Making Network Hardware with AI-Enhanced Computer Vision
Did you know that integrating artificial intelligence into the PCBA process can boost inspection accuracy by up to 98%? In the world of electronics manufacturing, the bar for quality and efficiency is constantly being raised. At Accton , we pride ourselves on constant innovation to add value for our stakeholders. This blog details how deploying high-resolution cameras and machine learning algorithms has elevated our manufacturing quality, reaffirming our position as the premier partner for networking hardware.
The Challenges
In Automated Optical Inspection
Traditional AOI systems have served us well, but they come with limitations. Typically, these systems handle only single-sided inspections and detect issues like missing or misaligned components. However, they often miss subtle yet critical defects such as scratches, stains, and other imperfections that can compromise product quality. Another recurring complaint about AOIs is the high number of false positives, resulting in inconsistent quality control, increased rework, and potential customer dissatisfaction. These limitations slow down production lines and drain resources, leading to higher operational costs and reduced efficiency.
In Final Assembly
In the final assembly stage, operators use poka-yoke guides and follow standard operating procedures to secure the board to the chassis with proper alignment and secure mounting. However, these methods too have limitations. Misalignment can lead to overheating, reduced performance, and premature failure of a very expensive network device. Also, it’s not uncommon for an operator to miss a screw during assembly. This oversight can cause mechanical stress on the PCB, leading to cracks, warping, or connection failures. These issues highlight the challenges of manual assembly processes, where precision and consistency are critical but hard to achieve.
In Packaging
Packaging, the final step before products are shipped to customers, also presents specific challenges. Everyone has experienced receiving a product missing accessories or with the wrong items included. It's alarmingly easy to pack the wrong power cable, accessories, or documentation specific to the product coming off each line at any given time. Such mistakes not only frustrate customers but also increase return rates and operational costs.
These cumulative issues and others across different stages of manufacturing underscore the necessity of a solution that ensures quality, efficiency, and repeatability.
Introducing Accton’s AI Inspection Platform
Following the hierarchy of human senses, where we trust our vision to perceive about 80% of the world around us, the camera sensor emerged as a rich source of data capable of revealing key insights about production. High-resolution cameras add super power-like capabilities to the operation – seeing details too small to the human eye, maintaining precision at high speeds, and functioning reliably for any number of human shifts.
Solving Optical Inspection Challenges with AI
To address the limitations of traditional AOI systems, Accton has integrated computer vision, augmented with artificial intelligence. Our high-resolution cameras and image recognition algorithms detect even the most subtle defects, such as micro-scratches, stains, peeling, label defects, tiny misalignments and more that traditional AOIs often miss. Additionally, our system significantly reduces the occurrence of false positives, ensuring that only genuine defects are flagged for rework.
By integrating AI-enhanced computer vision, we validate inspection results up to 98% faster and drastically reduce the operators' workload and visual inspection by 85%.
We continually improve the solution by retraining the model with new data using the latest data pipelines and AI tools available on the AWS cloud and vast amounts of data from the over 8 million devices we ship every year. This continuous loop of data collection, model retraining, and deployment allows us to reach superior inspection quality and adapt quickly to any changes in processes and new product designs.
领英推荐
Automated Inspection in Final Assembly
One standout feature of our solution is real-time assembly verification. Using live video feeds from cameras positioned above the assembly lines and custom-designed edge computing devices, we let AI compare each assembly step against the correct sequence of parts and actions. This early error detection increases overall assembly quality, improving line throughput and minimizing waste of materials and time, leading to more economical and effective operations, and happy returning customers.
Precision in Packaging
Our AI-powered inspection system also excels in packaging verification. It inspects each package to confirm it contains the correct device, accessories, and manuals, reducing the risk of shipping incomplete or incorrect orders, enhancing customer satisfaction, and minimizing returns.
Real-World Benefits of AI-Enhanced Computer Vision
By integrating AI and computer vision across various stages of manufacturing, Accton has achieved:
Incorporating computer vision and AI into our manufacturing process embodies Accton’s commitment to innovation and quality. By tackling the limitations of traditional AOI and enhancing our final assembly and packaging processes, we've achieved remarkable improvements in efficiency and precision. These advancements optimize our operations and ensure that our products consistently meet the highest standards.
Our technological leadership extends beyond Accton. Other companies in Taiwan that have adopted our AI inspection platform—including cameras, computing hardware, image processing software, models, and other IoT devices—have seen similar transformative results.
At Accton, our dedication to incorporating the latest technologies means we are continually evolving, always ready to adopt new advancements to stay ahead. This focus on excellence ensures that we deliver extraordinary value to our customers and maintain our position as a trusted partner for networking hardware.
?? Learn more ?? https://www.accton-iot.com/pallas/en/index.html?
?? Stay connected with us ?? https://www.accton.com/newsletter/
?? Contact us ?? https://www.accton.com/contactus/
Senior Project Manager 位於 Foxconn Industrial Internet
8 个月Very informative