The Time is Ripe for AI in Industrial Manufacturing

By Hanson Chen, Foxconn VP

To remain competitive in global markets, industrial manufacturers have been quick to adopt new technology to streamline production workflows. With the advent of artificial intelligence, manufacturers are presented with a tremendous opportunity to enhance every step of operations, from product development to supply chain logistics, to meet consumer demand and provide the highest quality of products.

The potential of AI in manufacturing is only strengthened by the vast amount of data now available within the industry. The massive network of sensors on machines and the sheer scope of day-to-day operational data industrial manufacturers collect can equip AI with insights to augment workflows.

The key to unlocking this promise relies on the computational power required to process large datasets and our own human imagination.

Luckily, at Foxconn, we have both.

AI Inspection Instruments at the Edge

At Foxconn Technology Group, one of the key ways we’re using AI is for automated high-precision inspection of our products’ components and tools.

There are numerous defect types that need to be inspected, which is challenging under demanding production line speeds. It’s overly laborious work that causes worker fatigue, compromises safety and quality control.

To address this, we have developed an AI Inspection Instrument (AIII), a high-performance computing and deep learning technology focused on surface defect inspection.

With AI-enabled optical imaging, it solves pain points in the chassis process of the desktop computer, including stamping, tapping, and painting processes. With AIII, we’re able to protect our workers by giving them the chance to work on more creative tasks, while it takes care of the more menial tasks and boosts overall product quality.

Mechanics of AIII

By simultaneously checking defect types and locations, the AIII can deliver accelerated and accurate inspection standards.

Our team of data scientists used NVIDIA CUDA-X AI software libraries to train our models, and for inference we used NVIDIA’s EGX platform to deploy new AI capabilities at the edge. Our optical imaging component, equipped with multi-camera and LED lights to complete dynamic optical shooting, helps us achieves a complete visual analysis of a product without stopping the production line.

Speed is critical in production. By using NVIDIA EGX platform, the models and image data processing ran over 40X faster than earlier hardware.

Future of AI in Manufacturing

AI-based optical inspection is fast becoming the driver of high-performance computing in intelligent manufacturing and we expect that it will become the key to building smart factories.

The possibilities of AI in manufacturing are limitless: 3D vision robotic application, logistics robot navigation, and the industrial big data brain are just a few areas that manufacturers can explore to improve product quality and operational efficiency.

With AI, the industry has yet another chance to ignite the next frontier of innovation.