Originally published at: A self-driving Robotaxi reportedly blocked emergency vehicles during a mass shooting | Boing Boing
…
Much like flying cars… autonomous cars are never going to be a thing that makes sense or makes our lives better. We should be focusing on public transit, not putting more individual vehicles on the road that can clog up our streets. I doubt individual cars will ever fully go away, but more public transit options is the way to go to deal with both our environmental problems and our commuting problems.
You’d figure that someone from the company could remote into the car and carefully direct or drive the car to park, or move it out of the way in the case of an emergency.
If full self driving isn’t ready for use i am baffled that these vehicles have been allowed to operate.
“Dispatch the robo-tow! …nooo! don’t tow the ambulance!! what’s the self destruct sequence again!?”
I’m surprised that remote human backup isn’t required. “Within a half hour” is a long time when people are dying.
Unless the driver is shot, it’s a lot quicker to just have that driver pull a uey (assuming they haven’t already done that because, y’know, gunfire). Even if the driver is (unfortunately) shot, a human driver can get in and move the car out of the way in one minute rather than 30.
Ticketing GM for blocking emergency vehicles is a joke of a remedy. Maybe if anyone’s injuries were worse due to the blockage they can file a civil suit against GM.
Yup. If any city should be pushing mass transit instead of individual cars, it’s SF. Even self-driving buses might be a modest improvement.
Sometimes you gotta take extreme measures to speed things up.
Most people don’t need them and improving public transit needs to be the top priority, but if and when they become safe enough for general use they absolutely will open up a lot of new options for people with vision issues or other physical limitations that prevent them from driving. Even the best public transportation system is unlikely to make grocery shopping in a modestly-sized rural town very convenient without a private vehicle involved.
And there’s no way to do that in these, otherwise why does the company need to come?
Yeah, it needs to be a few million fine for not having the ability for emergency crews to override. This needs to be fixed BEFORE someone gets hurt.
That situation required a good robotaxi with a gun.
Yeah, clearly the tech still needs work, I don’t think anyone who has looked into it who doesn’t get paid by the companies involved would think otherwise. It’s also a reminder that none of our laws are actually written in a way that you can unambiguously execute them and have sensible behavior be the result, including traffic laws, and any attempt to drive in the variability of real world conditions requires general intelligence, not just narrow, driving-specific intelligence. Since we don’t have artificial general intelligence, that needs to be a human who can take over, even if remotely, on short notice. I don’t understand what goes on in someone’s head thinking a half hour delay could possibly be ok in the context of a car blocking a street, emergency or not. If you can’t continue driving the vehicle, you pull out of the traffic lanes. That is part of the traffic code, and an important part.
Also: this is part of the statistics, and a problem that needs fixing, but the stats are still very clear that current autonomous vehicles are safer for occupants. pedestrians, and everyone else, per mile driven, than human-driven cars. It’s not even close. Please don’t lose sight of just how bad humans actually are at driving, and how many dumb mistakes we make. We need to demand continuous improvement, but we need to do it without banning a massively-net-life-saving technology just because it introduces a different set of tradeoffs than the old one.
Came for JohnnyCab ref, not disappointed.
This, to some extent. I have a friend who is legally blind. Neither Lyft nor Uber can figure out how to pick him up in his apartment complex without making him cross a busy street with no nearby crosswalk, which kind of negates their value. Would an autonomous-driving car be able to cope better? Not yet, and I dunno how long the improvements will take.
What specifically do you mean when you say “current autonomous vehicles?” Not all “self-driving” cars are equal, especially when it comes to Tesla, which has been shown to cherry-pick their statistics and actively mislead people about who was in control at the time of an accident. The NHTSA found that the Tesla Autopilot automatically disengaged within 1 second of impact in hundreds of crashes, and therefore the company had claimed that all those crashes were the fault of the drivers.
Oh for sure, I never believe a word Tesla says without hard data. That’s just good practice. Glad they kickstarted the race for everyone else to build electric cars and better batteries, that’s been great. Otherwise not a fan.
With current systems, I’d say that at a minimum, if you don’t have LiDAR on board, you have no business claiming you’re capable of autonomous operation. I’d also say that the official list of companies participating in autonomous vehicle tests according to the NHTSA list is a good starting point. Tesla isn’t on it.
Based on NHTSA data, 2/3 of “autonomous” vehicle crashes and all fatalities have been in level 2 autonomy vehicles (mainly Tesla). I haven’t found a breakdown between level 3/4/5, but I have not found any reports of serious collisions involving level 4 and 5 autonomous vehicles despite tens of millions of miles driven in total. The incidents I see in news reports mostly seem to be like this one, where the software is overly timid about breaking one rule to satisfy another. Seems like the better failure mode to have, for a new technology?
I’d also say any system that requires a human to step in on short notice while in motion is not acceptable for use on public roads. Human minds don’t work that way. We know this. It’s not mysterious. Asking humans for help after coming safely to a stop is, or should be, a totally different issue, and that’s what they seemingly couldn’t do here. Driver assist less capable than full autonomy is a recipe for humans being unable to maintain safe driving skills, unable to maintain focus behind the wheel, and therefore unable to step in at the times when they’re needed. It’s hard enough for professional pilots, let alone the rest of us.
There’s a third option you’re ignoring, though- less driving, period. If we pursued public transportation as aggressively as we did car culture, we wouldn’t need self driving car cars to save 40k lives a year.
Except that that’s how the world was for a hundred years or so, post Industrial Revolution and pre-automobile and it worked just fine. In British Columbia, for example, (to choose a random example that I happen to be familiar with) there used to be an extensive rail and steamship network blanketing the province. Every tiny mining town, village, and big city alike was deeply interconnected with hourly trains and lake/river ships. All of that is gone now. There aren’t even regional busses in most places anymore. People hitchhike between towns now because that’s all they have left if they can’t afford a car.
We mustn’t lose sight of history and think that slight variations of how things are now is the only way forward.
“But if you do that, the poor will benefit.
I want to feel good about my green-washed environmental destroying purchases.”
A further reminder that real self-driving cars do not exist yet, and we pretend they do at our own peril.
That’s true, it’s very possible to organize society in a way that you can do that. To make it work nowadays, though, you’d have to undo a lot of the development choices we’ve already made in the past 100 years. Some of them we definitely should, they were terrible choices! But I’m not sure that all the tradeoffs are worth it, even though we should move in that direction on the margin. In any case, I suspect we’ll end up with self-driving commercial vehicles anyway, so that e.g. Amazon et al can send a truck down the street on its own while drones and wheeled robots cart packages to individual doors. And once the tech exists, we’ll figure out where we can best use it or other things.