The infrastructure needed to support completely self-driving cars won’t be ready any time soon. If and when it does happen, that infrastructure is at least decades away–and it will come with a multi-billion-dollar pricetag.
Pointer from James Pethokoukis.
I think it’s worth it. Look at all the money spent creating HOV lanes. My guess is that lanes for self-driving cars would have much more social value. If you could use your car in self-driving mode on the major commuting highways and then drive it yourself when you get off that would be a nice improvement.
It irks me to see money spent creating bike lanes when instead money could be spent on making places ready for smart cars.
I wonder if the first users of fully self-driving cars won’t be seniors who’ve lost their license but still want to get to the store or the doctors office. Or blind or handicapped people. Capable adults who can drive might be among the last adopters.
I suspect that trying to build infrastructure to suit self-driving cars is akin to the problem of trying to start building an interstellar spaceship–by the time it happens, next-gen automated cars will be smart enough to not need it.
Such cars are probably further away than we think, but making cars that rely on some sensors embedded in the road is not really “interesting,” it’s basically a train without the tracks, wohoo, and for the cost of building and maintaining that you could just make the roads ever-safer for manual drivers.
I agree with James Carruthers. The problem is that they are getting in too big a hurry. It would be better to wait until cars can truly drive themselves, without needing all that specialized infrastructure.
I still believe self-driving cars are 10 – 15 years away as the programming has only worked in optimal driving conditions. And the sub-optimal driving is probably twice or three harder here. So it will happen but we have to remember most technology takes 10 to 20 years of integration. (Just think the internet was invented in 1969.)
So why would we build infrastructure around relatively vaporware at this point? Also I believe they will grow the Uber/Lyft/Taxi first as they would have more interest in programming at $25K per car.
It is similar to the automation problem generally. Humans are flexible so we have flexible traffic rules. Requiring programming if there were sensors everywhere and very precise traffic rules is easier than having to deal with interpreting computer vision and heuristics. I thought machine-learning was supposed to help with these.
They are confused by electric drive which has nothing to do with self driving. Assume combustion tech and re-compute and the author will have something worth reading.
For example, do not worry about recharging, just fill up. With that hassle gone, then missing white lines are easily replace with low cost beacons, or simply GPS. The ride sharing aspect changes nothing, you still have roads.
The real infrastructure required are for the high speed lanes with vehicles running 120-`140 MPH, the new high speed bus transit. There is little gain without speed. These vehicles will all be combustion.
Seems like the major hurdle is when self-driving cars and unpredictable human drivers start mixing it up
If all the cars were automated and can communicate it is am easier problem.
“It irks me to see money spent creating bike lanes when instead money could be spent on making places ready for smart cars.”
I wouldn’t support spending on either. Special lanes wouldn’t be for ‘smart’ cars, but ‘not so smart’ cars that can’t handle even the simplified environment of limited-access expressway driving (no intersections, pedestrians, bicyclists, street parking, etc). Truly smart cars shouldn’t require any additional infrastructure spending beyond what we’re already doing.
“Self-driving cars need better roads, better lane markings, better traffic-light timing, and better maintenance to behave predictably. Lane striping must be clear and regular. The road surface must contrast crisply with the striping. With today’s technology, the roads must be clear of snow and ice and so should the car’s sensors that are reading those roads.”
Sorry — self-driving cars that need that level of road perfection to operate safely really don’t belong on the roads.
“Truly smart cars shouldn’t require any additional infrastructure spending beyond what we’re already doing.”
+10
Virtually all technologists I’ve heard talk about self-driving cars are not planning on specialized infrastructure. They are working toward a large, connected machine learning system that will inform the processors on individual cars. These cars will (and do) use regular signs, markings, and street lights to navigate. Precipitation is a challenge, but it’s also a challenge for humans – and there is no possibility of upgrading humans with sensors that can better deal with the elements.
There’s another way for self-driving cars to deal with bad weather: have a few empty cars/automatic maintenance vehicles go ahead of each pack of passenger-carrying cars as scouts. These can send back information via direct or centralized communications to the passenger cars which are following at a safe distance.