The only backseat driver you don’t want to kick out

Mikael Ljung Aust, one of the leading researchers at Volvo Cars’ Safety Centre, has always been fascinated with the human mind. It’s evident not just from his extensive education – a Masters in theoretical philosophy and religion, another in cognitive science, and a PhD in crash causation and countermeasure evaluation to cap it off – but from the way he speaks. Our interview is peppered with stories of friends, colleagues, family and television shows, with topics ranging from children’s floorball games to Jeeves and Wooster episodes to the definition of luxury. All of these anecdotes hold keys to areas of humanity that captivate and intrigue Mikael, and they’re all told with the aim of illustrating how an understanding of what drives people in life can help him and his team at Volvo Cars figure out how to safely drive them in our cars.

--

Let’s start with the basics. How did you get started at Volvo Cars? Why was a car company looking for someone with your background?

Well, I was looking into my thesis work at Chalmers University in Gothenburg, which is also where Volvo Cars is based. At that time Volvo was just starting a new Accident Investigation Team with a focus on why crashes occur and not just how people are injured once they occur. That project needed a good methodology, so I did my PhD thesis work there by developing a method for accident causation investigation.

I started around 2002; we had a car and an early smartphone system to which the local emergency services sent alerts whenever there was a crash. We’d get there as soon as we could and interview people, trying to piece together what happened and then how to formally express that in a method that you could use to make accidents comparable and see what patterns keep recurring in most crashes.

Things look very different now, but that’s kind of where I've been hanging out ever since.

So you’ve been working on the research and development of the Driver Understanding Systems that’ll be debuting in the new Volvo EX90 being revealed on the 9th of November. Can you tell us a bit about that?

With our new eye-tracking and posture-tracking technology as part of our Driver Understanding System or “DUS” inside the car, you can say that what we’re tracking is whether your attention is on the road ahead or not. It's that simple.

By adding DUS to the equation, we can make the car more attuned to you. At one end of the scale, we can take away warnings or interventions that you can do without, because we can see that you’re in control even though the situation is tight. At the other end, we can boost the ones you really do need. This means that the proportion of times when the car warns you of something you really need help with, as opposed to something you’re already seeing, gets higher. And that builds trust and confidence. It's one of DUS’ main jobs, if not the main job; to help build trust.

Why does it matter whether a driver trusts the car or not?

It's kind of like having a co-driver. If they keep screaming “watch out!” for nothing, your confidence in them goes down, and probably your offers of a ride go down, too. You just ignore them, and that one time they scream for a real reason, you’re pretty likely to ignore them as well. The results could be really, really bad. But if that person is actually on the money, say, eight times out of ten, you do want them in the car. You trust them.

So if our cars get more attuned to you through the DUS you feel more comfortable trusting it, and the chances of you acting on a warning rather than getting irritated increases a lot.

What about when the car doesn’t just beep at you, but takes action?

The funny thing we found in a study we did around ten years ago was that if the car does the right thing, people tend to attribute that action to themselves.

We had six high-end cars in which we were comparing different infotainment systems. All of them had Lane Keeping Aid features, but back then it was pretty much first generation, and people didn’t really know about it.

We [the testers] turned Lane Keeping Aid on in all cars but didn’t tell the test drivers. We had sixty drivers, and almost every single one had an event where the car steered them back into the lane. They weren’t dangerous situations, just the cars steering them back from the side of the lane.

Only one person of the sixty asked about this at the end of the drive.

Because when you feel the steering wheel pull a little, you look back up, the car is headed back into lane. And of course, you believe you're an excellent driver, so it must have been you doing that excellent thing.

When things just work, when you don't have to think about it, that's luxury.

So can we use this driver ego and play into it with the safety features?

Yes! It’s even better if the driver thinks they’re doing it themselves, actually.

It’s about making sure that the way we present these safety features tie into the driver’s normal perception and actions, to sort of boost their senses and abilities when needed. Sort of like a strength enhancing exoskeleton. You still have to decide on what you’re lifting, but this gives you those enhancing powers if needed.

My colleague says about adaptive cruise control, you started off seeing it as just a gimmick, but as you use it more and more, it becomes indispensable. Soon, without it you feel almost like you’re naked in public. Then we know we've created support at just the right subconscious level; you're using it, not thinking about it, and you feel slightly uncomfortable without it. That's the sweet spot. When things just work, when you don’t have to think about it, that’s luxury.

Can you tell me more about that? How is that luxury?

Do you know Jeeves and Wooster? It’s an old comedy show about a gentleman from around the 1920s and his incredibly highly-skilled butler, Jeeves. Wooster turns around thinking that he wants a drink and Jeeves is already holding one for him.

I think that's where we want to be; providing what you need when you need it. To make that happen, we need to be a couple of seconds ahead of you, which means we need to know intent and understand desire. We have a good reading on what's pleasurable, what's actually a good response to this intention that you seem to have. It's a very exclusive thing, it’s a luxury, but it's worth striving for.

You talk about getting a reading off people, and being able to predict them through the Driver Understanding System, but how do you gather the data that’s needed to make these predictions?

It’s crucial to note that absolutely no data leaves the car without the driver’s consent. Even though we have cameras in the car as part of the DUS, it’s a total closed loop, which means that that no actual recording is saved anywhere, and it never leaves the car. For us, it’s not about recording a video, it’s about the fact that a camera is the most accurate way we have of measuring your eye movements and body language. That’s what information gets used by the car to keep you safer.

Now that the cars and their sensors are getting cloud connected, it could be possible to get access to more live data from how people actually drive. In every driver’s case, they would of course have to choose to opt in to having their data collected, and choose what level of data they are willing to share. Without that consent, as I said, we don’t even see their data.

But if they did choose to share, it could mean such exciting learnings. We could track the margin of choices people have in real time in a way that wasn't even remotely possible ten years ago and make constant, game changing safety advancements.

For example, we could study at what distance to an object people tend to brake. And then you add a bit of machine learning and artificial intelligence and clever algorithms on top of that. In many cases you can track down a set of parameters into which many people fall. Some behaviours may be common to people who like to brake way in advance of getting near something. Then another set of behaviours can be common to people braking right at the last minute and so on.

If you can characterise such driver ‘types’, then you can also say: “this person typically would have braked by now”. If the car then takes control and brakes by itself, they'll be alright with that because when they look up and see where they are they will be saying “Oh ****!”

We work with data types, rather than anything about an individual, because quite honestly, personal data isn’t not useful in providing that luxury treatment. If you check into a five-star hotel, the staff doesn’t need to know where you were born or your education to treat you as a valuable customer and predict your needs, because most people have the same needs when in a hotel and behaving professionally toward them does not require a pre-read of their life story. It's the same in traffic. We can be professional and personal but privacy-friendly.

We have to remember though that within those ‘types’, people vary, so there's a fair bit of difference between people. Also, as a person, you can have a good or bad day and that impacts your behaviour, your focus. So being able to better track where your comfort zone boundary is today, and adapting to that, would be the next step, I think.

What are you most excited about in the next five to ten years?

Again, it comes back to trust. I think one thing we might struggle with in the short term is public recognition and acceptance, because some people think we’re trying to act like a super invasive cop or nanny. However, what we’re really doing is taking care of you in a personal way based on your behaviour that day.

So, what I would hope in five to ten years’ time is for the public to say that if someone is driving like they’re drunk, it's not only acceptable but even expected for a car maker to activate safety systems and intervene as needed.

This would be similar to how attitudes have changed towards, for example, helmets when skiing. Nobody used to wear them, but now you feel almost naked without one. It’s a normalised safety feature. I want the same for cars, for people to say “I wouldn't want to travel in a car that can't tell if I'm having a heart attack or can’t tell if I'm falling asleep or losing focus. It would feel like driving a car without a seat belt.”

Technology is only beneficial if people see the use in it and feel safe using it.