Can you be nudged to look the other way?
MeBeSafe has developed a nudge in the car to increase the driver’s attention when cyclists approach. The nudge has now been built into a vehicle – but can we make sure that it is really working?
A driver approaches a dangerous intersection with blocked view. There may or may not be cyclists hidden somewhere out of sight, who could run out just in front of the car. MeBeSafe wants to make the driver more attentive to this threat by lighting up a nudge in the car.
A stylized image of an intersection will show up, with an indent showing from which direction an unseen cyclist is most likely to approach. This information will be based on historical data for now, but MeBeSafe is also developing the technology necessary to spot cyclists in real time.
This seems really good in theory – but will it work in real life? Could such a simple nudge really help to refocus the attention of the driver? A simulator study made by CRF showed promising results – but now it is time for the real roads.
For the nudge to be a success – it must work no matter who’s being subjected to it. Therefore, a good mix of people with different gender, age and background are being recruited to come and drive the car as a part of the Field Trial. And the first challenge starts already at this point. Should the drivers know what they are about to test? Olaf Op den Camp from TNO has not yet decided upon this.
“If the drivers are naïve, we will capture their actual first-time reactions, which provides a fair comparison with the previous simulator study. But if they know about it, maybe we could simulate how they would behave in the long run when they are used to the nudge? I’m however leaning towards the first option.”
The nudge should work particularly in intersections where the view of crossing traffic is blocked. It therefore makes sense to test it in an area with a lot of such situations. And fortunately, there are perfect areas for this in the Netherlands.
“A lot of people moved to Eindhoven hundred years ago, so huge areas of small houses were built – very close to the streets to save space. Views around the corners are blocked all the time, so this is an ideal place to test the nudge” Olaf asserts.
The drivers will be taking a trip around this tight city, while the researchers observe their behaviour. Normally, you would measure speed or braking in order to assess how safe you drive and if collisions can be avoided. But speed or braking is not the target of this nudge – it aims to make an impact already in directing the attention. And one way to asses this is to actually measure where people look – by their gaze.
“Of course people shift their gaze all the time, but if we can move the overall distribution of gaze towards the side that we nudged for, it seems that we have succeeded”, Olaf explains.
But what if the drivers are cautious good drivers and choose to look in that direction on their own most of the time? How can we know if it was them or the nudge that was responsible for the gaze patterns? One way to find out is to use so-called false positives.
“The false positive is a very powerful tool. It means we light up the nudge out in the open where there’s no reason to look any extra in a particular direction. If they still look more towards that direction, we know that it was the nudge that made them do it”, Olaf op den Camp proclaims. And if the nudge will work to redirect attention, it will then make people look at the right spot.
Behind this nudge lies the knowledge of which intersections are dangerous, which is built into a hazard model that the nudge makes use of. Of course, it is not feasible to model this for each single intersection around the world. Traffic is moreover very dynamic, and no two situations are alike.
That’s when a dynamic model developed by Cygnify and TNO comes in. It contains equipment to scan the surroundings in real time, detect cyclists and predict if they are about to cross in front of the car. This will however take a few more months to complete and is expected to be tested in a FIAT car after that.
But the principle is very much the same. If people are nudged by the rather simple static model, they will also be nudged by a more sophisticated future model. And this model is not only good for nudging – it could also play a crucial role in a more automated future.
“This model can not only be used to determine when to nudge a driver. It could also form the basis for how a self-driving car becomes aware of cyclists –and then influence which decision it will make” Olaf proudly states.
So in the end, this trial could actually lead to two potential applications. Both the obvious of nudging drivers to change their gaze pattern and detect cyclists. But also as a future way not to nudge you – but to nudge your car.