NXP and insurance tech firm Moter are planning to change the face of risk-based car insurance. Steve Rogerson discusses the implications.

How much do you want other people to know what is going on in your car? If the answer is not a lot, then maybe you should be a little bit worried about the many sensors that are being deployed in vehicles not just to monitor what is happening on the road outside the car but what is going on in the cabin too.

OK, we are not talking a Paradise by the Dashboard Light scenario here, or are we? The idea is that cameras and sensors inside cars can detect what the driver is doing, whether they are distracted, on the phone and so on. Now, the logic behind this is to check after an accident what the driver was doing in the seconds before the crash happened.

To do that, the sensors and cameras have to be monitoring the driver, and maybe the passengers, all the time, and that is a lot of data, which is one reason privacy worries have not really come to a head because who is going to plough through all those data. Well, they might not have to.

I was talking last week with Michael Fischer, chief digital officer at insurance tech firm Moter, and Brian Carlson, global marketing director at NXP Semiconductors. These two companies have put their heads together to use artificial intelligence and machine learning to analyse these streams at the edge, the edge in this case being in the car.

The interest from the insurance industry of course is that using such data means they can build up a profile of how good or bad a driver someone may be and how often they are distracted while driving. Getting such information would be a nightmare if all those streams had to be uploaded to the cloud for processing. By doing this at the edge, they can get information such as how often a driver uses the phone or looks out the side window, or even sees paradise. Then, only the results are passed on for analysis.

Insurance companies love the idea because it means they can reward good drivers with lower premiums and, possibly, punish bad drivers with higher premiums. However, privacy rules mean the driver is going to have to agree to this, and how many bad drivers are going to do that?

Fischer isn’t worried, as he believes most people are decent drivers and would be up for it if it meant cheaper insurance. And those who do not opt in would eventually pay more as they would not be entitled to the discounts. And, thinking about it, most bad drivers think they are good drivers anyway.

Those who may not have a choice about this are those who work as drivers, because fleet operators are very keen to keep a stronger eye on their employees. The up-front message is it will mean they can offer training to drivers who are being too aggressive or making mistakes, but there is of course the sneaky suspicion that this is a Big Brother move and employers are more interested in making sure their employees are doing what they should be doing and not taking unnecessary breaks.

Carlson says by using machine learning, the video processing can be reduced even further as recording can be set tit to trigger on certain events, such as mobile phone use. This can also be combined with existing technology that is being rolled out that monitors the driver for signs of drowsiness. Fischer even showed me a video example of it triggering on a driver tailgating.

And for those who think some of this is still a long way off, Carlson said the service-oriented gateway technology that allows this edge processing to happen will be going into vehicles later this year and early next year. So enjoy paradise while you still can.