Industrial Inspection Solutions
Algorithm performance, optical design technology and customized hardware are three core technologies of advanced industrial inspection solutions. Lacking one of the three will harvest pretty much nothing. Introducing the mechatronics-based machine vision inspection equipment and AI nerve-networks, AiV has written a new chapter in advanced industrial inspection solutions.
“Our innovative mechatronics-based machine vision inspection equipment works itself to find defects and to learn and accumulate data for future reference. No matter how complicated the inspection is, it makes the results standardized and automated” says Sung.
How Does It Work?
Follows are how this cutting edge equipment works: modeling of the original product and its defected version - gathering information from the imaging - building rule-based algorithm machine vision image data on which sanitized similar defect image data and augmented data can be added according to its need. The data accumulated goes through generalization and objectification while the labeled original and defected product image data creates deep learning models. And the self-learning goes on and on towards more enhanced solutions. The accumulated algorithm converges with its own optical technology and mechatronics elements to be customized for each need and want of the manufacturer. The optical technology not only analyzes 2 dimensional images but gives 3 dimensional construction for information, appearance, shapes, etc. It also comes with developer mode algorithm, deep learning-based pelletizing and de-pelletizing that can help maximize advanced automated logistics system. AiV inspection solutions satisfy the standards used in car parts, electronics, precision optical products that require micron-level precision and accuracy.
100% Defect Inspection Rate
AiV has successfully carried out 24 of PoC (Proof of Concept) of which 22 have been turned into mass-produce projects. It is encouraging that Central introduced AiV’s solution for ball joint inspection early this year and Smakee EV also for EV battery pack inspection. The ball joint of Central is supplied to Hyundai Motor, GM, Fords, Mercedes Benz and Porsche.
“We must be proud to say that our solutions detected defects of ball joints by 100%. It not only contributed to better performance of a complete product but increased efficiency of manufacturing and quality while reducing costs. The end-plate inspection at Samkee EV also showed 99.9% defect inspection rate” says Sung.
AiV listed its name in the ‘Unsupervised Anomaly Detection’ TOP based on the 2021 2Q Industrial Domain Data Set of MVTec. The company now plans to provide its core services ‘Vision Inspection SW’ as in web-services and experts at the company are ambitious to introduce more advanced solutions that can find and remove defect-cause over defect inspection. <PowerKorea>
신태섭 기자 firstname.lastname@example.org