Advantech, AMD and Mentor partner on AI
March 6, 2019
Advantech, AMD and Mentor announced at last week’s Embedded World in Nuremberg that they are working together to make artificial intelligence (AI) technology more accessible and easier to implement.
“We can’t do everything by ourselves,” said Rex Lee (pictured below), director of Advantech’s embedded IoT group. “We are going to be an enabler and do joint ventures. That is why we are doing co-creation with AMD and Mentor. We are producing a unified platform to accelerate AI development.”
AI technology is predicted to take embedded systems to the next level with higher efficiency and smarter systems designed to improve people’s lives. For instance, diagnostic errors among the 300 million diagnostic radiology images that are captured in the USA contain errors of up to four per cent. AI-infused image recognition using machine learning can see far more detail in MRI and x-ray images than human eyes, so it can improve diagnostic accuracy and help prioritise treatment.
AI technology in embedded system is often integrated in service model innovation rather than the manufactured product and it is highly sophisticated in the numerous tasks it performs including data collection, data analysis, pre-trained models and inference. In terms of machine learning, programmed algorithms rely on powerful and reliable computing units for big data consumption. That’s why edge computing is an important medium to satisfy the connection between the cloud and sensor devices.
The partnership between Taiwanese board maker Advantech, US chip company AMD and Siemens subsidiary Mentor can help accelerate AI implementation by integrating each party’s products and services. Each is devoting resources to make edge computing easier to apply so users can concentrate on AI application development for their hardware and middleware.
“Silicon is only as good as the system you build around it,” said Stephen Turnbull (above), director of product marketing at AMD. “We are bringing together our high-performance quad-core CPUs and high-performance graphics.”
Advantech is providing an embedded system-on-module platform equipped with the AMD Ryzen Embedded V1000 processers and the Mentor Embedded Linux operating system. The V1000 processor supports frameworks, libraries, tools and compilers for machine vision applications that leverage Vega GPU technology.
“IoT and cloud are integral to everything we do,” said Scot Morrison (below), general manager of Mentor’s embedded platform technology business unit. “We have been one of the longest term Linux providers.”
The three companies are embracing open standards such as Open VX and Open CL supported on the Linux kernel, so users can migrate machine learning across diverse hardware architectures for various AI applications.
According to ABI Research, shipments of devices with edge AI capabilities will grow by fifteen times to 1.2 billion units by 2023. AI at the edge enables real-time decision making without the limitation of latency, cost, bandwidth, power consumption and other considerations.
• Advantech has revealed hardware and software integrated edge AI products powered by Intel Vision Accelerator Design with Intel Movidius Myriad X VPU and Intel’s distribution of the OpenVino toolkit.