Artificial intelligence (AI) isn’t a new phenomenon – it has been around for nearly half a century – but in more recent years it has become increasingly prevalent across a range of Internet of Things (IoT) applications. This acceleration has been driven mainly by the evolution of computational power based on GPU (graphics processing unit) computing. This particular architecture provides more cores able to process thousands of operations in a way that was not possible with the previous generations of technologies. At the same time, devices with features such as the ability to work in extended operating temperatures have allowed edge computing to deliver computational power exactly where data are generated.

AI and 5G networking have begun to revolutionise rail transportation around the world by enabling and empowering many smart functions such as visual recognition, deep learning, low-latency communication and many diverse intelligent applications for safer, more convenient transportation infrastructures. 



In 2020, the ARC Advisory Group reported that while AI was suitable for helping to make a range of transportation modes safer, smarter and more reliable, the technology could be applied to improve efficiencies and reduce costs across a variety of train control, safety, supervision and asset management systems. The group said that AI “provides a powerful force to help unlock the potential further digitalisation of railways” and one of the most important areas where AI could be used was in automated train operation (ATO) systems.

Minimising accidents and fatalities 

While ATOs enable access to real-time, accurate data to deliver reliable rail networks, a vital function of any AI system must be to improve safety levels, minimising accidents and even fatalities.

In Taiwan, for example, where there are more than 400 railway crossings on roads, hundreds of crossing accidents occur every year, causing many deaths and injuries. Railroad barriers are not closed by default and often do not close fully when required, causing unnecessary delays and putting road and rail users at risk. It has been accepted that manpower alone cannot guarantee safety at these level crossings so a new project that applied advanced technology was created. The solution needed to take three main factors into account: geographic conditions, weather conditions and driver blind spots to improve safety at entrances to level crossings. The solution put forward to solve the problem was an application that used video analytics.

Accuracy levels of 99.9% 

One of the first problems to be encountered was that railway crossings in Taiwan tend to be spread out across the country so achieving guaranteed safety levels using automated systems would prove very difficult. There were too many variables to consider which meant that response times were sometimes simply not fast enough. To address this, the railway operator in the country looked at edge computing, AI and 5G communications using digital cameras as part of a system that could quickly identify objects, people and the movement of vehicles. The aim was to achieve 24/7 automated railway crossing monitoring without human intervention and with accuracy levels of 99.9%. The solution needed to be rapid enough to compute big data collected from cameras and sensors, integrate it all, process and perform AI video analytics via a VIADIO platform, and provide instantaneous alerts and warnings to vehicles and train drivers. 

The VIADIO AI vision platform came from IronYun USA, a software company that has helped users around the world to deploy tens of thousands of cameras for monitoring applications across many vertical markets. IronYun sells its end-to-end private cloud VIADIO platform to enterprises and governments and the sectors it works in include transportation as well as harbour management, nuclear power plants and public healthcare. 

Powerful edge AI solution

Looking for a powerful edge AI solution to work with its VIADIO software, IronYun reached out to Advantech because of its reputation for highly reliable industrial computers. The edge AI platform needed to support multiple I/O and 5G networking but also – most importantly – had to be capable of handling massive amounts of video with near zero latency. This was vital so that any potentially dangerous incidents at railway crossings could be detected in advance and acted upon in real time, with emergency alerts and warnings sent to operators and train drivers in order to avoid serious accidents.

The Advantech MIC-Jetson series is recognised as the best AI platform at the edge yet. Powered by NVIDIA Jetson technology, the company’s AI edge solutions deliver superior performance per Watt in a compact, embedded footprint. Features of the Advantech offering include strict validation to ensure thermal, mechanical and electrical compatibility, plus industrial-grade anti-vibration, high temperature operation capabilities, I/O support and a compact, industrial design.

NVIDIA® Jetson™-based MXM RTX-3000 GPU card

In the Taiwan case, IronYun needed to install four cameras around railway crossings, connected and powered by EKI-7710G-2CPI PoE switches for image capture. The video data would be sent to an ITA-3650G railway wayside Al edge platform with an NVIDIA® Jetson™-based MXM RTX-3000 GPU card for real-time data analysis and a built-in AIW-355 5G module for data transmission, which could be installed in cabinets at the side of the railway crossings.

The ITA-3650G railway Al edge platform integrates massive amounts of video data from cameras every second and this is processed by VIADIO Al platform in real-time, identifying any objects and their status through multi-frame analysis using its highly capable algorithms. For example, if an object stays on to a railway crossing for longer than 30 seconds, an intrusion event would be identified and confirmed by VIADIO, immediately turned into alert warnings and then sent out through 5G to a train driver within just two seconds. The alarm notification can be sent from 1km to 2km away, giving the train driver or operator plenty of advance notice so that they can slow down the train or stop it altogether as required.

Edge AI platform the only viable solution

This successful installation has shown that for smart applications to be deployed in rural or remote areas, they must use a solution that can be easily implemented, can be highly reliable and can operate in harsh conditions, including extreme weather and widely varying temperatures. The solution must also be cost-effective. The investment for deploying an industrial-grade AI edge platform is a fraction of that when using a cloud-based service. Also, data transmission distances from the network edge to the cloud and back would not be timely and could cause problematic delays, so the high-performance edge AI platforms needed to be installed roadside so they could handle massive data from multiple cameras and sensors without communication issues. The IronYun installation has proved that edge computing is the only viable solution in critical rail applications where safety and speed of reaction are crucial. 

Apart from having a comprehensive product portfolio, the involvement of Advantech was key to this installation. The Taiwan-headquartered company has developed strong partnership alignments over many years and delivers products earlier than other providers. This meant that VIADIO could run pressure tests, quickly respond to test results and deploy all the surveillance systems on-site in the fastest time possible.

In addition, Advantech offers a diverse product portfolio of edge AI platforms in different form factors, including GPU cards and AI accelerators, which are all compatible with the VIADIO software platform and perform at different surveillance scales. This means that whether working on a new installation, a system upgrade or expansion, Advantech is able to offer a one-stop supply service for smart surveillance applications. This service includes R&D and support for both hardware and software, delivering minimal compatibility issues and low total cost of ownership (TCO), as well as worldwide backup from 14 repair centres and four logistics centres.

Helping to drive AI evolution

AI is evolving all the time and is a key component in providing better performances in the rail sector and many other industries and Advantech is continually bringing new products to market to help drive this evolution. That’s why, with its focus firmly on the future, the company has created the MIC-733-AO, its newest edge AI inference system based on NVIDIA® Jetson AGX Orin™. Designed to empower energy-efficient autonomous machines, the MIC-733-AO offers eight times the performance of NVIDIA® Jetson AGX Xavier™ modules while supporting multiple sensors and the latest high-speed interfaces.

Ideally suited to edge AI applications, the MIC-733-AO is a fanless system that is capable of enduring wide operating temperatures (-10°C to 60°C) while offering flexible expansion slots and multiple I/Os. To cope with growing high-performance AI demands, the MIC-733-AO also features four GbE ports for high-speed data transmission that allows 5G communication.

Enabling rail transportation to move forward 

It has long been understood that artificial intelligence is superior in terms of accuracy, reliability, and efficiency when compared to a manual approach. But there has been resistance in some sectors to adopt AI. The reasons have tended to be long system development times, compatibility issues integrating hardware components, and issues with maintenance and inspection that cannot be customised to specific needs. That’s why Advantech’s rail transportation AI solution is fully integrated – combining an industrial camera, processing unit and application software to deliver a complete package that simplifies the project development process and allows for rapid machine vision inspection, without any coding, through an easy-to-use program. This significantly reduces system implementation time and subsequent maintenance costs, which are essential requirements in the demanding and rapidly evolving world of rail transportation.

Backed by its record over many years of delivering powerful and flexible edge AI solutions supporting NVIDIA® Jetson™ and remote management, Advantech is enabling rail transportation to move forward into a new era where automation and autonomous operation can boost efficiency and safety levels higher than have been previously possible. 

About Advantech

Established in 1983, Advantech is a leading provider of trusted innovative products, services, and solutions. Advantech offers comprehensive system integration, hardware, software, customer-centric design services, embedded systems, automation products, and global logistics support. We cooperate closely with our partners to provide complete solutions for a wide range of applications in diverse industries. Our mission is to enable an intelligent planet by developing automated and embedded computing products and solutions that facilitate smarter working and living. The number of applications and innovations made possible using Advantech products is unlimited. 

Download this article in PDF

Read more articles

More Articles

Related to ADVANTECH

Follow Advantech