Stylez, headquartered in Tokyo, provides Java applications and open source solutions to industrial and electronics manufacturers. Its parent company, Ryoyo Electric, is the largest supplier of NVIDIA® Jetson™ edge computing devices in Japan.
At-a-Glance
Working with its parent company Ryoyo, Stylez is using Rancher Prime to support a global automotive parts manufacturer to streamline its assembly lines by using AI. The AI solution uses palm-sized, edge-computing devices to analyze video information in real time, enabling the manufacturer to scan parts for defects as they move through its factories.
The journey to containers
As in many other economies, the competition for talent in the Japanese job market has intensified in recent years. However, Japan also has a more significant structural challenge: its population pyramid is unusually top-heavy, with proportionately more people past retirement age. The country also has relatively low birth and immigration rates, and the working population is predicted to shrink by as much as 20% by 2040 (World Economic Forum, Feb. 12, 2019).
To help its customers tackle the challenges associated with limited access to skilled staff, Stylez and Ryoyo Electric are actively promoting intelligent automation and orchestration solutions. These are particularly important in manufacturing, where it is vital to keep production lines running smoothly — typically on a 24/7 basis. Halting production hugely affects cost, so manufacturers are keen to use technology to help them achieve more with the same human resources.
The good news is that the latest edge computing technologies make bringing AI-assisted solutions to the production line easier. These solutions can analyze live video feeds to inspect the quality of manufactured goods and are interconnected to create smart factories that can run with little or no human input.
However, while creating an edge device capable of monitoring manufacturing quality is relatively easy in theory, deploying and running large numbers of these devices in factories is difficult. Any instability in the device or the application could halt production and lead to losses. Devices must be highly tolerant of extremes of heat and vibration, and applications must offer fast and easy upgrade paths to minimize the need for on-site maintenance.
When a major automotive parts manufacturer approached Stylez and Ryoyo for an AI-assisted quality inspection solution in its factories, Stylez determined that a containerized architecture would be most appropriate.
Tetsurou Yano, expert at Stylez, says: “Containerization is now effectively the industry standard for industrial applications, wherever there is a requirement to flexibly deploy applications to multiple physical hosts. As a container orchestration platform, Kubernetes provides a single place for us to manage all containers and configuration information.”
“We estimate that using Rancher Prime improved the efficiency of our containerization set-up by as much as 40% compared with vanilla Kubernetes. This means we spend less time on configuration and more time solving challenges for our customer — our engineers have more time to deliver value.”
Why Rancher Prime?
Stylez wanted a Kubernetes distribution backed by long-term support from a recognized industry leader for two key reasons. First, it aimed to achieve a rapid rollout without needing teams to gain deep expertise in Kubernetes. Second, it wanted to ensure that its customers could achieve the highest possible levels of availability for edge computing solutions in their factories.
Tetsurou Yano says: “Rancher Prime is a comprehensive product. It simplifies the use of Kubernetes by reducing the need for highly skilled staff and extensive training. Rancher Prime lowers the barriers to entry for container orchestration.”
Stylez containerized the AI visual inspection technology and its automotive customer rolled it out through Rancher Prime. The containerized solution runs on palm-sized NVIDIA® Jetson Xavier™ NX edge devices, each equipped with a powerful GPU for high-speed processing of AI workloads. The relatively low cost of these devices, combined with Rancher Prime’s efficiency in managing large sets of containers, makes it possible for the automotive company to deploy a huge number throughout the multi-stage assembly lines in its factories.
Takeshi Nakamura, deputy, solutions business unit 5 at Ryoyo, says: “Our solution works reliably on the factory floor to perform quality inspections, monitor equipment and collect data on throughput.”
The impact of Rancher Prime
Unstoppable production
The AI-assisted, quality-inspection solution created by Stylez and Ryoyo is helping their customer achieve its vision of the smart factory of the future. The solution uses AI to analyze and interpret live video from cameras monitoring the production lines, scanning parts for defects as they move through the process.
Thanks to the automatic load balancing and self-healing functions in Rancher Prime, the solution runs reliably and without interruption — helping the automotive parts manufacturer move towards an “unstoppable” manufacturing capability.
A representative for the automotive parts manufacturer says: “With our Rancher Prime environment running on NVIDIA edge devices, we are one step closer to realizing our goal of full automation with zero defects and no interruptions to manufacturing.”
Resilient, modular, self-healing
To facilitate the central management of the widely distributed edge devices, Stylez separated the logic of the qualityinspection application into individual containers. This ensures that the company can apply updates or upgrades to individual functions without bringing down the whole system. The solution is also resilient against hardware failures: the automotive parts manufacturer keeps a number of the inexpensive NVIDIA edge devices on stand-by, ready to replace existing devices on the production lines.
Takeshi Nakamura says: “If a device fails, our customer can simply replace it with a spare, and Rancher Prime will automatically restore the application over the network.”
Efficient deployment
Backed by responsive support from SUSE, the smart-factory project has not experienced major issues or stoppages. The decision to use Rancher Prime has contributed to the smooth rollout, helping Stylez to hit the ground running and save time in deployments.
Tetsurou Yano says: “We estimate that using Rancher Prime improved the efficiency of our containerization setup by as much as 40% compared with vanilla Kubernetes. This means we spend less time on configuration and more time solving challenges for our customer — our engineers have more time to deliver value.”
What’s next for Stylez?
While continuing to deploy edge devices in the automotive parts manufacturer’s factories, Stylez and Ryoyo are also winning new customers for similar AIassisted edge solutions in other industrial settings. With Rancher Prime, deploying and managing resilient software across large numbers of devices is relatively simple. This means that the companies can dedicate more time and effort to helping their customers achieve more, even as the talent pool shrinks.