Time to trust the robot’s judgement?

TU-Editor

30th June 2017

A driver’s ego is a fragile thing and understanding how UBI can affect it is one of the biggest challenges providers face in their bid to widen the technology’s appeal.

Just as most motorists believe they are good drivers, it can come as a nasty surprise when a computer program tells them they are, in fact, less skilled than they believe.

So TU-Automotive used the example of how the technology would treat an experienced sporting driver’s habit of ‘rear-wheel steering’ though a bend when we caught up with Colin Smithers, founder and CEO of Redtail Telematics at Connected Car Insurance Europe 2017.

His company specialises in producing telematics devices that use a combination of GPS, accelerometer, gyroscope and cellular technology installed in vehicles to determine how well a driver is performing behind the wheel.

So how can the box distinguish between a car out of control or one purposefully put into a slide, where legal, to achieve optimum speed and sight lines?

Smithers said: “We started from the point of listing the major causes of accidents and then we let the machine work it out itself to see if these conditions are replicated. We look at the people who have the accidents, see how they driver and then say ‘well that’s not good’. Of course the opposite is true of drivers who don’t have accidents. So we are empirically deriving best driving practice. The more people let us process their data, the more accurate the system will become.”

The systems rely on a mathematic model of data and Smithers believes it, ultimately, has the potential to make complex judgements of what looks like good driving in the real world.

He explained: “The algorithms can tell if you have someone who is very inexperienced and, as an instructor, you wouldn’t teach a beginner to go side-ways round a corner. So there is no reason why our algorithms can’t look for junior errors and say ‘look this guy is not up to going sideways round a corner’ versus someone else who is.

“I work on the premise that if you were sat in the back of a car blindfolded could you tell bad driving or a beginner’s driving, just from your own senses? I’m sure you could and because you can, our box can. We can measure so much with our six dimensional accelerometer, three linear and three rotational, that there is not a lot more we need to know beyond what does good driving look like.”

Smithers said one of the biggest obstacles to spreading data collection services comes from the carmakers themselves. He said: “I could see this technology fitted to new cars and, it’s arguable, it’s already there. Airbags, for instance, know much of this stuff already but it’s just that their data is not being made available. It needs a mindset change in the OEMs for this data to be made available. Certainly, eCall will have all the technology to do this stuff it’s just that the OEMs have to get to the point where they want to do it and that’s just about having the will.”

The will, Smithers argues, will come from carmakers understanding how the data driven revenue streams can work to their advantage. “Like all telematics, no one use will pay for it, so OEMs have to find multiple uses for the data for it to make sense and they have to go through every individual strand and ask ‘what do I do with this information and who can I sell it to?’

“If they know everything about location and driving style, they are starting to know about lifestyle so they can start figuring out who you are, where you’re going and this is valuable and they have to find ways of selling it. Is an automotive OEM good at that? Not yet. Ford’s involvement with Verisk is the way forward for them and I think many more OEMs will buy the expertise and stop other people getting it.”

[Tele.Myles.2017.06.28]