Driverless Cars?

Published in IEEE Spectrum Magazine, July 2014

Back in the 1960s the television show Candid Camera had several skits where they faked driverless cars on unsuspecting people.  We thought it was funny, because everyone knew that driverless cars were impossible.  That was then, but now the great technical and social challenge of developing driverless cars has suddenly has been opened up to us.  (see Spectrum articles in May issue)

At the turn of the millennium I was a member of a committee of the National Academy to select the most outstanding engineering achievements of the last century.  First on our list was electrification, followed by the automobile as number two.  An important criterion was social impact, and we reasoned that the development of the automobile profoundly changed where we lived and how we lived.   While this was a retrospective list, at about the same time people were making lists of important future achievements for the new century -- grand challenges with social impacts.  But as far as I know, driverless cars were not on any of those lists.

The Model T Ford was introduced more than a century ago.  It is amazing that automobiles have changed so little over that century.  The basic function is still about the same, and some of those Model Ts are still on the road today.  But now we have the chance to really make a difference.  The technology is ripe, but it won't be easy.

Someone who had the opportunity to sit in the back seat of a driverless car on a freeway in San Francisco described his experience as initially terrifying, but later simply boring.  I was thinking about that description as I watched television ads for new cars.  The ads featured cars that could climb mountains, ford streams, and had so much power that they could go from zero to sixty miles an hour in almost nothing flat.  Moreover, they were fun to drive and sleek in appearance.  I imagined instead an ad for a new driverless car stressing how boring the car is.  I was wondering: are people really going to want these cars?

Every driver has an individual driving style and persona.  I'm comfortable when I'm being driven by some people and uncomfortable with others.  As a driver myself, I don't like being stuck behind a slow driver, nor do I like being pushed by an impatient faster driver behind me.  I'm wondering: will the driverless cars have an adjustable comfort level of speed and aggression?

Group behavior in traffic is a complex phenomenon.  Some drivers believe this is a zero-sum game, while others believe that courtesy helps everyone.  A classic case is a long line at an exit ramp, where the line moves slowly because other drivers speed by the line and cut in at the front.  Will the driverless car be tempted into such behavior, or will it be unfailingly polite?

Then there are the classic routing decisions.  Heading into New York City there is a bulletin that the Lincoln Tunnel is jammed, while at the Holland Tunnel traffic is moving well.  Hearing this, many drivers will nevertheless go to the Lincoln Tunnel in the belief that everyone else will go the Holland Tunnel.  Enough drivers are siphoned off that both tunnels even out.  But will all the driverless cars use identical routing algorithms and jam the tunnel that was said to be good?

Will the driverless cars know when it might be ok to break a law, like crossing a yellow center line to avoid an object?  Can they exceed the speed limit while passing another vehicle?  And if there is a traffic violation, who gets the ticket?  In the event of an accident, who gets sued?  And there are those ethical decisions that sometimes must be made almost instantaneously -- hit a deer or crash the car?

So as powerful the technology is, there will be social problems with adaptation.  Moreover, a lot of people like to drive and want to feel in control themselves.  I believe that driverless cars could save lives in the future, but I'm afraid that there could well be a growing market for old cars that still enable human driving, even when it is inept and dangerous.