Some 66% of traffic accidents are caused by distracted driving and the number of deaths on the road is rising in many countries. By combining tech that is readily available with some that will be vital to automated vehicles – video, AI and data analytics – Nauto has succeeded in preventing collisions right now, Annie Turner finds.
Jennifer Haroon, chief operating officer, Nauto, explained how we can make driving safer – and smarter – by using autonomous vehicle tech for drivers at the Move event in London in March.
She said research has found that psychologically speaking, distracted driving is often compared to drunk driving as people understand it’s dangerous, but do it until they suffer consequences. The AAA Foundation in the US noted that distracted driving leads to four times as many crashes when you’re talking on the phone and almost eight times when texting while driving.
Haroon said, “We think one of the reasons people drive distracted is because they think reading a text only takes a couple of seconds. But, it’s more like five and if you’re driving and 55 miles per hour, that is like driving blindfolded across an American football field – but of course, most roads we drive on are not empty.”
Failures to tackle the issue
Laws and regulations are one way of attacking this problem, but enforcement is difficult, rules vary wildly from state to state and country to country, and people often choose not to comply, just as they choose not to wear seat belts.
Automatic emergency braking is another approach to avoiding crashes caused by distracted driving. However, the US Insurance Institute for Highway Safety says it will take until 2045 before 95% of us have automatic emergency braking and, as Haroon says, we just can’t wait that long; the situation is too urgent.
Turn to technology
Some commercial vehicle fleets already use telematics in vehicles so managers can use data to detect some incidents, like hard braking and swerving, although they cannot see the cause of the behaviour, only if it is habitual.
Nauto developed a platform to process information from dual-facing video systems that show the context, inside and outside the vehicle to understand what’s happening. That way fleet managers can see that one driver’s hard braking was a fast reaction to avoid a crazy cyclist, but another had to brake hard because they hadn’t noticed that traffic ahead had slowed.
Haroon noted, “Fleet managers and fleet owners are starting to see that there are some real benefits to technology like this. We can use some of the same technology that goes into autonomous vehicles, computer vision, deep learning and see benefits now: we don’t have to wait for full autonomy.”
Although the tech is the enabler here, the real smarts are about the effort put into understanding human behaviour. We are bombarded constantly by warnings and sign posts on the road while driving and on train journeys it’s hard to get more than three minutes without pre-recorded warnings and instructions, as well as some staff members liking the sound of their own voices. The overall effect is we block them out.
Yet Nauto has no shortage of statistics and case studies to demonstrate its approach works. For example, back in summer 2017, Atlas Financial Holdings, which insures commercial vehicles, signed an agreement to deploy Nauto’s cameras, telematics and data platform across multiple fleets of its taxi, livery and para-transit customers. One fleet insured by Atlas Financial Holdings saw a 35.5% decrease in collisions.
Nauto continued its research into distracted driving, looking at how to identify the tell-tale signs and the best way to get the driver’s attention without driving them mad or making them jump out of their skins.
On that first issue, Haroon explained, “We looked into what types of objects tend to be related to distracted driving. We’d already talked about mobile phones, then we thought what types of movements tend to be part of using a mobile phone when you’re in the car? Well, it’s very difficult to look at your phone down your lap while you’re driving so people look down and either to the left if they’re holding it in the left hand or to the right.
“The other thing people often do is the infamous head bob – looking at the road and texting at the same time. We looked at lots of different ideas, like certain head movements, and even the body position in the cab as an indication that the driver is being distracted.”
As for figuring out the best way to get the driver to refocus and to avoid collisions, or at least their severity, Nauto also drew on research carried out by third parties: for example, the US Department of Transportation tell in-vehicle infotainment system makers that user interfaces shouldn’t distract the driver for more than two seconds as the situation becomes much more dangerous after two seconds’ inattention.
This led Nauto to put a great deal of work into the design of audio alerts. They are already used widely in newer cars, but Nauto concluded people don’t like them and tend to switch them off. A good example is that in the US, most drivers turn off the lane-departure warnings in their brand new cars, because the system isn’t accurate and the alerts have no obvious purpose – people mostly leave lanes on purpose.
Nudge don’t shout – unless you have to
It designed a set of audio alerts that work by voice coaching if something is behavioural and potentially dangerous up to an urgent alarm and instructions when the situation is becoming dangerous due to inattention to get the driver to look up and act.
In pilots with early adopters, Nauto saw a 54% on average decrease in distraction events per hour and one driver had a 70% decrease in distracted events per hour. One of the European fleets the company has been running pilots with has seen a 59% decrease in the severity of collisions: a driver might get into a situation where avoiding a crash is impossible, but the alerts can still prompt the driver to do something that makes the crash less severe.
Haroon claimed, “As we use AI, the more companies use our system, the more we can help coach drivers and the more we can save lives. All the time we’re learning more about human driving behaviour, driving patterns, even things like road infrastructure. We can take that information and develop even more algorithms that can provide different types of coaching and even other types of real-time prevention.”
She concluded, “We’re looking forward to working with more fleets here in Europe this year and beyond. But more importantly, we’re looking forward to develop even more technologies that stop accidents.”
The author of this blog is Annie Turner, editor of IoT Transport