Tom Cruise escaped the law in one in the movie Minority Report. The Batmobile is another. For so long depicted as the stuff of science fiction, autonomous cars are here. And unlike jetpacks or self-tying shoelaces, we’ll be able to get our hands on them in the next few years.
How safe are they? Google’s own version has trundled around some 700,000 miles without major mishap, while Audi’s A7 concept car drove itself from Las Vegas to San Francisco.
Yet there is still some trepidation around the notion of leaving control of a vehicle to a computer processor. Here, we look at some of the realities.
Autonomous cars don’t cut out human error
Vehicle safety solutions have helped to reduce the dangers of driving and getting behind the wheel of a car today carries a lower risk than ever. But accidents still happen – so surely a robot car would eliminate the chances of one occurring? Well, assuming everything works as it should, that would be the case. Up to a point.
There is still a human element: the designers and programmers who built the car. They can only programme a computerised response to a situation if they have envisaged it happening. But as Chris Gerdes, programme director for the Centre for Automotive Research at Stanford, told TechRepublic.com, “The idea that you could take a step back and program the car in the comfort of your office, in a lower stress environment to handle all these stressful situations, is potentially a huge improvement in safety.”
Humans aren’t worse at driving than robots
While some driving mishaps are baffling, for the most part we don’t do too badly at guiding our two-tonne cubes around the roads. Where we have a real advantage over computers is in perception. Scanners and GPS systems both help autonomous cars make reactive adjustments to road systems and other vehicles in milliseconds, whereas we are better at understanding our environment and what might happen one mile down the road. Intuition isn’t something that can be coded (yet).
Automated cars can’t go everywhere
The guiding force for an automated car is the network of satellites high above its roof. GPS systems are fundamental to an automated car knowing where it is and where the road goes. Block that signal, say with heavy rain, cloud or a garage roof, and the computer piloting the car suddenly becomes blind. Until a work-around is developed, that severely limits an automated car’s practical use.
You’re not quite a passenger
The ultimate goal of automated cars is for the passengers they carry to have to pay no attention to the road. Pedals and steering wheels will disappear, front seats will no longer need to face forwards. But until then, we’ll still need to play an active, if reduced, role in many situations, monitoring what the automated car is doing, which lane it’s aiming for, how close it is to other traffic, and where that emergency services siren is coming from.
In other words, we’ll still need to keep our hands near the wheel, and our foot near the brake.