It's easy to see how self-driving cars would benefit society. Traffic jams gone. Accidents reduced. Leisure time increased! These are all good things. But there are a number of bad things that are keeping self-driving cars from taking over. What's the hold up?
Before diving into the details of the challenges facing self-driving cars, let's talk about the good things. The Eno Center for Transportation just released a comprehensive report on the state of self-driving cars that includes some pretty impressive figures. Authors Daniel Fagnant and Kara Kockelman of the University of Texas estimate that a future where just 10 per cent of the cars on the road are autonomous would prevent 211,000 accidents every year. If you bump that percentage up to 90, the number of crashes avoided soars to 4,220,000.
Then there's the money saved. Because self-driving cars can make split second changes to improve fuel efficiency and travel closer together to prevent traffic jams, they'd save 756,000 hours of driving time and reduce fuel consumption by 102,000 gallons in the 10 per cent scenario. This adds up to a total of $US16.8 billion of savings per year or about $US1320 for each self-driving car. All told, the country would save $US37.7 billion a year if just 10 per cent of the cars on the road were autonomous. These are spit-balled figures — "prognostications" according to Fagnant — but they make a point.
Unfortunately, getting to the point where one in 10 cars on the road is autonomous will not be easy. In fact, it's going to be super effing complicated. The roadblocks (heh) run the gamut from regulatory disarray to basic economics.
The technology is not quite there
The most obvious hurdle for self-driving cars is the technology that makes them drive themselves. We've obviously made huge leaps and bounds from 2004, when not a single self-driving car made it to the finish line on a seven mile course in a DARPA test. We still have a ways to go, though.
For example, we haven't built self-driving cars to deal with intense conditions like busy city driving and extreme weather. Snow counts as extreme in this case, and self-driving cars suck at dealing with it. They can't "see" the lines on the road and get confused. This goes for pretty much any quick changes in road conditions like accidents or construction zones. Cities are tough for similar reasons, since it's not uncommon for traffic police to direct traffic — and detours happen more often. In these cases, the self-driving cars could literally get lost.
But if there's anything that's been impressive about the rise of the self-driving car, it's how fast the technology has improved. Companies like Nissan say they'll be ready to sell self-driving cars to the public by 2020, and Volvo's right behind them.
These things are expensive
Self-driving cars are pretty awesome, and like many awesome things, they are not cheap. The equipment alone on Google's self-driving car costs between $US75,000 and $US85,000. That's significantly more than a brand new Tesla! Of course, when you add all the other costs of building a car on top of that, you're well north of $US100,000. Since the best-selling cars in America cost between $US16,000 and $US27,000, that price is going to have to come down, and come down it will. The only question is how quickly?
Car companies are bullish about this sticking point. Nissan, for instance, says that its self-driving car will only be $US1000 more expensive than the non-autonomous models. Meanwhile, one study found that it will take 20 to 22 years just to bring the price of a self-driving upgrade down to $US3000. There might also be added costs on top of this, like higher insurance premiums, though there will also be savings. No matter how you cut it, though, these things aren't going to be cheap for years to come.
The law is a mess — privacy too
Obviously, since it's only about a decade old, this technology is largely unregulated. This won't last forever. California and Nevada have already legalised self-driving cars on the road, while Florida and Washington DC have given the OK for testing. However, the laws that govern self-driving cars vary widely in these states. With the risk that standards could vary from state to state, many believe that we'll need national legislation to be enacted before we can truly open up the roads to self-driving cars.
There's also the issue of privacy. Because these cars necessarily collect a lot of data about when and where people are going, drivers ought to be concerned about how that data's being handled. In the words of Fagnant and Kockelman, "Without proper safeguards, this data could be misused by government employees for tracking individuals, or provided to law enforcement agencies for unchecked monitoring and surveillance." Sounds familiar, doesn't it?
Hackers gonna hack
This is where it gets real. Because self-driving cars are powered by computers and computers can inevitably be hacked, there are some serious security concerns here. In the same way that government spies can suck up the torrents of data being produced by self-driving cars, hackers could break into the network that would inevitably keep self-driving cars from crashing into each other. Terrorist-types could simply disable some safety features and cause huge amounts of damage — or, as the Eno study's authors suggest, create a nightmare scenario when all self-driving cars on the road accelerate to 70-miles-per-hour on command.
The encouraging thing about these security concerns is that we've been pretty good at keeping hackers away from our large-scale infrastructure. Even though President Obama wants you to know that a cyber attack could kill tons of people one day, it hasn't happened yet. (Knock on wood.) And so with any of these potential problems with self-driving cars — and there are many more — they remain, for now, potential problems.