Timothy B. Lee writes an interesting fantasy about the future, where cars drive by themselves. He sees plenty of benefits to such a system:
In short, a car that drives as well as the best human drivers would save tens of thousands of lives in the United States and hundreds of thousands of lives worldwide. And most likely, we’ll be able to do even better than that. Computers have much faster reaction times than humans do, and they will be “looking” in all directions simultaneously. Self-driving cars may be able to avoid many of the mistakes that even experienced human drivers make. They won’t have blind spots, they’ll have better sensors, and they will be able to react almost instantaneously to unexpected problems, giving them the ability to recover from dangerous situations that no human driver could have handled.
Dramatically fewer accidents is the most obvious—and probably the most important—benefit of self-driving technology. But self-driving technologies will also bring significant changes to peoples’ daily lives.
He also envisions several other benefits to robot cars. Fewer parking lots would be needed since a car could drop off its passengers and continue out of the way. Cars could drive through the night while passengers nap; retail stores could make deliveries with their robot cars.
His descriptions might sound nice, but it is a nonsensical idea that automatic cars will replace human drivers. In an earlier article, the author admits that there are major technological hurdles in the way of a driver-less society:
There are also bicycles, animals, parked cars (which might pull into the road at any moment), potholes, stalled vehicles, emergency vehicles, and so forth. Navigating successfully around each of these types of obstacles, safely and effectively, requires a lot of knowledge that may be obvious to human drivers but may be difficult to capture programmatically.
Nevertheless, the Ars Technica author concludes that cars driving themselves is ultimately a matter of “when not if.”
Lexus may have a car that assists drivers with inferior parking skills, but I do not ever foresee a time when the human driver is obsolete.
Issues
- How will off-public road data be collected for the system, e.g. driveways and garages?
- People like to be in control. That is a major reason people are more afraid to fly than to drive, despite the safety statistics. Will you trust a machine more than yourself?
- Cars and motorcycles can be fun to drive. Will people give that up?
- Today’s GPS navigation systems are unreliable. Will GPS and the other computer software and sensors ever be more reliable than the human body?
- While some companies are experimenting with technology, would they really take full responsibility for all accidents and expose themselves to that liability?
If you want to see a robotic car, I suggest watching Night Rider (the classic episodes not the new series).
9 Comments
1) Why do we care about this road data, and why do robocars interfere with it?
2) People in many cities are fine with not being in control in taxis, transit and robot-driven transit. They trust those things, and we won’t have robocars either until they are safe enough to get at least that much trust. But that is a question of when, not if.
3) No. They don’t need to. But they’ll do it mostly in fun to drive places like mountains, not on city streets at rush hour.
4) Yes, much more reliable. They already are. GPS is decades old.
5) Of course, as long as they don’t have to take special liability, beyond what drivers have. Robocars will only be accepted when they have an accident rate far lower than human drivers. The human driver liability cost is well understood, and is close to the cost of insuring all those drivers. That can easily be built into the price of operating the car, it will be a lower price than people pay today. Unless, for some reason, our liability system makes the company pay more per machine caused accident than insurance companies pay for human caused accidents. This is quite possible of course, but is a legal choice society makes.
Brad, thanks for your comment.
1.) I assume that if cars are going to automatically drive people from their homes to their destinations, robo-cars would need to be programmed with detailed information about driveways, garages, and so forth. Seems like a huge task.
3.) So you’re talking about robo-cars in cities only? Not all 5.7 million miles of paved roadways in the U.S.?
4.) While GPS can be useful, I think it leads to reliance on technology that can fail. Satellite TV seems to go out in bad weather. It would not be pleasant if a robo-car’s navigation system fails while everyone is sleeping in their seats.
Thanks again for your comment.
Ah, now I understand about the data. Robocars will be able to tell pavement from other surfaces, and see all obstructions, so they should not need road data at all, any more than humans need maps to drive in people’s driveways. However, if you want road data, owners of homes and driveways can certainly provide it if they want their robocar, or anybody else’s, to make use of it.
Getting it’s pretty easy. Go on something like google maps in satellite mode. Take a mouse and mark where your driveway is.
I do need to make it clear, we’re talking about software that doesn’t need GPS any more than a human needs GPS. In other words, it likes GPS in order to navigate and figure out how to get to places, but it doesn’t need it to go down the street, or to read street signs.
Though it would be handy to put in new-generation location systems which use local transmitters to get centimeter accuracy, but no robocar will depend on just one way of knowing about its environment. I expect they’ll all have a dozen ways of knowing about their environment (GPS, LIDAR, cameras in all directions with stereo and focus for distance, audio, network communication with wireless nodes and other cars, lane markers, street signs, inertial guidance, accelerometers, human input, etc.) and they’ll want them all to pretty much agree.
Robocars should work all all roads (paved or otherwise) but they will start on some roads and expand to others. It’s a matter of debate which roads they will start on first.
A navigation failure would at worst mean the car asks one of the occupants to navigate. Which does not mean drive, it means “tell me which exit I should take.”
Thanks, Brad. It sounds like you’re talking about cars driving people around responding to voice commands or a destination entered into a computer. The article I read by Timothy Lee talked about cars dropping off people at work and driving themselves back home by themselves. I don’t like the idea of robocars running around free of onboard human input.
As for examples given where autopilot cars will be “better,” some of them are where I just don’t agree that it’s better. One is that cars being able to park themselves will reduce space constraints on providing parking garages as the cars would park themselves with on one onboard. A remote controlled car could do the same, and autostacking garages offter a greater benefit. While neither solution may seem low-tech, autostackers have existed since the early 1960s. And remote controlled vehicles have existed since the 1930s.
Once you dedcate the road space absolutely the autopilot cars, doesn’t that raise the question of why not either use a fixed track cabin transit, or something that is driverless by its very nature, such as ropeways or pneumatic tube transport.
Also SP, do you also not like the idea of a convoy full sized family cars each with only one person in it?
Right, full vehicles are more efficient but not always possible for every trip.
The space saving parking is an interesting idea, I wonder how expensive that is.
Thanks for your comments!
The space saving parking is quite old. As far as I know the first one was the British autostacker, built in 1961, but it never worked properly, which was probably due to a design fault, not because it wasn’t technically possible, this was before modern (digital) computer control existed.
I’ve read that article and multiple features are touted as to why autopilots will make cars better, and some of them are examples I don’t agree make them better. To understand why, try to think of the consequences of the large reduction in public transport that happened mostly during the 1950s and 60s, this itself greatly degraded the lives of many people who were either unable to drive themselves or found it tedious and frustrating.
Post a Comment