macrovector - Fotolia

At IBM’s Watson lab, customers marry the power of AI with the IoT

IBM formally inaugurated its Watson Internet of Things headquarters in Munich, Germany in February, and Computer Weekly went to meet the customers and partners who are using Watson AI to deliver on the promise of the IoT

This article can also be found in the Premium Editorial Download: Computer Weekly: Google power – how the internet giant builds datacentres

At about lunchtime on an unseasonably warm February day, a small commercial drone hovered alongside Highlight Tower; a striking, angular glass block soaring 126m over a suburban Autobahn on the outskirts of Munich, with equally striking views.

Inside the building, Harriet Green, general manager of Watson Internet of Things (IoT) at IBM, and Ilse Aigner, deputy prime minister of the state of Bavaria, linked hands over a big red button, watching the video pictures relayed by the drone.

As it climbed to the very top of the tower, the familiar, linear IBM logo swam lazily into view, and to applause from their audience, Green and Aigner pushed the button, lighting up the sign and officially inaugurating the IBM Watson IoT Headquarters. Although, since it was midday and sunny, it is unlikely many people noticed it at first.

It is a little over a year since IBM first announced its intention to locate the global base for its Watson IoT platform in Munich, and since then it has been very busy, not only fitting out its new home, but advancing IoT technology and drawing businesses such as automaker BMW, outsourcing and facilities practice ISS, and ICT channel distributor Avnet to its banner.

At the inaugural event, which saw customers, collaborators and journalists descend on Munich en masse to find out more about what the IoT can deliver for enterprises, IBM announced a new round of enterprise partnerships at the Watson IoT lab, described by Green in her opening remarks as a “collaboratory”.

It would probably be fair to say that, despite its long history, IBM is not renowned for its expertise in networking, and nor has it sought to be. But it doesn’t need to, for what IBM is bringing to the picture is not connectivity, or a desire to push a proprietary network standard, but the growing power of Watson.

IBM wants to offer a whole range of offerings, capabilities and partnerships to extend the power of cognitive computing (so memorably demonstrated by Watson on US quiz show Jeopardy) to the IoT via the Watson IoT Cloud.

Read more about IBM Watson

Andy Stanford-Clark, IBM distinguished engineer and a key figure in the Watson IoT project, argues that while there have indeed been many rushed IoT projects that pushed to market the minimum viable product – often riddled with massive security holes – and got the IoT a bad name. Therefore it is important to take the time to get the IoT right, which is where IBM comes in.

The whole point of Big Blue’s involvement is not that it can’t resist trivial use cases – such as a Red Dwarf-style talking toaster that can “print” the weather forecast on your toast, or any number of silly applications – but rather that if one of its partners sees a business opportunity and a monetisable stream for selling something as a service, then IBM will have their back, according to Stanford-Clark.

“The Watson tools like speech enablement, conversational understanding, image processing, whatever API [application programming interface] it is gives them the building block to make that product,” says Stanford-Clark. “We’re not sitting here saying, yes you can do it but if you want to do it, we’re here to help.”

Gabi Zodik, IBM Research director of IoT and mobile platforms, says: “One way to think about the IoT is not just as an isolated thing, but as a gateway to cognitive systems. How do we access cognitive systems? Through the IoT devices that are around us all day, so that’s the shift. The IoT is not out there on its own, it’s in cloud and cognitive.”

SNCF plots a path to the cognitive IoT

Customers flocking to the IoT lab include Dutch drone maker Aerialtronics, industrial hardware supplier Bosch, entrepreneur funding platform Indiegogo, Finnish lift and escalator builder Kone, office technology supplier Ricoh, French state railway operator SNCF, and financial services giant Visa.

Public transport has been at the forefront of use cases for the IoT pretty much since the term IoT was first coined, so for SNCF’s CTO, Raphael Viard, the concept of connecting anything and everything he can is not new. Indeed, we have previously reported on its efforts in this area.

Sitting down over coffee with Computer Weekly, Viard explains how SNCF is using Watson IoT on the IBM Cloud to address its biggest challenge, maintenance.

Like other rail operators, SNCF suffers from two big problems: its maintenance depots are generally located on constrained sites near city centre termini; and its trains are increasingly sophisticated and carry more and more components. All this adds up to means that its maintenance windows are far longer than they used to be, and with no room to grow, its support infrastructure is at capacity.

On its most heavily trafficked Parisian commuter routes, SNCF is outfitting new generation trains with 2,000 sensors that forward 70,000 data points every month, so its engineers can remotely monitor the stock to get early warning of issues with doors or air conditioning, for example, freeing up their maintenance capacity.

Improving communication around delays

For passengers, says Viard, the harvested data can be used to improve its communication around delays and helping them plan around any problems on the rail network. He is also looking to use the IoT to monitor lineside trees.

“We are doing a lot with big data for vegetation, trying to analyse where we may have issues due to high winds and trees falling on the catenaries [the overhead power supply], and so on, so we’re doing a lot of experimentation around that,” he says.

Right now, SNCF has not started to use Watson’s cognitive capabilities, but Viard is keenly observing developments and coming up with use cases for the railways.

“In the first phase we are looking much more at analytics and big data, but when we have done this, cognitive may be involved,” he says. “In addition when you do cognitive you need to have all of your data in Watson, and right now we don’t, we just have our IoT data there.

“We are gathering data and when we have enough to analyse in that way we’ll look more at cognitive, but the challenge for us at this time is trying to gather more data.”

For SNCF, cognitive computing will likely find a strong use case in passenger facing areas, according to IBM general manager of global technology services Remi Lassiaille. “You could imagine cognitive in the CRM sphere to manage customer relationships, and maybe also in the call centre,” he says.

EEBus uses cognitive to improve energy efficiency

SNCF may be at a relatively early stage in its use of cognitive computing, but for EEBus, a major new pan-European IoT alliance initiated by the German government to explore initiatives in the energy industry, the benefits are already writ large.

The members of EEBus, which besides IBM itself include leading connected home stakeholders such as Bosch, Miele, Schneider and Vaillant, are exploring a standardised and common language for an interoperability of devices that support communication between, for example, smart heating systems, electric vehicles, solar panels and household appliances.

“Our solutions and algorithms learn from the consumer side, and connect to cognitive systems to combine the knowledge and experience from many homes to improve the user experience, because we can learn very quickly about the user and what he or she needs,” says Sven Schreiber, executive vice-president of residential business at EEBus member SMA Solar Technology, and also a member of the initiative’s executive board.

For Schreiber, it is vitally important that future smart energy systems are able to run as unobtrusively as possible, so that they become essentially a background component of people’s lives.

“At the end of the day there should be no loss of comfort for the end user, and they don’t want to have to go back to university to be able to run a domestic system, so it must be an intuitive system, hence there is big potential to connect the intelligence of Watson,” he says.

Bernd Wunderlich, IBM Watson IoT industry leader for Europe, says IBM’s involvement in EEBus will accelerate the use of the IoT in the energy industry.

“We have one showcase that we have built already for a heat pump, where we have developed a minimum viable product in just a few weeks, which shows how we are bringing external APIs into the game, such as from the Weather Company [a recent IBM acquisition] to make such systems much more intelligent,” he says.

Cognitive keeps users happy

Like Viard at SNCF, Hugues Delval, executive vice-president of services at Kone, a leader in escalator and lift technology, is already using IoT sensors to monitor the health of the firm’s installed base of equipment, checking whether or not a lift car is levelling correctly, or noting abnormal vibrations, alarm system function, and so on.

For Kone, which makes half its €8.8bn annual revenues from recurring services contracts, bringing cognitive analytics into play to extrapolate trends in its datasets that may indicate an oncoming problem is a key differentiator for its businesses.

“We are able to tell our technicians that this elevator’s doors are vibrating abnormally, it’s probably coming from the door lock, so please make an intervention in the next 48 hours and make sure you have this component in your van,” Delval tells Computer Weekly.

“This enables us to guarantee the availability and performance of the equipment, and for customers it is also very important. We can provide the information to the customer on his mobile so that when he, as a tenant of the building, calls Kone to say he has noticed this issue, we can say we’re already intervening.

“It brings value for us in terms of our ability to serve our customers in terms of performance, safety and availability, and it also helps us make our people look more professional, because they come with the right information,” says Delval.

In the future, Kone hopes to expand its use of Watson artificial intelligence (AI) to introduce services that will enhance the user experience still further. In years to come, Delval predicts that lifts in apartment blocks will recognise mobile device owners when they walk through the front door, and knowing that said device is associated with a property on the eighth floor, will have a car ready to whisk the resident home without them having to wait for it.

More prosaically, Kone hopes to use Watson to analyse inbound calls to its contact centres to help its agents ask the right questions and find the potential root cause of a problem quicker. Such a system may also be used to identify how the caller is feeling – angry, calm, and so on – which the contact centre agent can then use to tailor his or her approach and line of questioning.

“We will augment people’s capabilities to better serve our customers, that’s a key element for us,” says Delval. “We are entering a period where we can really tap into the cognitive analytics capabilities to design these new services, and the further we go down this road the more cognitive and AI capabilities will go with us.”

A guided tour of the IoT

Visitors to Watson IoT’s new home can be left in no doubt as to how lavishly IBM has spent on the experience. While other suppliers tend to demonstrate small-scale use cases, often involving model cars and cuddly toys, IBM pulled out all the stops to impress the press corps, enlisting some of its top technologists, such as electrical engineering veteran John Cohn, to run guided tours of the facility.

At the client experience centre, IoT explorers can peruse every single connected component of BMW’s i8 electric supercar, hear music generated by real time IoT data flows from the trains on Munich’s U-bahn network, and see case studies projected onto a digital tabletop.

Elsewhere, a clearly excited Cohn demonstrated some of the IoT capabilities IBM has built into its offices, using a text-to-speech interface to book a hotdesk, while overhead, lights lit up to guide him to it.

Meanwhile, strategically placed mood lighting lit up in shades of red and green to let busy IBMers know how long the queue at the downstairs coffee bar was getting – it is illegal in Germany to show anybody’s face over a webcam without their consent, but not to display the fact of their presence in the space, so a network of infrared IoT sensors reads the ambient temperatures in the room to see how many people are there, and changes the lights to achieve the same effect.

In other parts of the building, IBM engineers used Watson cognitive computing and the IoT to convert speech from English text to Chinese calligraphy written by a robotic arm, and showed off Olli, a US-built electric vehicle that uses Watson IoT not only to analyse and learn from high volumes of transportation data, but also interact with its passengers using Watson developer APIs: speech to text, natural language classifier, entity extraction and text to speech.

Read more on Internet of Things (IoT)