Monetizing the data that’s always been there

Brad Surak, Chief Product and Strategy Officer, Hitachi Vantara, talks about a wide range of issues including Industry 4.0, Artificial Intelligence, 5G and their tie-up with Disney theme park rides.

On the latest industry buzz words and solutions…

Industry 4.0, cloud and Digital Transformation are all overloaded terms. We talk about the outcomes that we've seen in our own operations and focus on very pragmatic step by step outcomes. We don't believe there's any kind of silver bullet to solve these problems. It's an evolution, a continuous improvement, a rollout. Like, IoT is not a product. It's not a market category. It's an architectural disruption, the same way cloud is. We look at it from an engineering standpoint. IoT is about blending the Edge into the IT enterprise architecture. We see that Edge as this continuum and different customers have different needs.

We try to look for architectural patterns that help solve problems. If a customer has a different way of looking at their operation, we can adapt to that as well, but we'll come in with a point of view. Our customers have very old systems and some newer systems and so we have to adapt it always. But the same kind of architecture can address a number of different problems for the customer. They can get the most return on their investment on that technology.

It's not a product solution where you go and implement a product for a customer that fixes everything. There's a systems integration component to it. Rather than try to solve a point solution, oftentimes we're brought in to solve a point problem. We try to broaden it so that the systems we put in place address a wide number of problems for the customer. Most of our customers want to hear about visionary outcomes, but they want to focus on the conservative and pragmatic outcomes.

Industry 4.0 and Artificial Intelligence

One of the reasons why I came to Hitachi was this construct around social innovation. Social, environmental and economic value is what we think about. It's drawn from the Japanese culture and is a social contract with the workforce. We think more deeply about the responsible application of technology. We put AI systems in place in our factories for safety purposes. We can detect if the workers are wearing a hard hat and are they supposed to? Are they getting too close to a piece of equipment? Video analytics and AI can identify a movement that may lead to a repetitive stress injury. We talked about this happiness study. IoT and advanced analytics was used to see how people respond to certain environmental factors.

We employ robotics to take away either the monotony of the job or if it's related to safety. How do you evolve the capabilities of the workforce as you're automating out some of the activities that people don't get a lot of joy out of? We have some very advanced solutions in a factory in Japan that manufacturers our storage arrays. It’s not a lights out factory, but a lights out warehouse and storage management solution. We use robotics to go and pick all the right parts loaded on to robotic carts that drive it to the right place. We do studies of where the bottlenecks are in a in a process and then we come out with very advanced automation solutions that get a high ROI.

But we have to mind is the bias that gets built into the algorithms from the training data. This is the problem everyone is starting at. It's like raising a child. It's not programming in a traditional sense. It's delivering a set of experiences to an algorithm that optimises around that set of experiences. If you deliver the wrong set of training data, you're going to introduce bias into the judgement of the AI and that can have negative consequences.

So, how do you become a digital industrial company? That's a mixture of the traditional skills in manufacturing, with new contemporary digital skills. It’s creating a workforce that's more connected and more fluent with both sides of the technology stack. How do we bridge that gap between in skill sets of our employees, between operations technology and information technology? Once you have workers that understand both sides, you've got a differentiator. Then you’ll find that innovation is going to come daily from the operators in the factory. If you're blending the skills between industrial and digital, you're now opening up this new way to create a whole lot more value, by selling the capabilities in your company in a different way.

The data economy and digital twins

It's about monetizing the data that has always been there, but you've never maybe captured or leveraged it as a business model. We also do elevators and escalators. So you could take the sensor data from an escalator and you could figure out the weight of the people and how many people are riding at any one time. You can then monetize that as kind of people flow and you can sell that to a shopping mall operator who might change the rents on their stores based on how many people are flowing past the door. There are different types of sensors from vibration sensors and pressure sensors to temperature sensors. These are all new business models that traditional companies now have available to them. It's not just software companies that are doing this. They led the way with these models, but it's now impacting other industries.

Then there’s the digital twin. When you're designing a piece of industrial equipment, first you have to use a CAD programme to model it. You can simulate the physics of the piece of equipment in that type of a programme so that you can understand how it will perform. You can actually create almost infinite sensors virtually inside of those models. Then you can use these digital representations and simulations and then extrapolate sensors in other parts of the equipment, use those to train your algorithm.

You can synthesise data from the simulations and use that to make your algorithms more accurate. You can actually simulate many more years of operation of that equipment in high performance computing environment. Where do I have to put physical sensors and what can I what problem can I solve from the physical sensors that exist so I don't have to go back and retrofit the equipment with those sensors and how can I leverage simulations digital twins to extrapolate.

5G, latency and innovation

In the US and Europe, we're seeing some 5G corridors, which are enabled first. Speed is one thing, but it’s about the latency. If you think about some of the applications in the industrial world where you have an existing factory, it’s very expensive to run wires and connect those machines, to collect the data from all those sensors and bring it all together and analyse it. Wireless connectivity is a great option.

Where latency becomes important is when you need real time feedback to an operator of a piece of equipment so they could change how they're operating it. With a machine, you can start to optimise an operation based on the sensor data and the readings. You can look at the quality of the finished goods and then change how the machine operates. These kinds of control inputs become much more possible with 5G. It’s really that network and compute layer that's allowing this to move applications that had to be sitting right on the machine. Now you have the ability to use more powerful analytics in a cloud type environment, because you have that communication, or at least speed reliability. 5G is going to open up a whole new frontier of innovation.

On VSP 5000, the world’s fastest storage array

It's actually the fastest, the most reliable and the most scalable. The teams in Japan really did a great job of coming up with this next generation architecture for the storage array. Always the competition is around speed and reliability. But we're also thinking about how does it participate in this architecture that borders the lines between the cloud and the Edge? There's a lot of capabilities that we have planned that can extend it so that it can operate in a cloud aware way. It can be able to extend some of the capabilities that we've developed some of that performance to some of the older infrastructure. Companies today are investing in the cloud. They're trying to keep their data centre technology longer and sweat their assets. Our solution will help them get better performance out of their legacy assets.

The partnership with Disney

We've sold storage to Disney and Hollywood for a long time. We've started working with the part of Disney that designs, builds and operates the attractions, the rides. Now you can think of Disney as a rail company because of all the rides and roller coasters on rails. So the physics of analysing the performance of those, predicting what maintenance is required, all of that is very similar to bullet trains and rail networks. We have a lot of expertise in that area.

We help Disney analyse the data coming off the rides. They've got a very complex set of technologies. Everything from the original rides on Pinocchio with copper wires to the new Star Wars attraction which is part video game part physical ride; it’s a wide range of different technologies to pull data from. They are also extremely good in simulation. Disney is world class at the idea of the digital twin.

The problem that we help them solve was how to go analyse the data from these rides and predict when they needed maintenance. You don't like it when you buy a ticket to the park and it's down. They can open a ride earlier in the day and can charge a little bit more, called extra magic hours. That really was the beginnings of this. We're the official data analytics provider for the attractions. We're deploying our technology across their various parks. We manufacture the monorail for the Tokyo Disney. It's eventually going to become a platform for ride analytics. Each of these rides is created custom: There's only a one of a kind. They're very different.