Not exactly an independent source, I think?
I’d love to have a self-driving car, like my own personal taxi.
But I don’t think we are very close yet.
Well. I’d say the conclusion is more nuanced than that, for example they say in the conclusion that their numbers are clearly better for police reported/any injury reported crashes, but the benchmarks for property damage or injury crashes have such wild swings in them that it’s hard to draw a conclusion. All of that they speak to themselves, for example, that as they used a phone survey to adjust for people who don’t report small accidents,
I would describe this entire study as an actual, good faith effort, with some large underlying assumptions that they are very clear about. All of that, of course, will get boiled down to a headline.
I appreciate your skepticism, but for that particular example give me the robot with multiple sensors over almost anyone checking their texts every time. Sure, not perfect, but human drivers have never been a high bar and there is nothing going on today that is making them any better. To me, multiple always-on sensors is just a superior arrangement to the typical human. And unlike most humans, they WILL improve with experience.
I rather like the idea in a Heinlein book (can’t remember which one now) where one of the characters expresses incredulity that people were once allowed to drive without even passing basic calculus. 
BMW has apparently given up on the “self-driving” part and wants to outsource a desk jockey to park your car for you.
Congrats on inventing valet parking with extra steps.
Honestly wtf
The secret is that the car is not truly parking itself. This isn’t a limited case of driving autonomy. You’re actually handing over control of your car to a real human sitting in a sort of call center full of sim racing rigs.
Do they really need a full racing rig with full haptic feedback? I was picturing something like this.
ETA: That’s also how I picture remote valeted cars ending up…
Hehehe, the funniest part of that to me is that the car can do the parking just fine by itself. The human is used to get from the door to the parking space.
Nah, that’s one of the things that Ars points out. Their setup is nicer than my sim racing rig, and they max out at about 6 MPH when using the system. The pedal is also directly mapped to speed. Floor it, you get 6 MPH. Half throttle gets you 3 MPH.
Hilarious, but it does actually work as advertised.
Yeah, but the valets will get paid $2/hr instead of $20. Whether that’s a boon or not perhaps depends on one’s views about the global economy.
This kind of thing has basically been a punchline to jokes about self-driving for years now: if it’s too hard, we’ll just outsource the drivers to India or wherever. I suppose I shouldn’t be shocked that someone’s actually going for it.
Hmm, can the drivers use a speaker to yell at pedestrians?
I often joke about using “the Braille method of parking” when driving in person.
I suppose you’d need haptic feedback so the remote drivers know when they’ve run up on the curb or tapped the car ahead or behind.
It’s all well and good until there’s a lag spike and your car continues driving in a straight line at constant speed until it hits a wall.
Will BMW owners have to pay for this as a monthly service, do you think? Remember that a BMW already has its own parking system if you’re just trying to get into a space.
I also don’t understand where you would use this. I don’t think you could use it at a place like a hotel, where valet parking has its own spaces, but otherwise you have to drive into a garage and take a ticket. And if you wanted to park, say, on a street in Boston, you would have to figure out some way to pay the meter.
I guess I could drop you at the door of a grocery store or such.
We don’t have the technology or the infrastructure for self-driving cars. I doubt we will within the next 20 years.
Huh.
I guess they didn’t get the memo:
Now, now, you didn’t let them explain that “we” was a hermit in a remote crag of a Himalayan glacier.
Or anyone not living in SF or PHX…
What technology and infrastructure are missing? Paved roads? I get that up-to-date maps and GPS help, but I didn’t realize that there are only two metropolitatan areas (worldwide?) that are so equipped.
You don’t really need infrastructure to have self-driving cars.
The whole point of the current designs is that they do not require specialized roads on which to run. Why not? Because that path to adoption simply will not happen. They have to run on ordinary roads in ordinary cities surrounded by ordinary humans behaving ordinarily while on foot, bikes, mopeds, motorcycles, or driving cars, busses, or trucks.
You just need a company willing to perform their experiments in your city. And a city and state government willing to let themselves, and their populace, be the guinea pigs. Nice weather helps, since all this is harder in snowy or intense rain conditions.
And most elusive of all, rational minds. If the self-driving cars can be shown to be statistically safer, on average, than human drivers, wouldn’t that be the logical make-or-break requirement? Yes, but logic isn’t how we’re wired. We need technology and infrastructure until we’re blue in the face. And let one robot bump one pedestrian, and we’re outraged (never mind the flocks of drivers who magically appear at bar closing time every night).
Well yes, that is one of the larger problems.
This 1500-post thread covers 6-1/2 years of progress and non-progress with self-driving vehicles. The public acceptance / liability issue has been debated several times along the way. And this is far from the only self-driving vehicle thread. It’s the longest lived and closest to an omnibus thread, but there’s plenty more if one cared to search.