The problem with driverless cars is the 80/20 rule. I believe we have the 80% covered but I think everyone is extremely overly optimistic on when the remaining 20% will be fully functioning and flawless enough to fully trust.
That’s why I’m confident in Antlers timeline as well, because the scope of those scenarios is just so much lower that those scenarios have to be working properly first before you can even think of driving around town freely in a driverless car.
I don’t even know how you solve the liability problem. Who is at fault if an autonomous car crashes? The owner when they weren’t driving? The car marker? If it’s the latter, then the insurance costs kill that idea.
That’s was my whole point about the second hand driver. Pay someone minimum wage to take on the liability of the car in an accident. Uber certainly doesn’t want the liability. That they immediately blame the co-pilot here already shows exactly what they’re going to do.
I don’t think this is that big a deal - if a crash happens because of a manufacturer error, then it’s their fault, right? This might add a bit of liability insurance to the manufacturer, but if so, it reduces the cost of auto insurance and the manufacturer can charge some sort of autopilot maintenance fee or whatever that covers the liability + upgrades / updates to the software.
I don’t think it’s that far. We don’t need the technology to be 100% safe or efficient, just safer on average than human drivers, and we are not far away from that at all.
You just need one city or country to authorize them and if in one year or so of of operation accidents are statistically lower than regular vehicles, most objections will go away, including insurance costs.
It’s a bit of a fight to get manufacturers to even to admit to errors now; it’s also a long drawn out process. Why would we think they’ll take responsibility when they can point to the person in the seat and say it’s their fault. Where is the burden of proof going to fall, whether or not the kid in the car or whatever or whomever was struck has enough money to take on the company and make them prove it wasn’t their fault?
They don’t have to be perfect, but they do have to work without requiring a human driver to intercede multiple times on a typical intra-city trip, and that is well beyond the current state of the art.
That maybe true… In fact I’m sure that’s true. However, the state of the art is constantly advancing so will that statement be true in three years time? Who knows?
I think this is the big point - you can always point to tech that people thought was imminent and then failed to get there, but this is a field where advances have started to come quickly, investment has scaled up dramatically, and there’s basically a massive global race to be the first to market. The regulations will be the thing that needs to keep up, with licensing, liability rules, and perhaps some way to account for software upgrades and security for the “drivers” of autonomous vehicles. 10 years is really not a long time in the manufacturing world, so anyone who isn’t testing cars that are almost there is going to be left behind by those who are.
Here’s an optimistic view (fully autonomous cars by 2019):
and a more pessimistic one (that the cars currently announced in the 2020/1 time frame will be level 4 – everything except “extreme” weather / bad road conditions and weak maps data – and that level 5 automation will take until 2028):
Level 4 might be enough for Uber, though, even if it means shunting the cars away from certain locations and enlisting drivers when bad weather interferes. Even if it takes 12 years, though, the only way to get there is to start testing it in real conditions.
I think it’s very hard to predict. Once the technology is there (that’s the big if) adoption rates are difficult to measure since it’s a paradigm shift. It might be slow, it might be fast or it might be explosive (think smartphone market). Who knows. There has never been a change in the product as significant as this would be (well, going from carriages to cars, I guess).
Interest from the broader market will likely have hiccups, supply issues, price-related issues, etc, but none of that matters for Uber, which could easily roll out the automated driving as a cheaper alternative and jack up the price for a human driver.
Oh, once (if) the technology is there (and allowed) I would expect most transport companies (except long distance passenger transport) to shift really, really quick.
I agree. The trucking industry will almost certainly flip over to driverless as soon as it is feasible.
This actually terrifies me. You ever see an accident with a semi-truck involved? They’re not small, and I just wonder what a driverless truck does when the snow starts and they need to… chain up for the mountain pass.
This seems relatively easy to solve.
Or just wait for it to pass, unless you get snowed in / iced over.
Yeah, I think the dispatch center makes that call. If it’s a passing storm they just sit tight. Otherwise they send out a human to take over.
You ever live by a mountain pass? No one knows for sure when those will close. When the pass closes here, you see sometimes miles of semi trucks and cars of people waiting for it to open. Sometimes they’re just waiting for ODOT to tell them it’s even safe for chains.
Think about how accurately they predict hurricanes… as in how many times does NJ prepare for a Hurricane and how many times Sandy actually hits.
You ever see a semi tip when it takes a corner too fast one that looks like they should be able to do it but can’t?
How about seen a semi actually dragging a car, because some idiot thought they could sneak by but gets caught in corner of the trailer? These are usually sports cars… and you can’t see them.
LOL, ok:
Real rule #1: Humans only on the paths that require going through mountain passes that sometimes close with no warning.
Simple.
Can we continue working on the other 99.9% of the use cases now?
These aren’t rare cases on I-5, which is an interstate. Semi truck accidents here are not super rare with humans trying to accommodate other humans who act unpredictably.
Let me put it another way, a number off these accidents are not the fault of the semi driver. The semi-driver knows though that if he or she is an accident, people are very likely to die. They’re going to try and adjust for the stupidity of people driving vehicles that will not “win” against an 18-wheeler.
When you’re bored, just type in I-5 semi truck accident and browse the news section. It’s a daily occurrence, usually more than once a day. You might think a computer would do better, but the problem is, if you’re carrying several tons and someone does something stupid… it’s not if there is an accident but what kind. Is it worst to roll the roll the truck trying to turn it or plow through the cars in front of you. You decide… or I guess some developer?