Testing is an indispensable parts of the modern industrial production process. Each step of mass production process has a certain defect rate. Though the individual figure is rather small, when multiplied, the result can constitute a bottleneck for a factory to break to improve yield. In the past, workers in the electronics manufacturing industry had to hold complex printed circuit boards in hands, and spend a long time checking the appearance of the boards and detecting defects. This practice certainly did not meet the requirements of high-speed production.

Guangzhou Leichen Technology Co., Ltd. (“Leichen Technology”), a holding subsidiary of CVTE, applies robotic applications, machine vision and automated testing to inspection systems, to develop high-end, flexible, intelligent industrial production and inspection equipment. Leichen Technology integrates deep learning algorithms with big data training intelligent models and applies them to automatic optical inspection (“AOI”) to facilitate efficient and accurate industrial visual inspection.

Have you heard of AOI?

AOI is a visual inspection technology based on optical principles that replaces manual visual inspection with machine vision. The system has a built-in high-resolution industrial camera, which performs intelligent image analysis after scanning the surface of electronic circuit board, to detect faults, leaks, reverses, and other defects on the plug-in components of electronic circuit board. It is able to detect details that the human eye cannot see. 

Deep Learning=Smarter

Traditional AOI is automated, and with manual settings, can replace a large amount of repetitive visual inspection. However, it is difficult to use traditional AOI to process and train the massive data generated in industrial inspections. Consequently, it may lead to waste of data and stagnation of the applications.

To make inspection more "independent and intelligent", Leichen Technology independently developed deep learning and artificial intelligence solutions and applied them to AOI. This deep learning algorithm imitates the works of the human brain, constructs a deep neural network, and extracts automatic features from the raw data. When an undetectable defect occurs, the inspection system can collect data through the device and perform fine-tuning training of the model under certain manual intervention. During the training process, the test results are continuously optimized as the data set is expanded and the model strengthened.

Strong Adaptability and Compatibility with Many Industries

The production processes and technologies vary widely in each vertical field of the manufacturing industry, and the cost of the application of AOI to each field is high. The adoption of traditional machine vision in industrial inspection still faces the challenge of adapting to different production environments. Even the slightest variance in the production environments may affect the detection accuracy.

To solve this problem, Leichen Technology developed deep learning algorithm of strong compatibility and flexibility to adapt to different industries. For example, when the AOI system of Leichen Technology performs component polarity identification, it is not affected by differences in imaging angle, position, illumination, texture, color, and size etc., and is compatible with a variety of industrial environments. Now the AOI system has been applied in many industries, including automated inspection of LCD, visual inspection of automotive electronics, and electronic products assembly and processing.

The "black technology applications" mentioned above are inseparable from the technical support behind it: the Central Research Institute of CVTE and the professional machine vision R&D team of Leichen Technology. The former conducts basic research in various fields such as visual computing, data mining, and robotics to explore unknowns and discover new knowledge. Based on its underlying technology, Leichen Technology's R&D team productizes the technology, and introduces these “black technology applications” into more industrial fields.

In the future, factories will be equipped with flexible collaborative systems, which run the entire production process autonomously, optimize throughout the process, and adapt to new environments in real time. CVTE will harness the power of cutting-edge technologies such as visual computing and data mining to improve the manufacturing models and industrial systems, to inject new energy into the construction of smart industries, and shed the light of technology into our daily lives.