Google’s self-driving cars are probably better at three-point turns than you

 Arrow on tarmac depicting turn in road

In the latest monthly report from its self-driving car programme, Google says that its autonomous fleet has been focused on one particular manoeuvre recently – mastering the three-point turn.

“The ability to navigate elegantly around tight spots is essential to building a driver that can get you from door-to-door,” says Google.

Human drivers often struggle with three-point turns as they “tentatively inch forward and backward, in tight spaces, without a full view of the road”, noted the report. 

The company's self-driving cars, on the other hand, “can see a full 360 degrees, measure distance down to a few centimeters, and precisely calculate the quickest path for the car”, says Google.

Self-driving cars may help to avoid situations like this

Its self-driving fleet regularly practices its three-point turns, carrying out around 1,000 of them on city streets every week. 

However, one thing is holding the cars back from executing these turns too efficiently: human behaviour. Autonomous and human drivers approach the manoeuvre very differently, it turns out.



“Our challenge is to teach our self-driving cars to choose the option that’s not only the quickest, but one that feels natural to passengers,” says Google.

In order to make their turns feel more natural, Google has programmed its self-driving cars to mimic how people currently drive, opting for wider forward arcs instead of a series of short back-and-forth movements. 

Our goal is to a develop a fully self-driving car that can handle every part of driving, and that means teaching our car to handle advanced maneuvers like these multi-point turns.
— Google self-driving car project monthly report – October 2016

The sensors on its self-driving vehicles would actually allow them to perform most of a three-point turn in reverse, says Google.

But human drivers prefer to drive forward where they have better visibility of the road, so its cars will reflect this behaviour. 

In the race to develop fully driverless cars, this is unlikely to be the last time we see a trade-off between the most efficient way of carrying out driving tasks and established human behaviour.

And when it comes to one of the trickier driving manoeuvres, it seems that – for now at least – the established human approach is leading the way.


2,230,275
The number of miles driven autonomously by Google’s self-driving car fleet to date.