Exclusive interview: Hitachi Europe boss explains how company uses AI for IIoT

Exclusive interview with Ram Ramachander, chief commercial officer for social innovation in the EMEA region, Hitachi 

Hitachi is one of the largest companies in the world, employing more than 300,000 people globally and generating revenues of more than $80 billion a year across all its many business units.

Classed as a conglomerate, Hitachi has interests in a huge variety of markets, including, of course, heavy industry.

The areas Hitachi has been particularly interested in lately could be summed up as smart cities and transportation, which the company tends to group under “social infrastructure and industrial systems”. 

This is the area of responsibility for Ram Ramachander, whose official job title at Hitachi is chief commercial officer for social innovation in the Europe, Middle East and Africa region.

 

Robotics and Automation News caught up with Ramachander at a recent conference in London, England, entitled the Hitachi Social Innovation Forum, at which the company’s chairman, Hiroaki Nakanishi, made the closing remarks, briefly sharing the stage with the Hitachi’s humanoid robot, Emiew.

Ramachander was one of the keynote speakers at the event. In fact, he opened the conference with the first talk of the day.

“The four areas that I talked about is where we’re wholly focused for success,” says Ramachander.

“So it’s manufacturing, transport, smart energy and video intelligence and public safety – or smart spaces and video intelligence, if you want to call it that.

“So those are the four areas where we are very laser-focused on business growth, digitisation, incubation – everything that we’re doing in the EMEA region.”

A lot of what Ramachander does revolves around developing business concepts and models, based on technologies that are either relatively new or entirely new.

Ramachander explains: “Firstly, we establish the market opportunity in the region for any one of the areas I mentioned earlier.

“We then look at all of the technology requirements, and plot the ecosystem around that.

“So, for instance, in smart energy, if you’re looking at electric vehicles, we plot the entire ecosystems around EVs, and we look at how technology can enable the opportunities in that area.

“And then we work with customers to make it real in delivery. And then we commercialise that, we scale it across the region, in terms of what we do.

“So, following that initial period of incubation, our responsibility then is to commercialise into a real business – launch it as a business in the region.

“That’s primarily what we do.”

Setting the stage 

The Social Innovation Forum was hosted by The Telegraph newspaper and provided a platform for Hitachi and many of its partner companies to discuss what new technology can do for the private sector as well as the public – or government – sector.

For Hitachi, the line between public and private can become blurred because it’s often involved in large-scale transportation and infrastructure projects, which almost always require the approval of some sort from regulatory authorities connected to the highest levels of government.

For example, Hitachi is a large-scale manufacturer of rolling stock, or trains, and partners with Mitsubishi Heavy Industries on international and intra-city railway systems. And railways are, of course, a political issue in most countries.

Among the many corporate moves the company has made in recent years is the acquisition of Ansaldo, an Italian rolling stocking manufacturer.

This is interesting because Ansaldo recently partnered with mining giant Rio Tinto to build an autonomous freight train, a technology that’s likely to become more important, in light of growing international logistics networks.

One of the most ambitious plans for logistics networks is China’s so-called Belt and Road initiative, which envisions at least three direct routes from the Pacific coast of China into southern and northern Europe, some of which will be traversed by railway lines.

The Belt and Road project is said to represent a revival of the ancient Silk Road, which one imagines as being little more than a dirt road back in the day, but now, railways will arguably be a more critical component in the network than roads and sea lanes, considering the gigantic landmass we’re looking at – Eurasia – and how much freight each train can carry compared with the average truck.

Such long journeys – whether by road, rail or sea – will almost inevitably see autonomous transportation systems being developed, if only for freight services at first.

It will also coincide with the development of stronger data networks and individual conurbations of high connectivity along the road – meaning there will be many more smart cities and towns in Europe and Asia in the future, possibly on logistics networks such as the one being propagated by the Communist Chinese.

One of China’s largest telecommunications companies, Citic, recently bought a Netherlands operator of internet points of presence, Linx Telecommunications, a deal which perhaps indicates one potential plot, or “roadmap” as the Chinese call it, or direction of development.

Where this is all leading to is difficult to describe in detail – mainly because it’s such a large, long-term, dreamlike idea – but certainly it will involve a variety of industrial sectors, such as construction, manufacturing, logistics, and telecommunications.

And the desired end result is simple: much-improved land transportation connections between China and Europe, perhaps competing with shipping lanes, which still account for the vast majority of freight transport today.

Platform shows

Hitachi is, of course, a Japanese company, but it has extensive operations in Europe, partly due to its growing rolling stock manufacturing business – the company is planning to build a “bullet train” for the region, much like the famous Japanese Bullet Trains, or Shinkansen, which were originally built by Kawasaki Heavy Industries.

Hitachi faces some serious competition, however, from European giants of the railway business, as it does in other regions and other markets, much like any other conglomerate.

In the industrial internet market, for example, Hitachi could be said to be a new entrant, with its recently launched Vantara enterprise and its Lumada platform – a business model reminiscent of GE Digital and its Predix platform, or Siemens and MindSphere, or ABB and Ability, or the many others which have sprung up in the past year or two.

Such platforms essentially enable direct monitoring of industrial machines, the data from which can be used to analyse performance and predict and plan maintenance requirements.

General Electric claims to have been the first, with Predix, launched in 2015. But many others have quickly followed suit and all have realised it’s a digital goldmine.

Hitachi has combined three of its business units – Hitachi Data Systems, Hitachi Insight Group, and Pentaho – into one single unit, to create Vantara, which instantly makes it a $4 billion company.

It’s probably too early in the development cycle of the industrial internet platforms market to dwell on numbers, but it so happens that GE Digital claims its Predix platform has helped it generate $4 billion of revenue a year.

The qualification here is that General Electric is such a gigantic company that even if GE Digital only supplied its parent company, it would probably make $3.99 billion a year.

A similar thing could be said of Hitachi and Vantara, or Siemens and MindSphere, or ABB and Ability. But that’s probably more a reflection of our lack of knowledge about the business ecosystem as well as an indication of the nascent nature of the technology.

Rather than, or perhaps as well as, comparing and contrasting these companies on how much revenue they claim to be generating, another measure could become how well their platforms and associated technologies are performing technically.

That might be easier since we’re talking about computers and the internet, where bandwidth and data transmission speeds are fairly widely used metrics.

Another key differentiator might become how fast their software can analyse the vast quantities of data the connected sensors are collecting from the machines. Meaning, how well their respective artificial intelligence systems are performing, since only an AI system can be expected to assimilate such large amounts of data – or big data, as people call it.

Indeed, how much data they are collecting might itself become a differentiator, although this may not be an indication of anything other than quantity. Meaning, just because you’re collecting many terabytes of data doesn’t mean any of it’s useful.

These technical developments in the direction of computerisation are often collectively referred to as “digitalisation”, and it’s something Hitachi is heavily involved in, according to Ramachander, not least because building digital versions – or “digital avatars” as he calls them – of physical assets is potentially worth billions of dollars.

Ramachander says: “When I talk to the Lumada guys, they say, ‘If you could turn every one of our products which we create – and we create hundreds and hundreds of physical assets – into digital avatars, that would be amazing’.”

See our article on Digital manufacturing for more information about digitalisation.

Use it or lose

For Ramachander, digitalisation is one of the most important trends in large enterprise today. “It’s the key priority,” he says. “You have to be digitising your business to survive in the future.

“And you have to think about not just your own products being more digitally efficient but also how you’re helping your customers enable that.

“So in the case of organisations like ours and GE and others, we’re getting better and better at defining what that means to us.

“We’ve moved from a product business to a solutions business. I’m very clear in my mind that where we focus on digitisation is where have existing leadership in the hard, physical world.

“What I mean by that is, if it’s rail, we have the right to really talk about digital rail because we just understand that whole world, end to end, and the solutions we bring to market off the back of digital rail are real-life solutions that can solve real problems.

“We’re not just sending out concepts and the platforms and saying, ‘Well, you could do this, you can do that, you can do anything’.

“But we’re doing it ourselves first, and then we’re taking it to the market to help clients do it themselves.

“So that’s where we think organisations like ourselves can really help customers. We bring that deep capability around digitisation.

“We’re not only on the top layer of just platforms, but we’re also right there on the hardware layer on the asset level as well.

“Anything that we do, I make sure that we’ve got that entire stack. We’re not just selling that horizontal layer.”

Applied intelligence 

A “stack” is a combination of software and programming languages that an individual or company uses to build some sort of solution. It could be likened to the word “toolbox”, or “toolkit” as Ramachander calls it.

For many businesses now, the development of AI solutions is of paramount importance because, as mentioned, it’s what can analyse all that data and make sense of it.

The better the AI is, the more likely it is that the platform it’s implemented on will be more attractive to potential customers, perhaps because it’s faster or more insightful.

Ramachander says: “We have our own AI capability, but we also use a range of different tools. Part of the Lumada offering is that we also want to use an ecosystem of partners.

“We have some tools that we deploy in Japan – and our research and development guys are deploying a number of AI capabilities there – and then we use some tools ourselves [in this region].

“The thing with AI for us is that it’s not really about the toolkit, it’s really about how we’re applying it.

“Where is it being applied really well?

“So, if you take our MPS AI, used by the video intelligence guys, that’s an AI capability around video – it’s multi-perspective search that’s using AI.

“So that’s using AI to really solve a problem, not using AI for the sake of AI.

“And then we create new solutions. Our data scientists use a range of toolkits, including our own capability.

“We have something called H, which is just one example of our own AI capability.

“We use it quite a lot in Japan, in a lot of things we’re doing around AI, although we haven’t deployed it so much in Europe.

“Any toolkit – it’s how you’re using it that’s important than the tool itself.”

Body systemic 

While AI is, understandably, much discussed, Ramachander says it’s just “one component element of a bigger stack”.

He says: “We don’t look at AI on its own – we look at it as part of an internet of things and big data stack, in terms of what we do.

“So, when I talked about Lumada earlier on, I talked about the four key elements of what we have.

“The first is around using IoT to connect to assets, and then virtualising those assets.

“One of the key things that we do is to create digital avatars – a digital representation of the asset so that we can monitor its behaviour, and capture all the data from it.

“When we take it into the analytics layer, we take in all this external data to create new predictions and forecasts on how that asset is going to actually work.

“What we use AI for is the optimisation element of that asset, or a combination of assets to deliver an outcome.”

Ramachander says Hitachi was looking into the German energy market, and says it’s an illustration of how AI can be used. “Around 40 per cent of energy in Germany comes from renewables.

“So that creates volatility in the grid and also has a negative impact on fossil fuel generation in the existing market.

“So we used IoT to connect all these distributed new energy generation assets, and we used AI as the brain.

“We used advanced analytics to take the data – from weather data, demand data, market trend data, and all this additional data coming into the database, connected to the asset behaviour data.

“Then we used AI on top of it to determine when it’s the best time for you to generate your own energy, when it’s the best time to buy from the market, and when it’s best to trade your energy into the market.

“And that’s how we use AI – it’s essentially the brains, and IoT is the arms and legs of the whole solution.

“So we see it as a tool that unlocks the value around these IoT solutions.”

Control your cell 

When you mention IoT, many may initially or perhaps exclusively think of it as a connectivity and monitoring platform. It can be used for those things, but uppermost in Ramachander’s mind seems to be a proactive approach, to the point that he mainly thinks of IoT as a means to control.

“The IoT enables us to control the device,” he says.

“So you’ve got two different kind of devices out there – you’ve got the intelligent device which is doing a lot of processing by itself at the local level, and you’ve also got the dumb device, which is primarily pushing data in and out.

“But we want to optimise that device.

“At the end of the day, it’s all done on the physical level, your optimisation is coming at the physical level, not at the virtual level.

“So whatever you’re doing with your AI and stack, it’s in order to understand and virtualise the asset itself, understand how it behaves and how it behaves in different conditions, predict and forecast its behaviour – and then optimise its behaviour in the right conditions at the right time.

“So, the IoT aspect is not just about connectivity, it’s also about providing a digital representation of the asset so that we can look at different scenarios.”

The key tool necessary to make best use of that control, suggests Ramachander, is machine learning, a branch of AI which, as the name implies, learns as it analyses the incoming streams of data, as well as from the simulations of various possible scenarios.

“It’s learning continuously from all of that data,” says Ramachander. “So, we’re running and re-running the scenarios until it gets the right kind of scenarios.

“It’s the machine learning aspect of it which is important.

“We can run a million simulations relatively quickly around how you can run that asset, and it can continuously run, so the AI can learn the right kind of scenario.

“Then we can physically manage the asset.

“That’s the great thing about AI. We can virtualise this entire environment, run millions of scenarios overnight, and look at what is the most optimal scenario.”