5 ms·

Some problems with autonomous vehicles: * Time. Lets be optomistic and say they are perfected in 5 years time AND can be added to new cars for minimal cost the

by slyall 12y ago

Some problems with autonomous vehicles: * Time. Lets be optomistic and say they are perfected in 5 years time AND can be added to new cars for minimal cost then then it'll be say 2030 before all the cars on the road have them (assuming laws outlaw manual cars when 90% of people are using driverless ones) * Now at that point we are going to assuming that driverless cars are so good that they get twice the efficiency on your local road and highways resemble NASCAR races (without the 5% chance of a crash each trip) * The problem is that now since driving is so easy the amount of traffic goes up so you are back where you were previously. Anywhere with enough traffic to have problems already has a tonne of latent demand just sitting around ready to generate more journeys. Note: * 5 years is really optimistic for driverless car perfection * 10 years is to ban all manual cars? * Ever tried crossing a NASCAR raceway on foot? * Your assuming driverless vans/buses won't also happen. Share your rid cost with a dozen others, semi-fixed routes (easier to survey/programme), fleet management.

saraid216 12y ago
> * Every tried crossing a NASCAR raceway on foot? This is one of the interesting advantages a public transit system has over a self-driving fleet of cars (which, incidentally, I also look forward to). Mass transit depends on the concept of "stops" or "stations". The way people get between destinations (like their home, the shop, etc.) and the transit station is largely by walking. Thus, mass transit necessarily builds in pedestrian friendliness. This isn't true for cars, which do not have such nexi. (Indeed, this lack is generally agreed to be an advantage.) This means that such designs will have to deliberately include pedestrian friendliness in order to have it, rather than being forced into it by the facts of the design requirements. I'm not saying it's a sufficient advantage for one or the other, but it's a definite thing.
bsder 12y ago
> 5 years is really optimistic for driverless car perfection I don't think that's true. This is likely one of those problems whose solution accelerates after introduction. Let's assume that 99% of the problem is solved (probably not, but we're worst-casing here). The problem is that you need to stumble on a 1% case and fail in order to fix that bug. At this point, I suspect that we're probably limited by the fact that buggy cases just aren't coming up often enough for the small number of autonomous vehicles that exist. Look at what happened in the DARPA challenge. The first year, nobody got anywhere. It didn't take very long until that challenge went from nobody completed to worrying about things actually moving too quickly.

by slyall 12y ago

saraid216 12y ago

> * Every tried crossing a NASCAR raceway on foot? This is one of the interesting advantages a public transit system has over a self-driving fleet of cars (which, incidentally, I also look forward to). Mass transit depends on the concept of "stops" or "stations". The way people get between destinations (like their home, the shop, etc.) and the transit station is largely by walking. Thus, mass transit necessarily builds in pedestrian friendliness. This isn't true for cars, which do not have such nexi. (Indeed, this lack is generally agreed to be an advantage.) This means that such designs will have to deliberately include pedestrian friendliness in order to have it, rather than being forced into it by the facts of the design requirements. I'm not saying it's a sufficient advantage for one or the other, but it's a definite thing.

bsder 12y ago

> 5 years is really optimistic for driverless car perfection I don't think that's true. This is likely one of those problems whose solution accelerates after introduction. Let's assume that 99% of the problem is solved (probably not, but we're worst-casing here). The problem is that you need to stumble on a 1% case and fail in order to fix that bug. At this point, I suspect that we're probably limited by the fact that buggy cases just aren't coming up often enough for the small number of autonomous vehicles that exist. Look at what happened in the DARPA challenge. The first year, nobody got anywhere. It didn't take very long until that challenge went from nobody completed to worrying about things actually moving too quickly.