LUSTER: AI-Powered "Lizhu" Makes an Appearance on CCTV's "Smart Manufacturing in China"
In ancient legends, there was a person named Li Zhu who had the ability to discern the finest details from a hundred paces away. Today, an AI version of "Li Zhu" is meticulously examining every detail on the production line of a Chinese manufacturing facility.
On August 8th, the second episode of the CCTV-2 documentary series "Smart Manufacturing in China" titled "Visionary Insight" was aired. LUSTER, representing China's leading machine vision company, made a glorious appearance.
From minutes to seconds : A groundbreaking achievement
The frame in a mobile phone, often referred to as the "backbone" of the device, serves as the framework that holds various components together. The precision of its appearance significantly impacts the quality of the final product.
In the past, the visual inspection of the appearance of the frame in mobile phones relied on manual inspection, resulting in high rates of missed defects and making it challenging to ensure product yield. Data shows that the rate of missed defects in manual visual inspection can be as high as 10% or more, based on the yield calculation of backend products. A single mobile phone frame inspection involves approximately 88 criteria, with the smallest flaw measuring only 0.5 micrometers. It is akin to searching for a grain of rice on a football field, even for the most keen-eyed quality inspector who would require the assistance of a magnifying glass and take approximately 2 minutes and 20 seconds to complete the inspection of a single mobile phone frame. The prolonged search for such minute defects surpasses the limits of human vision.
Zhao Yan explained, "In just 5.5 seconds, the device can complete 88 types of inspections, leaving no room for defects as small as 0.5 micrometers to escape its scrutiny. One device is equivalent to the workload of 40-50 workers, making it significantly more efficient."
The Curved Edge: A Great Innovation - The Significance of the Device
The curved edge, also known as the R angle, is one of the most important design features of a mobile phone frame and has long been regarded as a crucial criterion for evaluating the aesthetic appeal of smartphones. The perfect curvature of the R angle has become a sought-after feature among major manufacturers, leading to fierce competition in its implementation.
The smoother the curvature of the R angle in the mobile phone frame, the more challenging the production process becomes. To meet the curvature requirements, multiple processing steps are performed repeatedly during production, which can result in imperceptible flaws such as vertical lines. This, in turn, adds difficulty to the subsequent quality inspection.
LUSTER's device for the visual inspection of mobile phone frame appearance utilizes collimated light source imaging technology and artificial intelligence. With the assistance of visual alignment compensation detection methods, it controls the positional deviation of the mobile phone's R angle motion trajectory within 0.05mm and the angular deviation within 0.05°. This achievement enables precise imaging of defects that are difficult for the human eye to detect, such as vertical lines and white spots, even under high-speed operation. As a result, it provides crucial support for achieving high product yield in the backend.
Today, LUSTER's device for the visual inspection of mobile phone frame appearance has become a powerful tool for international mobile manufacturers in improving quality and increasing efficiency on their PVD production lines.
Vision + AI : Unlocking the Benefits of Intelligent Manufacturing
The device for the visual inspection of mobile phone frame appearance consists of a vision imaging module and a detection platform module. The vision imaging module ensures that defects are captured in images, while the detection platform module, combined with customized combination light sources, enables high-speed scanning of the entire mobile phone outline during operation. It achieves micron-level precise imaging of defects such as white spots, black spots, and dark dots. Additionally, leveraging LUSTER's F.Brain deep learning platform, the device enables AI analysis, training, and intelligent grading of defect data, thereby reducing false positives and minimizing missed detections.
Furthermore, LUSTER has established the Taihu Industrial Big Data Center, which houses over 60 industrial-grade intelligent AI detection algorithms for defects. Even in situations with limited NG defect samples and a small number of samples, it achieves pixel-level detection accuracy. Through scientific classification and grading of defect images, it can classify 78 types of defects.
In the future, LUSTER will closely follow national policies and industry development trends, leveraging the integration of "Vision + AI" technology with high-complexity industrial scenarios. This integration aims to make machine vision more perceptive, AI algorithms more intelligent, and robotic systems more adept, thereby adding momentum to the innovation and development of China's industrial manufacturing industry.