The Machines Talk: Kone's machine learning and IoT journey

Kone Machines Talk

By Trevor Clarke, TRA Director

You know that weird feeling when you step onto an escalator or travellator thinking it is, but it isn’t moving? The one where your muscle memory tells you one thing, but the fact it isn’t moving makes you do a mini jolt or stumble? Kind of embarrassing and awkward. Kone wants to help you avoid it.

How? By empowering its customers and engineers better leveraging IoT sensors and machine learning.

Unlike airlines, shipping, automobile manufacturers, or public transportation providers, Kone does it with escalators and elevators.

The company is, what I would term, an excellent example of an enterprise in an operational technology domain going digital. A pursuer of OT and IT convergence, of a sort.

You see, in most places that have escalators, travellators or lifts, the building managers often only found out they weren’t moving when one of their staff noticed or a customer complained – machines weren’t connected for the most part so notifications were analog. The subsequent action to rectify the situation could only ever be reactive - the customer frustration was already felt.

In some instances, a machinery failure or stoppage, can pose a risk, especially in high traffic areas or elevators.

This was a challenge that Kone identified and is attempting to address with 24/7 services it launched in February this year. Services that combine internet of things (IoT) sensors, with machine learning (on IBM Watson) and data analytics.

“Our aim is to use these technologies by combining or connecting our service base, where we have 1.1 million elevators and escalators and then provide better maintenance services with these technologies and also new people flow solutions,” said Samu Salmelin, Head of Services and Solutions R&D, KONE.

“It is 24/7 connected services. What we are providing there with for our customers is really better predictability for their buildings, for their assets. We are providing them better service because we are able to take actions before equipment fails. We are able to be much more on top of the things on behalf of our customers. So if you think of the current way that we often operate, it's actually customer or users who identify a performance gap, a noisy elevator, something that might feel unsafe or even a malfunction. And now with better connectivity, and by using machine learning, implementing algorithms that are running in the cloud, we are able to predict the failures and we are also of course able to act faster according then to the service contracts we have with our customers, with the new technologies.”

One thing Kone has not done as it has rolled out its services is replace staff with technology; one of the big fears of robots and AI. Previously, Kone maintenance staff would rely on their professional skills, intuition, experience and analysis of the issue at hand. The challenge with this, is not that Kone staff couldn’t do the job – it’s that it was always reactive.

“Now we are able to provide them really actionable information. Instead of them taking on preventative maintenance tasks, they are really able to do predictive maintenance, predictive repairs, and also focus the maintenance efforts to the areas where we have seen the condition deteriorate,” Salmelin said. “It can be the condition of a piece of wearing parts, or it can be a condition that is by the customer feeling for example, vibrations in the elevator, and then we are able to have recommendations to our technicians.”

To help in the predictive analytics side Kone has connected a lot of its machinery to the internet with a range of sensors. You can see an example of it at a campaign website here. The site lets anyone listen to a narration of the movement of an elevator and the sensors it is hosting sending its position and status to the Kone cloud. From Shanghai to France voyeurs and elevator tragics (is there such a thing?) can get their kicks out of monitoring a host of locations.

This of course is only part of the story. Using machine learning the company is now able to predict failure of components along with getting a better set of data on their customers. For Kone this is key – no two customers are really the same. To offer a more compelling service and achieve its own goals , Kone needs to tailor services to individual clients.

“I would obviously start from the customer, and thinking about how the technology has really enabled to renew your business model and the ways of working. I think there are wonderful new technologies out there, but you have to really think if it is for your business. At the same time you have to be visionary. You have to be able to do things which are uncertain that have not been proven yet,” Salmelin said.

“But, keeping the customer and the business in the, center, especially the customer value -- that's the key. And then the technology, they are tools -- means -- to achieve things. I think that's the key. And easily you can maybe fall into that where you get fascinated by the technology, but you still need to make it serve your business.

“The technology is, in a way, the easy part. But how do you really make it serve the business? How do you then take the people on board and think about -- how does it really enhance your people? Because our business is people-business and we have to really help our people to serve our customers, and then this technology has to enable -- make it easier than what it has been before and give you more insights to do the work. I think that those are after all, the essential things.”