DRIVERLESS ED

TEACHING CARS TO DRIVE THEMSELVES IS A LOT LIKE TEACHING PEOPLE.

FOR MANY TEENS, driver’s education class is a rite of passage. It’s the first step toward passing a test, scoring a license, and feeling truly independent behind the wheel. If driverless vehicles catch on, will cars have to take driver’s ed too?

That’s a joke with some truth to it. Because to succeed, a driverless vehicle (or more precisely, that vehicle’s onboard computer) will need to learn all skills that, until now, only human drivers have known.

A Lot to Cover

Onboard computers have to learn the rules of the road, plus all the skill and judgement human motorists use when driving. That includes everything from accelerating and changing lanes to parking and sharing the road with bikes. Sure, computers are “smarter” than people in many ways. But this won’t be easy. If you had to write it all down, even simple driving tasks would require long, detailed instructions. Yet that’s exactly what programming a driverless car will require.

How can we do it? Think science. Most of the actions that we call “driving” are based on physics and other scientific disciplines. So, in a way, driver’s ed is really a special kind of science class—where practicing driving is the lab! And we can use what we know about science to teach vehicles to drive themselves. Let’s test how that could work, using a classic driving situation.

Brake Time

You’re driving along, and the car ahead of you comes to a sudden stop. Now you need to stop your car at a safe distance behind the other vehicle to avoid an accident. How does a human driver accomplish this? Let’s examine the two most important factors.