Tiziano on the VIA transition from processor platforms to Hard Hat AI Solutions

From Processor Platforms to Hard Hat AI Solutions

Tiziano Albani, the head of VIA Europe, recently had an interview with the Italian technology website Edge 9 Hardware Upgrade in which he talked about our transition from processor platforms to Hard Hat AI solutions. Here is a translation of Tiziano’s remarks. You can read the original Italian version here.

Using AI to solve forklift problems

VIA was one of the most prominent names in the processor world in the 90s and early 2000s, especially within the x86 world, then gradually withdrawn from the consumer and server markets. What is the current position of the company?

Over the past ten years we have transitioned our business from semiconductor design to intelligent solutions for automotive, edge, industrial, and building applications. Our strategy is to provide highly integrated solutions that can be rapidly deployed by the customer to solve critical real-world problems. For example, our VIA Mobile360 Forklift Safety System combines the system hardware, cameras, cables, mounts, and other accessories with AI-powered pedestrian detection and driver fatigue and distraction monitoring applications as well as the VIA WorkX Connect Cloud Management Service in a single package. Installation of the complete solution on EV or combustion powered forklifts takes less than fifteen minutes in most cases.

Our silicon design expertise has been invaluable in the development and optimization of the core computer vision algorithms for VIA Intelligent Solutions. Over ten years of embedded system design experience has enabled us to create ruggedized low power systems that deliver reliable operation in the most demanding on-road and off-roads environments.

Using AI for inspection solutions

What kind of processors does VIA use in its solutions?

We use a variety of processors from MediaTek, Qualcomm, Intel, and other suppliers depending on the target application of the system. The VIA AMOS-3007 system, for example, features the 1.5GHz Intel® Atom® quad-core processor.

We see strong potential for MediaTek Genio processors for industrial, commercial, and consumer edge AI and IoT systems and devices, and have developed a growing range of SOM-9X Series modules and systems to meet growing market demand. In addition to offering a choice of hardware platforms, we are also building a rich development framework called VIA LenZ that makes it easy for our customers to create and deploy machine vision applications such as helmet detection and defect detection to meet their specific safety and production quality needs. Providing a complete hardware and software platform that enables the rapid integration and optimization of small footprint data models for different applications is critical for meeting the diverse needs of the market.

Using AI for PPE and pipeweld inspection models

Large artificial intelligence models require extremely high computing power, but they are not the only ones to give benefits to users and companies. What is the position of VIA with respect to AI? What do its products offer in this segment?

The proliferation of large language models such as ChatGPT is leading to a bifurcation of the AI market. On the one hand, we are seeing the emergence of what I call the Hollywood AI segment, which is focused on making it easier for individuals and companies to generate huge volumes of new content in textual and multimedia formats for articles, videos, advertisements, homework assignments, business presentations, and analyst reports.

While the potential for Hollywood AI is enormous, we are already seeing the emergence of a number of serious issues that will need to be addressed by governments, enterprises, and users, including concerns over IP ownership (as is already being seen in Italy!), the integrity of the data and content, and, as you point out, the massive volumes of compute power that are required to run the large language models. Although there is no doubt that adoption of Hollywood AI will grow very quickly, these issues will need to be addressed if the market is to develop on a healthy and sustainable basis.

Using AI for large language models

On the other hand, we are also seeing the growth of what I call the Hard Hat AI segment, which is aimed at providing solutions for practical problems at the physical edge such as preventing accidents on building sites using people and object detection or detecting defects in product quality in real-time before the goods leave factory. This is the segment that we are focusing on.

With our VIA Intelligent Automotive Solutions, for example, we provide a comprehensive range of smart safety systems that can be installed on vehicles of all shapes and sizes – from forklifts and passenger cars to gigantic haulers and loaders used for mining and quarrying. With our VIA Intelligent Industrial Solutions, we enable leading petrochemical companies to prevent potential leaks with AI-powered pipeline inspection systems and make sure that staff are wearing the appropriate safety gear before entering hazardous zones in factories and warehouses.

If the cloud processor market is in the hands of AMD, Intel and other names like AWS and Ampere, the edge world is more open to other manufacturers. What is VIA’s offer in this area?

As mentioned above, we offer a growing variety of Hard Hat AI solutions for intelligent automotive, industrial, and building applications. By providing highly integrated packages that combine advanced AI and IoT technologies with strong domain knowledge and expertise, our aim is to make it fast and convenient for customers to deploy our solutions at scale across their global operations.

VIA Technologies, Inc.
VIA Technologies, Inc.