I love the corpospeak. why say “crashed into” when you can use “made contact” which sounds futuristic and implies that your product belongs to an alien civilization?
I wasn't asking about the car's logic algorithm; we all know that the SDC made an error, since it [checks notes] hit another car. We already know it didn't do the correct thing. I was asking how else you think the developers should be working on the software other than one thing at a time. That seemed like a weird criticism.
Ideally they don’t need actual accidents to find errors, but discover said issues in QA and automated testing. Not hitting anything sounds like a manageable goal to be honest.
Honestly, I’m pragmatic, if less people die in accidents involving autonomous car, then yes.
The thing is we shouldn’t be trusting the manufacturers for these stats. It has to be reported by a government agency or something.
Similarly Autonomous car software should have to be certified by an independent organization before being deployed. Same thing for updates to the software. Otherwise we would get deadly updates from time to time.
If we deploy and handle autonomous car with the same safety approach as in aviation I’m sure this transition can be done fairly safely.
Rules are written in blood. Once you figure out all the standard cases, you can only try and predict as many edge cases that you can think of. You can’t make something fool proof because there will always be a greater fool that will come by.
Unexpected or not, it should do its best to stop or avoid the obstacle, not drive into it.
An autonomous vehicle shouldn’t ever be able to actively drive forward into anything. It’s basic collision detection that ought to brake the car here. If something is in the position the car wants to drive to, it simply shouldn’t drive there. There’s no reason to blame the obstacle for being towed incorrectly…
Same thing applies to a human driver. Most accidents happen because the driver makes a wrong assumption. The key to safe driving is not getting in situations where driving is based on assumptions.
Trajectory calculation is definitely an assumption and shouldn’t be allowed to override whatever sensor is checking for obstructions ahead of the car.
The car can’t move without trajectory calculations though.
If the car ahead of you pulls forward when the light goes green, your car can start moving forward as well keeping in mind the lead cars trajectory and speed.
If it was just don’t hit an object in its path, the car wouldn’t move forward until the lead was half way down the block.
The car knew the truck was there in this case, it wasn’t a failure to detect. Due to a programming failure it thought it was safe to move because the truck wouldn’t be there.
If you’re following a vehicle with proper distance and it slams the brakes you should be able to stop in time as you’ve calculated their trajectory and a safe speed behind. But if that same vehicle slams on the brakes and goes into reverse, well… Goodluck.
It’s all assumptions assuming the detection is accurate in the first place.
If you’re following a vehicle with proper distance and it slams the brakes you should be able to stop in time as you’ve calculated their trajectory and a safe speed behind.
You dont need to calculate their trajectory. It’s enough to know your own.
If a heavy box falls off a truck and stops dead in front of you, you need to be able to stop. That box has no trajectory, so it’s an error to include other vehicles trajectories in the safe distance calculation.
Traffic can move through an intersection closely by calculating a safe distance, which may be smaller than the legal definition, but still large enough to stop for anything suddenly appearing on the road. The only thing needed is that the distance is calculated based on your own speed and a visually confirmed position of other things. It can absolutely be done regardless of the speed or direction of other vehicles.
Anyway. A backwards facing truck is a weird thing to misinterpret. Trucks sometimes face backwards for whatever reasons.
It would be interesting to know how the self driving car would react to a ghost driver.
You dont need to calculate their trajectory. It’s enough to know your own.
This doesn’t make sense. It’s why I was saying the car won’t move at a stop light when it goes green until the car is half way down the street.
If the car is 2.5 seconds ahead of me at 60mph on the highway, it’s only 2.5 seconds ahead of me if the other car is doing 60 mph. If the car is doing 0mph then I’m going to crash into it.
It needs to know how fast and what direction the obstacle is going, and how to calculate the rate of acceleration/deceleration and extrapolate from there.
2.5 seconds at 60 mph is more than enough to come to a full stop. If the car in front of you dropped an anvil (traveling at 0 mph) on the road, you could stop before crashing into the anvil. You do not need to drive into the other cars trajectory path.
Yes you can. It is a stopping distance. 2.5 seconds at 60 mph is 220 feet. A car can brake from 60 to 0 in less than 220 feet. It will take longer than 2.5 seconds to do, but it won’t hit the object which originally was 2.5 seconds ahead.
Maybe a straight behind isn’t as good an example, although it is calculating the likelihood of it continuing to go straight.
An oncoming car, drifting out of the lane towards your lane.
It’s not going to hit you until it’s in your path, but the trajectory of it coming towards you is in your path.
If you don’t consider where it’s going and how fast it’s going, you won’t know if it’s going to enter your lane before you pass it.
If you’re only trying to avoid hitting objects and its not in your path until the last quarter second, you won’t take appropriate actions because you don’t know it’s coming at you.
All these measurements are taken as time between you and them and it uses that info to calculate the trajectories.
Yes I know and it should. What I am saying is that the trajectory calculations should never be allowed to override the basic collision calculations, like it did in this case.
It does not matter if the towed truck appeared to have a different trajectory than it actually had, because it was very obviously in the range of collision.
Do you have a reverse sensor in your car that beeps when you’re close to stuff?
It was the self driving car that drove into the tow truck. All it’s sensors must’ve been beeping, and it still decided to keep driving.
The description of an unexpected/(impossible) orientation for an on road obstacle works as an excuse, right up to the point where you realize that the software should, explicitly, not run into anything at all. That’s got to be, like, the first law of (robotic) vehicle piloting.
It was just lucky that it happened twice as, otherwise, Alphabet likely would have shrugged it off as some unimportant, random event.
It’s great though - that’s how you get amazing services and technological advancement.
I wish we had that. In Europe you’re just stuck paying 50 euros for a taxi in major cities (who block the roads, etc. to maintain their monopolies).
Meanwhile in the USA you guys have VR headsets, bioluminescent houseplants and self-driving cars (not to mention the $100k+ salaries!), it’s incredible.
True, but your savings on non-luxury bones helps with the fees associated with luxury ones, I’m sure. I can’t do anything for my bones with a $30 glowing petunia.
Yeah it’s $40 for an Uber in Columbus or Cleveland as well. There isn’t a monopoly on taxis creating that price, thats just how much it actually costs to rent a car for cross city travel.
If you want a no regulations/free market at the helm, you want to move to India. They have all the rules you love.
It should of course not run into anything, but it does need to be able to identify obstacles at the very least for crash priority when crazy shit inevitably happens. For instance, maybe it hits a nice squishy Pomeranian that won’t cause any damage to itself instead of swerving to avoid it and possibly totalling itself by hitting a fire hydrant.
Or maybe it hits the fire hydrant instead of a toddler.
At any rate, being able to identify an obstacle and react to unexpected orientations of those obstacles is something I think a human driver does pretty well most of the time. Autonomous cars are irresponsible and frankly I can’t believe they’re legal to operate.
I didn’t read it as them saying “therefore this isn’t a problem,” it was an explanation for why it happened. Think about human explanations for accidents: “they pulled out in front of me” “they stopped abruptly”. Those don’t make it ok that an accident happened either.
After an investigation, Waymo found that its software had incorrectly predicted the future movements of the pickup truck due to “persistent orientation mismatch” between the towed vehicle and the one towing it.
Having worked at Waymo for a year troubleshooting daily builds of the software, this sounds to me like they may be trying to test riskier, "human" behaviors. Normally, the cars won't accelerate at all if the lidar detects an object in front of it, no matter what it thinks the object is or what direction it's moving in. So the fact that this failsafe was overridden somehow makes me think they're trying to add more "What would a human driver do in this situation?" options to the car's decision-making process. I'm guessing somebody added something along the lines of "assume the object will have started moving by the time you're closer to that position" and forgot to set a backup safety mechanism for the event that the object doesn't start moving.
I'm pretty sure the dev team also has safety checklists that they go through before pushing out any build, to make sure that every failsafe is accounted for, so that's a pretty major fuckup to have slipped through the cracks (if my theory is even close to accurate). But luckily, a very easily-fixed fuckup. They're lucky this situation was just "comically stupid" instead of "harrowing tragedy".
They've already been testing on private tracks for years. There comes a point where, eventually, something new is used for the first time on a public road. Regardless, even despite even idiotic crashes like this one, they're still safer than human drivers.
I say my tax dollar funded DMV should put forth a significantly more stringent driving test and auto-revoke the licenses of anybody who doesn't pass, before I'd want SDCs off the roads. Inattentive drivers are one of the most lethal things in the world, and we all just kinda shrug our shoulders and ignore that problem, but then we somehow take issue when a literal supercomputer on wheels with an audited safety history far exceeding any human driver has two hiccups over the course of hundreds of millions of driven miles. It's just a weird outlook, imo.
People have been hit and killed by autonomous vehicles on public streets due to bad practices and bad software. Those cases aren’t hiccups, those are deaths that shouldn’t have happened and shouldn’t have been able to happen. If a company can’t develop its product and make it safe without killing people first, then it shouldn’t get to make the product.
People have been hit and killed by human drivers at much, much higher rates than SDCs. Those aren't hiccups, and those are deaths that shouldn't have happened, as well. The miles driven per collision ratio between humans and SDCs aren't even comparable. Human drivers are an order of magnitude more dangerous, and there's an order of magnitude more human drivers than SDCs in the cities where these fleets are deployed.
By your logic, you should agree that we should be revoking licenses and removing human drivers from the equation, because people are far more dangerous than SDCs are. If we can't drive safely without killing people, then we shouldn't be licensing people to drive, right?
I’m all for making the roads safer, but these companies should never have the right to test their products in a way that gets people killed, period. That didn’t happen in this article, but it has happened, and that’s not okay.
People shouldn't drive in a way that gets people killed. Where's the outrage for the problem that we've already had for over a century and done nothing to fix?
I’m getting tired of implementing technology before it’s finished and all the bugs are worked out. Driverless cars are still not ready for prime time yet. The same thing is happening currently with AI or companies are utilizing it without having any idea what it can do.
You’re right there should be a minimum safety threshold before tech is deployed. Waymo has had pretty extensive testing (unlike say, Tesla). As I understand it their safety record is pretty good.
How many accidents have you had in your life? I’ve been responsible for a couple rear ends and I collided with a guard rail (no one ever injured). Ideally we want incidents per mile driven to be lower for these driverless cars than when people drive. Waymos have driven a lot of miles (and millions more in a virtual environment) and supposedly their number is better than human driving, but the question is if they’ve driven enough and in enough varied situations to really be an accurate stat.
My point asking about personal car incidents is that each of those, like your car tap, show we can make mistakes, and they didn’t merit a news story. There is a level of error we accept right now, and it comes from humans instead of computers.
It’s appropriate that there are stories about waymo, because it’s new and needs to be scrutinized and proven. Still it would benefit us to read these stories with a critical mind, not to reflexively think “one accident, that means they’re totally unsafe!” At the same time, not accepting at face value information from companies who have a vested interest in portraying the technology as safe.
I obviously do since I said look at how the data is collected, what is counted and what is not. Take your own advice and look into that. It’s not this one accident that makes me think it’s unsafe, and certainly not ready to be out there driving.
Here’s an article saying that based on data so far, waymo is safer than human drivers. If you have other information on the subject I’d be interested to read it.
Also your own article “But it’s going to be another couple of years—if not longer—before we can be confident about whether Waymo vehicles are helping to reduce the risk of fatal crashes.”
tired of implementing technology before it’s finished
That’s is every single programme you’ve ever used.
Software will be built, sold, used, maintained and finally obsoleted and it will still not be ‘complete’. It will have bugs, sometimes lots, sometimes huge, and those will not be fixed. Our biggest accomplishment as a society may be the case where we patched software on Mars or in the voyager probe still speeding away from earth.
Self-driving cars, though, don’t need to have perfectly ‘complete’ software, though; they just need to work better than humans. That’s already been accomplished, long ago.
And with each fix applied to every one of them, it’s a situation they all shouldn’t ever repeat. Can we say the same about humans? I can’t even get my beautiful, stubborn wife to slow down, leave more space, and quit turning the steering wheel in that rope-climbing way like a farmer on a tractor does (because the airbag will take her hand off).
That’s is every single programme you’ve ever used.
No software is perfect, but anybody who uses a computer knows that some software is much less complete. This currently seems to be the case when it comes autonomous driving tech.
And with each fix applied to every one of them, it’s a situation they all shouldn’t ever repeat.
First, there are many companies developing autonomous driving tech, and if there’s one thing tech companies like to do is re-invent the wheel (ffs Tesla did this literally). Second, have you ever used modern software? A bug fix guarantees nothing. Third, you completely ignore the opposite possibility - what if they push a serious bug in an update, which drives you off a cliff and kills you? It doesn’t matter if they push a fix 2 hours later (and let’s be honest, many of these cars will likely stop getting updates pretty fast anyway once this tech gets really popular, just look at the state of software updates in other industries).
I understand your issue with these cars - they’re dangerous, and could kill people with incomplete or buggy software. I believe the person you are responding to was pointing out that even with the bugs, these are already safer than human drivers. This is already better when looking at data rather than headlines and going off of how things seem.
Personally, I would prefer to be in control of the vehicle at all times. I don’t like the idea of driverless tech either.
Well, has anyone done good statistics to show all the self driving cars are more dangerous than regular distracted humans as a whole?
We can always point to numerous self driving car errors and accidents, but I am under the impression that compared to the number of accidents involving people on a daily basis, self driving cars might be safer even now?
I’m thinking of how many crashes took place in the time it took me to type this out. I’m also curious about the fatality rate between self or assisted driving vs not.
I think we tend to be super critical of new things, especially tech things, which is understandable and appropriate, but it would be nice to see some holistic context. I wish government regulators would publish that data for us, to help us form informed opinions instead of having to rely on manufacturers (conflict of interest) or journalists who need a good story to tell, and some clicks.
Currently there are many edge cases which haven’t even been considered yet, so maybe statistically it is safer, but it doesn’t change anything if your car makes a dumb mistake you wouldn’t have and gets you into an accident (or someone else’s car does and they don’t stop it cause they weren’t watching the road).
I’m against driverless cars, but I don’t think this type of errors can be detected in a lab environment. It’s just impossible to test with every single car model or real world situations that it will find in actual usage.
An optimal solution would be to have a backup driver with every car that keeps an eye on the road in case of software failure. But, of course, this isn’t profitable, so they’d rather put lives at risk.
Idk I kinda like the idea of a madman living in my graphics card. I want to be able to spin them up and have them tell me lies that sound plausible and hallucinate things.
They say it works without an internet connection, and if that’s true this could be pretty awesome. I’m always skeptical about interacting with chatbots that run in the cloud, but if I can put this behind a firewall so I know there’s no telemetry, I’m on board.
You can already do this. There are plenty of vids that show you how and it’s pretty easy to get started. Expanding functionality to get it to act and respond how you want is a bit more challenging. But definitely doable.
engadget.com
Oldest