Hitachi Rail and Nvidia are developing AI-powered train tech that can ‘read any possible problem,’ says the railway company’s group CEO
Hitachi says it installed its AI platform on more than 2,000 train carriages.
Hitachi Rail, part of Hitachi Group, provides services for urban, mainline, and freight railway operators, including for signaling, rolling stock, telecommunications, supervision, ticketing, and payment. Hitachi Rail has 24,000 employees working in more than 50 countries.
Situation analysis: What problem was the company trying to solve?
The number of train passengers is rising. Giuseppe Marino, the group CEO of Hitachi Rail, told B-17 that rail operators faced challenges such as reducing maintenance costs, improving reliability, and extending trains’ life cycle.
Hitachi Rail has been using digital sensors to monitor the health of transportation infrastructure — train cars, rails, signaling systems, and tunnels — and improve their performance. But the company wanted to use artificial intelligence to speed up the analysis of that data.
“We needed to find a solution that leveraged the existing infrastructure,” Gajen Kandiah, the president and chief operating officer of Hitachi Digital and a co-lead of the Hitachi AI Transformation Center, told B-17. “The question really became: We are capturing all of this data on the edge because the trains are rolling at all times, we have digital capabilities internally — how do we connect the two?”
Gajen Kandiah is the president and chief operating officer of Hitachi Digital.
In September, Hitachi Rail partnered with Nvidia to launch the Hyper Mobility Asset Expert, or HMAX, an AI-powered digital-asset-management platform. The goal is to improve trains, signaling, and infrastructure management.
Key staff and partners
Teams within Hitachi Rail and Hitachi Digital worked with Nvidia to develop HMAX. Kandiah said it combined Hitachi’s existing sensors and rail data with Nvidia’s IGX edge AI platform, which enables the devices to process the data where they are instead of at a central facility or data center.
“Their team worked hand in glove with our teams,” he said, adding that Hitachi Rail began working with Nvidia on the platform in February.
AI in action
HMAX integrates with rail companies’ operations and maintenance systems. Hitachi Rail said the platform combines live data collection from train sensors and cameras with AI to help operators predict problems, improve their networks, and ensure trains and related infrastructure run more efficiently.
“We’re collecting a huge amount of information,” Marino said. “With Nvidia, we can accelerate the way we’re applying artificial-intelligence systems to read any possible problem in a very smart way.”
Giuseppe Marino is the group CEO of Hitachi Rail.
Previously, rail operators had to bring trains back to stations and manually offload the data collected by sensors and cameras, such as the temperature of the train’s motor or how much the train is vibrating. Marino said the AI platform automatically pulls that data into a central location, or data lake, for analysis.
For example, with HMAX, operators can monitor overhead lines, quickly identify cables that need repairs, and prevent disruptions. Nvidia said it traditionally took about 10 days to process a day’s worth of video collected by trains.
Kandiah said that eliminating the lag time between data collection and analysis improves the efficiency, reliability, and longevity of rail infrastructure.
Did it work, and how did leaders know?
Marino told B-17 that over the past few months Hitachi had installed HMAX on more than 2,000 train carriages.
Marino added that HMAX was reducing service delays by more than 20%, train maintenance costs by 15%, and fuel costs at train depots by 40%.
Kandiah said the tech was also lowering the total cost of ownership of the trains, “so there’s the opportunity for an operator to run the train longer” and replace trains less frequently.
What’s next?
Marino said Hitachi Rail was working on expanding HMAX to other rail operators, including the Copenhagen Metro in Denmark, adding that the technology is applicable to other industries with large infrastructure, such as the energy sector.
The company says it’s also increasing the quality and number of data points HMAX is collecting.
“As much as we are doing, we’re learning,” Kandiah said. “We are starting to see the potential of the product and of the technology — how do we start to solve different use cases?”
