Science fiction author Charlie Stross recently penned a blog piece on the future impact of autonomously computer-driven cars. Let’s call them “robocars.” I’ve pondered this before and Stross’s post is the perfect jumping-off point for a discussion of the many issues standing between science fiction and the robocar future. Let’s take a look.
Firstly, based on the progress from DARPA’s Grand Challenge and now Google’s fleet of robocars, it’s eminently clear that computers are getting very good at driving. Robocars rely on a variety of different ways of sensing the road (stereo video cameras, laser range-finders, sonar, radar, GPS — you name it and somebody’s tried it). One of the ways that robots gain increased accuracy is by fusing data from multiple distinct sensors, and that means that the cost and reliability of those sensors will be one of the limiting factors before robocars hit the mainstream, never mind the pesky problem of all those sensors looking decidedly ugly.
Stross posits that once robocars become affordable, insurance companies and government regulations will immediately favor them over traditional cars, and it’s easy to see why. Take away the human driver and you take away driving while drunk, distracted, or drowsy. Long-distance trucking companies would immediately jump on the chance to have their rigs running all day and night without drivers who require food and sleep. Stross suggests that we won’t even bother owning cars any more, except for the occasional nut-job / TTAC aficionado who likes to race. In a congested big city where taxis are everywhere (New York, London, etc.), plenty of people already don’t bother to own cars, and new business models like ZipCar fill in where taxis don’t really cut it. Still, while I’ve only spent occasional time in New York, I have attempted to get a taxi there while it’s raining, and let’s just say that supply didn’t meet demand. With robocars, we can easily imagine ZipCar-like services where you pay more for higher priority when demand grows. Maybe we’ll see instant auctions: I’ll pay $100 for the first car that shows up right here, right now! Or reverse auctions: I need to go from here to there, who’s willing to take me for the least money? My kingdom for a ride downtown in a Mercedes! Make it so, number one.
I see a completely different impact on suburbia. With my own house, I made the tradeoff to live close to work versus having a glorious suburban starter castle and a one hour drive-time commute from hell. What if all that pain went away and I could rig up my car to be more like a rolling office? Now living in the distant suburbs wouldn’t be nearly as bad. Robocars could drive faster, with less separation between cars, and would be far less likely to get into wrecks. And when wrecks do happen, robocars wouldn’t slow down just to rubberneck. Expensive parking lots at work or the airport? Why bother? Send the car home and it will pick you up when you need it again, or send it a few miles away to a robocar-only parking lot that can really pack the cars in for a cheaper price.
Since I’m a computer security guy, I should probably spend some time on how things could go horribly wrong. Some of Stross’s commenters got into the dystopian aspects of robocars. Much like all the robots in the inexplicably lame I, Robot movie all going evil at the same time (poor Isaac Asimov, spinning like a hidden Iranian centrifuge in his grave), it’s easy to imagine that robocars would be required to have “back door” access for the government, both in terms of reporting your whereabouts (see, for example, attempts to tax cars based on miles driven rather than gas consumed) and in terms of being able to hijack your car for any of a variety of purposes ranging from instructing a bank robber’s car to go straight to the police station to various spy-vs-spy applications, up to and including murdering undesirables by driving them at high speed into any convenient brick wall.
Perhaps less ominously, it’s easy to imagine hacking your robocar to post bogus traffic announcements that cause other cars to reroute themselves away from you, giving you a clean shot at your destination. You might also send fake messages to a car from its tire-pressure sensors causing the target robocar to slow down and pull over because it thinks there may be a flat tire (the fake message part is already feasible). Computer security researchers have already determined that in-car electronics aren’t particularly well-engineered from a security perspective, which seems unlikely to change any time soon, so there may not need to be any sort of government-mandated backdoor. It will probably be there as a consequence of poor engineering.
Malicious behavior aside, teenagers will have great fun hacking their friends’ cars to take them to incorrect destinations and hacking their own cars to ignore speed limits or take them to the party while the electronic logs say they went to the library. Tinkers will still mod their robocars in a variety of ways, such as increasing the g-limit for acceleration, braking, and turning in non-emergency situations. Why? A robocar would make for a hell of a hoontastic experience! J-turn your way into every parallel parking space. Safely and accurately.
It’s equally easy to imagine the liability lawyers getting involved in all kinds of ways. If a software bug caused my car to misbehave and I got hurt, or if a car thief told my car that it wasn’t mine any more and it left me, whose fault is that? Can I sue the manufacturer for negligence? That kind of fear, alone, will slow down the rise of the robocars. It’s a safe prediction that robocars will first come to us as an evolution of taxis and ZipCar-like services, particularly when the technology is still expensive and immature. Another easy prediction: the big consumer demand will start when the Baby Boomers, now in their 60’s, hit the age where their kids agitate for the keys to get taken away. The Baby Boomers will proudly get up, shake their canes at us, and lead us into our inevitable robocar future. Let’s just hope all the security issues have been worked out beforehand.
Linguistic note: I’m using the term “robocar”, while ABC News, in the clip below, uses “self-driving cars.” If you think about the word “automobile” — “automatic” plus “mobile” — robocars are really a step toward realizing the original purpose of the car, namely to get you where you want to be, automatically.