We are based on the local area and look at the worldHomepage - English
Hon Hai Group
America
Europe
Czech Republic
Slovakia
Press Center
Foxconn Announces FOXCONN NxVAE, Unsupervised Learning AI Technology
2021/01/21
3+3

TAIPEI, 21 January 2021 - Foxconn Technology Group (Foxconn), a global leader in smart manufacturing, today announced the launch of FOXCONN NxVAE, a new unsupervised learning artificial intelligence (AI) technology that ensures higher levels of efficiency and accuracy in the inspection of defects in manufacturing production lines when compared with traditional practices. 

As a first step in applying this new technology, Foxconn introduced FOXCONN NxVAE to some handheld device production lines in mainland China. After eight months of research and development, those lines successfully reduced the manpower resources required for defect inpsection by 50 percent. The new technology will also be applied to broader manufacturing uses for verticals such as textiles and healthcare, as part of Foxconn’s support for accelerating transformation of a range of industry sectors. 

Figure: The automated production line applied for FOXCONN NxVAE.

New technology transforming the production line defect inspection process
Defect inspection during the production process is an important step to ensuring product quality. Traditionally, Computer Vision technology is used in defect inspection and that technology identifies defects by recognizing and defining a “Golden Sample” - a perfect product sample - and comparing each product with that sample. The “Golden Sample” comparison, however, varies when it is placed in different manufacturing locations. Slight changes, such as the difference of light sources and the positioning of the DUT, Device Under Test, sample, can result in a comparison failure and that is why there is a need for additional manpower for secondary detection. Supervised learning (deep learning) technology is being increasingly used to enhance the efficiency of this process. The machine learning algorithms are taught to distinguish defects through data labelling, training and inferencing. However, that requires the use of high resolution defect visuals for the machine learning process and that can be time consuming and hard to collect given the high-quality production standards within Foxconn. 

FOXCONN NxVAE, with unsupervised learning AI at its core, addresses the above challenges and brings unprecedented benefits to smart manufacturing. The unsupervised learning algorithms in the new solutions can identify products with defects by simply analyzing and clustering unlabeled data from a good product sample without the need for human intervention. The technology, and its increased ability to discover differences in data received, ensures that the inspection process is significantly enhanced, and manual checking becomes unnecessary. Moreover, its self-learning ability does not require collecting, categorizing, and labelling each of the detect visuals thus allowing for a more significant buffer in the usually tight production timeline. FOXCONN NxVAE has the ability to detect the 13 most common types of defects accurately without any errors. 

Continual investment and innovation of AI solutions 
The vision at Foxconn is to pave the way for next-generation AI solutions. The company has deployed several in-house-developed AI solutions on a number of different production lines, leading to an improvement in reporting accuracy from 95% to 99% and a reduction of at least one third of the operating costs for appearance defect inspection projects. The company’s vision is also reflected in the recent launch of “BOXiedge™”, the next-generation AI processing solution for video analytics. That solution provides market-leading energy efficiency for standalone AI inference nodes, benefiting applications including smart cities, smart medical, and industrial IoT. In the future, FOXCONN NxVAE will also be applied and integrated into “BOXiedge™” solutions. 

“The yield rate of our production lines has exceeded 99% and the unsupervised learning algorithm developed by the AI team not only enhances efficiency and reduces the challenges associated with introducing new products into the production line, it also marks an important production efficiency milestone for our industry,” said Gene Liu, Vice President of the Semiconductor Subgroup at Foxconn Technology Group. “This development also demonstrates our company’s vision of “3+3=∞” which symbolizes the infinite possibilities created by Foxconn’s industrial advancement and emerging technologies, and we remain committed to investing in these areas.” 

For more information, please contact: Max Deng (max.cm.deng@foxconn.com). 
FOXCONN NxVAE Product Information: https://nxai.com.tw/nxvae-unsupervised-learning-en/ 

Further Reading
QunaSys and Hon Hai Research Institute Announce Collaboration to Develop Advanced Fermionic Encoding Methods
2024/10/29
QunaSys and Hon Hai Research Institute Announce Collaboration to Develop Advanced Fermionic Encoding Methods
2024/10/29
Hon Hai Technology Group (Foxconn) Adds Two WEF New Lighthouse Factories
2024/10/11
Hon Hai Technology Group (Foxconn) Adds Two WEF New Lighthouse Factories
2024/10/11
Hon Hai Tech Day 2024 Opens To Record Attendance Superchip AI Server Capabilities and EV CDMS Strength On Display
2024/10/08
Hon Hai Tech Day 2024 Opens To Record Attendance Superchip AI Server Capabilities and EV CDMS Strength On Display
2024/10/08