Background
One of the biggest challenges facing today’s textile industry is the fact that it for the most part still relies on human vision and manual inspection for defect detection, meaning the industry not only operates on thin margins but also has a considerably high defect rate.
Smart textile, using artificial intelligence and computer vision, aims to act on the knitting machines and to reduce defective production rate by enabling real-time defects and irregularities detection, therefore bringing defective production rate to or close to 0%.
Requirements
A system integrator came to Lanner looking for a hardware appliance that can be relied upon for realizing the aforementioned high-tech solution, one that not only can be made ready for stopping the production of defects during production by way of using hardware and software but also delivers an intuitive management UI on which the production stats can be checked and verified.
The said smart textile solution is to be deployed as a connected platform that provides production visibility, error-free machine inspection and real-time data and helps make supply chains more transparent so that waste can be reduced, and quality can be ensured. In other words, textile output with fewer defects, higher quality and higher margins.
Lanner Solution
The EAI-I130 is an industrial-grade edge AI inference system equipped with NVIDIA® Jetson Xavier NX or NVIDIA® Jetson Nano System-On-Module.
The EAI-I130 is one of Lanner’s latest intelligent Edge computing platforms that deliver security, low-latency and a ruggedized design that withstands the harsh outdoor deployment settings. Depending on product configuration, this 5G Edge AI inference appliance may be equipped with a 6-core NVIDIA Carmel ARM®v8.2 64-bit CPU, 384-core NVIDIA Volta GPU and either NVIDIA® Jetson Xavier NX or NVIDIA® Jetson Nano, therefore delivering up to 21 TOPS and 48 Tensor Cores of AI performance.
In addition to 5G and WiFi6, it also comes with dual GigE PoE LANs that support IEEE 802.3 af/at PoE+ (By SKU). Furthermore, the EAI-I130 is fanless and IP40 rated. What’s more, this particular IPC is fanless, IP40 rated and features -40°C to 70°C extended operating temperature range.
The EAI-I130’s compatibility with both NVIDIA® Jetson Xavier NX and NVIDIA® Jetson Nano makes available supercomputer performance to the Edge in a small form factor System-On-Module, making the EAI-I130 the perfect platform for running modern neural networks (including computing vision) and for processing data from multiple high-resolution cameras and sensors.
Results
Powered by NVIDIA® Jetson Xavier NX/Jetson Nano that offers up to 21 TOPS AI performance, the said SI’s proprietary software integrates seamlessly with the EAI-I130, making available an all-in-one UI that includes performance dashboards for simplified management, remote control for 24/7/365 cloud-based access and business intelligence for better decision-making.
The EAI-I130 comes with 2x GigE PoE LANs that support IEEE 802.3 af/at PoE(+), 1x HDMI port, 4x DI/4x DO, 2x USB 2.0 and 2x COM ports, all ideal for connecting cameras, lights, sensors and other peripherals. High resolution cameras that make up the core and are the eyes of a smart textile production system; lights must be leveraged for detecting defects invisible to the human eyes.
Furthermore, the EAI-I130’s compatibility with both NVIDIA® Jetson Xavier NX and NVIDIA® Jetson Nano makes artificial intelligence and machine learning available to the next-gen textile production. Automatic defects detection systems with machine vision set up using the EAI-I130, one of Lanner’s edge AI inference platforms, enable detection accuracy improvement over time as irregutlatires are detected in real-time and automatically recorded on the X-Y roll map, greatly reducing their downstream impact in the supply chain.
Conclusion
AI inferences at the edge opens up vast amounts of new markets and innovative applications that can benefit from its high throughput, efficiency, and accuracy. The textile industry sure has a lot of gain from edge AI inference computing’s potential to create faster, more reliable and cost-effective machine vision systems that detect, intervene and prevent almost all detects, therefore leading to lower operating costs, minimized waste, better values and higher margins.