I feel the question of moding a AV does bring up the question of personally owning one and the whole ideal of owning a vehicle.
As say they did operate them in a circuit so you can use a bike or walk to the stops it would make the ideal of owning rather inconvenient
There is a youtube channel from someone that lives in the netherlands that talks about bad roads/infrastructure and ways to fix it called Not Just bikes
There are so very many tech "innovations" that companies are trying to push out for profit without actually testing it thoroughly in real-world conditions nor considering the legal and ethical ramifications of deploying the technology to the public - or even just to private entities. They just "disrupt" with a bunch of buzzwords and empty promises that people take as gospel, and we can only pray that legislation catches up to account for this new variable, nevermind all the chaos that comes as the limits of their highly-regulated testing parameters are exposed once real-world variables start popping up.
Faster reaction times for AI don't affect the time and distance needed for a vehicle to come to a complete stop, but those factors do increase with speed, which I feel needs stressing to those saying that automated driving could mean faster speeds and shorter follow distances between vehicles.
The main difference with stopping is that you won't rear-end the car in front of you. You could follow very closely behind them, and as long as they have time to stop, you have time to stop. This is specifically with AVs, it's very not the case with human drivers.
Now, that effect doesn't really come into play with something unexpected appearing in front of you. If a biker/runner suddenly cross the street from an obscured viewpoint and you need to stop quickly, the AV would still do it better, but you're right that physics start to really come into play, then.
But if we take a moment to assume "good" conditions on a highway? You could be going like, 100mph with 10 feet in between you and the car in front of you, and that's totally fine for an AV. A theoretical future AV that works well, not what we have now. That's kind of what I mean when I'm saying that the stopping distance calculations would change. It's unlike it'll change to be that extreme anytime soon, but hopefully the idea here is coming across. AVs are not ready for that now, again. But they could do more than humans in that direction.
No, that would be incredibly unsafe. You need to assume that the car in front of you may rear-end a semi that has stopped or a huge pileup or something and come to a stop more or less instantly at any time. As such you need to follow at a distance where you can stop before hitting them if that happens. This is actually worse for an AV because it can't see things like that coming as far ahead as a human can. A safely designed AV would have a greater follow distance than your average driver.
You're assuming the AVs wouldn't be passively talking to each other. The AV in a crash caused every AV behind it to slow/stop/reroute simultaneously so you don't get a pile-up. The reactivity would be predictive as you got farther from the crisis point.
You won't be able to assume that all cars on the road are also autonomous for probably at least 30-50 years.
I think it will be faster. Once it becomes an option for every car, it’ll be a quick jump to 25% adoption for the folks that have to have the newest best rides. Then it’ll be one car generation (5-7 years) for it to be standard and get 65% of the market.
That last 35% are going to be old, poor and/or conspiracy nutters.
But to get to that 65% mark, 15 years from now.
This depends pretty heavily on country, to be fair. The average age of Car /Light truck in the US is currently around 11 years, while in NZ it's currently around 14. Australia is slightly ahead at 10,, and Canada is at 9. But, even assuming all manufacturers rolled this out in all models, which is wildly optimistic, it still seems like it will take longer to roll through.
Oh hey, I didn’t know (or forgot) we had an AV thread.
Anyways I’m currently working on my second AV project. On both projects, my role was in brake control software development and calibration. So while I don’t directly work on AV control logic (“Should I apply the brakes right now?”) I do spend a lot of time interfacing with it and reacting to its commands (“Computer says decel at 5 m/s2, apply 6 mPa brake pressure.”)
Project 1 was for an startup making a driverless delivery vehicle. It was an utter shitshow, the highlight of which was when we delivered brake software to the client and had them acknowledge that it was for closed course use only and unsafe for public roads. They immediately put it on public roads anyway because fuck you, Silicon Valley don’t care.
Project 2 is for a very respected Japanese OEM building prototypes for closed course testing in a safe manner. It is also a shitshow, has been in development for 5 years so far and will be ready for 40 kph closed course testing this month. Maybe. Mostly it’s notorious around the test track for things like suddenly losing steering control.
Anyway this stuff is a long way off yet, and I sure as shit wouldn’t trust a word of what any techbro company is touting about AV performance. The legacy automotive OEMs are at least attempting to do it safely.
This morning at the red light next to me, a guy had his cellphone propped up in the AV cluster of his dash and was watching some sort of TV show. It was there when he pulled up and he drove away with it there.
Don't care if they are level 2 or level 5 AVs ASAP.
Using land travel to get between major metro areas just seems like a better argument for making light rail a fast consumer friendly experience instead of giving everyone automated cars. But again we just get into recursive argument of america being designed for cars meaning we can't do anything but support cars even though plenty of modern countries have shifted away from them
Using land travel to get between major metro areas just seems like a better argument for making light rail a fast consumer friendly experience instead of giving everyone automated cars. But again we just get into recursive argument of america being designed for cars meaning we can't do anything but support cars even though plenty of modern countries have shifted away from them
Especially since trains can provably gets to 2-300 km/h. With existing technology. From the '70s.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
Using land travel to get between major metro areas just seems like a better argument for making light rail a fast consumer friendly experience instead of giving everyone automated cars. But again we just get into recursive argument of america being designed for cars meaning we can't do anything but support cars even though plenty of modern countries have shifted away from them
is it?
Perry's commuter bullet between Dallas, Austin, and SA would have been awesome but the amount of eminent domain legal battling just to get the land made it impossible. Everybody in the cities wants high speed rail between them, and they also want all the people in between to give up their property in service to that goal.
It's a hard sell, even here where conditions are perfect.
Using land travel to get between major metro areas just seems like a better argument for making light rail a fast consumer friendly experience instead of giving everyone automated cars. But again we just get into recursive argument of america being designed for cars meaning we can't do anything but support cars even though plenty of modern countries have shifted away from them
The problem with rail is always what happens when you get there. Fixing car-based design at the local level kinda has to come before connecting hubs.
Is traffic between major city centers even that big of a problem? In my experience, traffic volume near cities is highest when people are going from and back to the low-density suburbs and rural areas around the city, not shuttling between cities. It'd be really cool if I could take a high-speed rail line from Raleigh or Durham to wherever I want to go in the US, or even just the southeast, but I'd still have to drive a half hour from my house at the end of a private road which is one of many short, private roads off a very long two-lane road mostly through farms and woods a good half a dozen miles from the nearest highway ramp to get to a train station.
I tried googling public transit from my house to downtown Raleigh and google just said, "No". The closest bus stop is 3 miles away at a park-and-ride place for a technical college.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
Did you buy a new car or did finance a new car?
I purchased it outright.
edit: I also called several other insurance companies to compare and all were higher than the company I was (and still am) already with (GEICO if it matters).
edit2: and the specific reason for the massive increase was that newer tech is more expensive to replace. (I went from a 2004 Saturn Ion to a 2020 Hyundai Elantra) ($45 a month to $120 a month with the exact same deductible) (100% clean driving record)
Well, how can you tell now if someone mods their car in an unsafe manner? You wait for it to get noticed by a cop or investigated after an accident. These are good questions, but they're not exactly new problems brought on by AVs.
You're overlooking the point of the entire thread, which is that those normal cars modded in unsafe manners still have human drivers, which are way better than computers at handling unexpected situations. Modded cars with no drivers have no such safety net. They are much more dangerous. It's not a 'new problem' but it will be magnified in new ways because of a lack of human oversight.
Again, the Tesla results, which are the worst of all AV companies, and have resulted in multiple deaths, are the closest we have to real world conditions. I think that at some point in the future AVs will be safer than human drivers, but as far as how they would do now, it's a crapshoot.
edit:: It's like saying that a child didn't get into trouble while their parent was around, so the child should be just as fine on their own.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
If by "interesting" you mean "same old problematic shit."
We already have algorithms deployed to "make sure" kids are paying attention during online classes or employees are not sleeping during Zoom meetings or people are not displaying "suspicious behavior" in stores or in airports.
These algorithms are all flawed and are especially egregious when deployed upon neurodivergent individuals and non-white persons.
Well, how can you tell now if someone mods their car in an unsafe manner? You wait for it to get noticed by a cop or investigated after an accident. These are good questions, but they're not exactly new problems brought on by AVs.
You're overlooking the point of the entire thread, which is that those normal cars modded in unsafe manners still have human drivers, which are way better than computers at handling unexpected situations. Modded cars with no drivers have no such safety net. They are much more dangerous. It's not a 'new problem' but it will be magnified in new ways because of a lack of human oversight.
Again, the Tesla results, which are the worst of all AV companies, and have resulted in multiple deaths, are the closest we have to real world conditions. I think that at some point in the future AVs will be safer than human drivers, but as far as how they would do now, it's a crapshoot.
edit:: It's like saying that a child didn't get into trouble while their parent was around, so the child should be just as fine on their own.
To be clear, citing that Tesla has resulted in multiple deaths does not mean it's not better than human drivers. It's entirely possible now that if we were to magically have Tesla Autopilot-equivalent tech in every car on the road, it would be much safer.
Or maybe not. Tesla's data implies it would, but also they aren't doing a good job of sharing all their data apparently? So instead we're left with anecdotes and hypotheticals. The goal is not perfection, it's improvement.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
If by "interesting" you mean "same old problematic shit."
We already have algorithms deployed to "make sure" kids are paying attention during online classes or employees are not sleeping during Zoom meetings or people are not displaying "suspicious behavior" in stores or in airports.
These algorithms are all flawed and are especially egregious when deployed upon neurodivergent individuals and non-white persons.
I gotta tell you that I don't really think we should make exceptions for if you're neurodivergent when it impacts your ability to safely operate a vehicle. Likewise "Person is not conscious" is a pretty low bar to get over when you're talking about a person sitting in the driver's seat and frankly just about anything that looks similar should also trigger stopping the car.
Well, how can you tell now if someone mods their car in an unsafe manner? You wait for it to get noticed by a cop or investigated after an accident. These are good questions, but they're not exactly new problems brought on by AVs.
You're overlooking the point of the entire thread, which is that those normal cars modded in unsafe manners still have human drivers, which are way better than computers at handling unexpected situations. Modded cars with no drivers have no such safety net. They are much more dangerous. It's not a 'new problem' but it will be magnified in new ways because of a lack of human oversight.
Again, the Tesla results, which are the worst of all AV companies, and have resulted in multiple deaths, are the closest we have to real world conditions. I think that at some point in the future AVs will be safer than human drivers, but as far as how they would do now, it's a crapshoot.
edit:: It's like saying that a child didn't get into trouble while their parent was around, so the child should be just as fine on their own.
To be clear, citing that Tesla has resulted in multiple deaths does not mean it's not better than human drivers. It's entirely possible now that if we were to magically have Tesla Autopilot-equivalent tech in every car on the road, it would be much safer.
Or maybe not. Tesla's data implies it would, but also they aren't doing a good job of sharing all their data apparently? So instead we're left with anecdotes and hypotheticals. The goal is not perfection, it's improvement.
exactly. Each 1% of improvement saves 300+ lives and prevents 55,000 crashes.
Well, how can you tell now if someone mods their car in an unsafe manner? You wait for it to get noticed by a cop or investigated after an accident. These are good questions, but they're not exactly new problems brought on by AVs.
You're overlooking the point of the entire thread, which is that those normal cars modded in unsafe manners still have human drivers, which are way better than computers at handling unexpected situations. Modded cars with no drivers have no such safety net. They are much more dangerous. It's not a 'new problem' but it will be magnified in new ways because of a lack of human oversight.
Again, the Tesla results, which are the worst of all AV companies, and have resulted in multiple deaths, are the closest we have to real world conditions. I think that at some point in the future AVs will be safer than human drivers, but as far as how they would do now, it's a crapshoot.
edit:: It's like saying that a child didn't get into trouble while their parent was around, so the child should be just as fine on their own.
To be clear, citing that Tesla has resulted in multiple deaths does not mean it's not better than human drivers. It's entirely possible now that if we were to magically have Tesla Autopilot-equivalent tech in every car on the road, it would be much safer.
Or maybe not. Tesla's data implies it would, but also they aren't doing a good job of sharing all their data apparently? So instead we're left with anecdotes and hypotheticals. The goal is not perfection, it's improvement.
I never said they're not better than human drivers. I'm saying that this rosy-cheeked optimism that they're better is based on biased data from biased sources, and a failure to imagine negative scenarios when they're in the wild. I don't know if they're better right now, and neither do you.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
If by "interesting" you mean "same old problematic shit."
We already have algorithms deployed to "make sure" kids are paying attention during online classes or employees are not sleeping during Zoom meetings or people are not displaying "suspicious behavior" in stores or in airports.
These algorithms are all flawed and are especially egregious when deployed upon neurodivergent individuals and non-white persons.
I gotta tell you that I don't really think we should make exceptions for if you're neurodivergent when it impacts your ability to safely operate a vehicle. Likewise "Person is not conscious" is a pretty low bar to get over when you're talking about a person sitting in the driver's seat and frankly just about anything that looks similar should also trigger stopping the car.
It's less that it would impact your ability, and more that the software was probably only tested on white neurotypical men, so it'll be more likely to read anyone else as unable to drive regardless of their actual abilities.
Well, how can you tell now if someone mods their car in an unsafe manner? You wait for it to get noticed by a cop or investigated after an accident. These are good questions, but they're not exactly new problems brought on by AVs.
You're overlooking the point of the entire thread, which is that those normal cars modded in unsafe manners still have human drivers, which are way better than computers at handling unexpected situations. Modded cars with no drivers have no such safety net. They are much more dangerous. It's not a 'new problem' but it will be magnified in new ways because of a lack of human oversight.
Again, the Tesla results, which are the worst of all AV companies, and have resulted in multiple deaths, are the closest we have to real world conditions. I think that at some point in the future AVs will be safer than human drivers, but as far as how they would do now, it's a crapshoot.
edit:: It's like saying that a child didn't get into trouble while their parent was around, so the child should be just as fine on their own.
To be clear, citing that Tesla has resulted in multiple deaths does not mean it's not better than human drivers. It's entirely possible now that if we were to magically have Tesla Autopilot-equivalent tech in every car on the road, it would be much safer.
Or maybe not. Tesla's data implies it would, but also they aren't doing a good job of sharing all their data apparently? So instead we're left with anecdotes and hypotheticals. The goal is not perfection, it's improvement.
I never said they're not better than human drivers. I'm saying that this rosy-cheeked optimism that they're better is based on biased data from biased sources, and a failure to imagine negative scenarios when they're in the wild. I don't know if they're better right now, and neither do you.
I must have misinterpreted your post, sorry. We agree that more data is required
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
If by "interesting" you mean "same old problematic shit."
We already have algorithms deployed to "make sure" kids are paying attention during online classes or employees are not sleeping during Zoom meetings or people are not displaying "suspicious behavior" in stores or in airports.
These algorithms are all flawed and are especially egregious when deployed upon neurodivergent individuals and non-white persons.
I gotta tell you that I don't really think we should make exceptions for if you're neurodivergent when it impacts your ability to safely operate a vehicle. Likewise "Person is not conscious" is a pretty low bar to get over when you're talking about a person sitting in the driver's seat and frankly just about anything that looks similar should also trigger stopping the car.
It's less that it would impact your ability, and more that the software was probably only tested on white neurotypical men, so it'll be more likely to read anyone else as unable to drive regardless of their actual abilities.
I guess this is just ignorance on my part. In my brain, I'm thinking that "driver is slumped over the wheel or no longer visible" is a problem regardless, but I guess we could come up with some circumstances where a person is awake and alert but looks asleep?
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
If by "interesting" you mean "same old problematic shit."
We already have algorithms deployed to "make sure" kids are paying attention during online classes or employees are not sleeping during Zoom meetings or people are not displaying "suspicious behavior" in stores or in airports.
These algorithms are all flawed and are especially egregious when deployed upon neurodivergent individuals and non-white persons.
I gotta tell you that I don't really think we should make exceptions for if you're neurodivergent when it impacts your ability to safely operate a vehicle. Likewise "Person is not conscious" is a pretty low bar to get over when you're talking about a person sitting in the driver's seat and frankly just about anything that looks similar should also trigger stopping the car.
It's less that it would impact your ability, and more that the software was probably only tested on white neurotypical men, so it'll be more likely to read anyone else as unable to drive regardless of their actual abilities.
I guess this is just ignorance on my part. In my brain, I'm thinking that "driver is slumped over the wheel or no longer visible" is a problem regardless, but I guess we could come up with some circumstances where a person is awake and alert but looks asleep?
For example, several years ago there's was the fancy new camera tech that could tell if the person blinked during the photo or not. Except for certain people of Asian descent where it thought their eyes were always blinking. What if your car just thought your eyes were always closed?
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
If by "interesting" you mean "same old problematic shit."
We already have algorithms deployed to "make sure" kids are paying attention during online classes or employees are not sleeping during Zoom meetings or people are not displaying "suspicious behavior" in stores or in airports.
These algorithms are all flawed and are especially egregious when deployed upon neurodivergent individuals and non-white persons.
I gotta tell you that I don't really think we should make exceptions for if you're neurodivergent when it impacts your ability to safely operate a vehicle. Likewise "Person is not conscious" is a pretty low bar to get over when you're talking about a person sitting in the driver's seat and frankly just about anything that looks similar should also trigger stopping the car.
It's less that it would impact your ability, and more that the software was probably only tested on white neurotypical men, so it'll be more likely to read anyone else as unable to drive regardless of their actual abilities.
I guess this is just ignorance on my part. In my brain, I'm thinking that "driver is slumped over the wheel or no longer visible" is a problem regardless, but I guess we could come up with some circumstances where a person is awake and alert but looks asleep?
For example, several years ago there's was the fancy new camera tech that could tell if the person blinked during the photo or not. Except for certain people of Asian descent where it thought their eyes were always blinking. What if your car just thought your eyes were always closed?
Or the neural net was never given training data on black people, so it just doesn't see them. Bam car never moves no driver. Lots of fail states from bad design and garbage in/garbage out.
I wonder inwardly what the statistics will look like if everyone has AVs
we know that there are more good drivers than bad right now by a fairly huge margin right? Why do we think that the myriad of problems AVs will introduce that will face all drivers across the entire country will result in less accidents?
Lets say that the country moves to all AVs in 15 years. Heck, lets say 25 years. Has there ever been an effort that would have to be more coordinated (and successfully coordinated!) in american history? Matching software between dozens of car manufacturers from several countries and encompassing hundreds of models across hundreds of thousnads of miles in places that still, in the year of our lord 2021, don't have internet .... I just don't see it happening.
and I could go on with a hundred other factors that would have to be accounted for
Well, how can you tell now if someone mods their car in an unsafe manner? You wait for it to get noticed by a cop or investigated after an accident. These are good questions, but they're not exactly new problems brought on by AVs.
You're overlooking the point of the entire thread, which is that those normal cars modded in unsafe manners still have human drivers, which are way better than computers at handling unexpected situations. Modded cars with no drivers have no such safety net. They are much more dangerous. It's not a 'new problem' but it will be magnified in new ways because of a lack of human oversight.
Again, the Tesla results, which are the worst of all AV companies, and have resulted in multiple deaths, are the closest we have to real world conditions. I think that at some point in the future AVs will be safer than human drivers, but as far as how they would do now, it's a crapshoot.
edit:: It's like saying that a child didn't get into trouble while their parent was around, so the child should be just as fine on their own.
To be clear, citing that Tesla has resulted in multiple deaths does not mean it's not better than human drivers. It's entirely possible now that if we were to magically have Tesla Autopilot-equivalent tech in every car on the road, it would be much safer.
Or maybe not. Tesla's data implies it would, but also they aren't doing a good job of sharing all their data apparently? So instead we're left with anecdotes and hypotheticals. The goal is not perfection, it's improvement.
I'll admit I haven't been following the various competitors and researchers as closely as I would like, but in the absence of trustworthy external verification, I wouldn't trust a damn thing that Tesla or Uber publishes. These two companies have been cobbling together driver-assist features and releasing cars into the wild with enough automation to lull untrained drivers into a false sense of security but not enough automation to ensure safety when the driver is completely checked out.
To contrast, Waymo/Google spent $texas and years building an entire closed-course simulated city for testing their vehicles, then slowly moved to limited tests with trained drivers in the wild, and then have been incrementally increasing the breadth and depth of those tests. They've also been working alongside UC Berkeley AI researchers who publish papers in peer-reviewed (and other) journals about AI safety, and they've released their city scanning data to the public so anybody who wants to use it in AI research may do so.
There's a reason that the most egregious safety violations (and deaths!) are coming from Tesla and Uber vehicles, while the worst we hear about Waymo are minor fender-benders with no/negligible injuries, mostly in situations where some other vehicle or pedestrian did something like run out in the middle of the road that would have also been difficult for a human driver to react to.
every person who doesn't like an acquired taste always seems to think everyone who likes it is faking it. it should be an official fallacy.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
If by "interesting" you mean "same old problematic shit."
We already have algorithms deployed to "make sure" kids are paying attention during online classes or employees are not sleeping during Zoom meetings or people are not displaying "suspicious behavior" in stores or in airports.
These algorithms are all flawed and are especially egregious when deployed upon neurodivergent individuals and non-white persons.
I gotta tell you that I don't really think we should make exceptions for if you're neurodivergent when it impacts your ability to safely operate a vehicle. Likewise "Person is not conscious" is a pretty low bar to get over when you're talking about a person sitting in the driver's seat and frankly just about anything that looks similar should also trigger stopping the car.
That’s not what it means though
What is being pointed out is the potential that the system misreads the physical cues of a neurodivergent person because they will not necessarily cue the same as a neurolotypical person. And that the systems also will not be able to see black people because, again, automated camera systems designed to recognize people instantly fail to understand that a black person is, well, a person.
In other words: the system will generate a lot of false positives and false negatives because they’re designed and tested on a very small subset of people and they’re not particularly reliable.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
If by "interesting" you mean "same old problematic shit."
We already have algorithms deployed to "make sure" kids are paying attention during online classes or employees are not sleeping during Zoom meetings or people are not displaying "suspicious behavior" in stores or in airports.
These algorithms are all flawed and are especially egregious when deployed upon neurodivergent individuals and non-white persons.
I gotta tell you that I don't really think we should make exceptions for if you're neurodivergent when it impacts your ability to safely operate a vehicle. Likewise "Person is not conscious" is a pretty low bar to get over when you're talking about a person sitting in the driver's seat and frankly just about anything that looks similar should also trigger stopping the car.
It's less that it would impact your ability, and more that the software was probably only tested on white neurotypical men, so it'll be more likely to read anyone else as unable to drive regardless of their actual abilities.
I guess this is just ignorance on my part. In my brain, I'm thinking that "driver is slumped over the wheel or no longer visible" is a problem regardless, but I guess we could come up with some circumstances where a person is awake and alert but looks asleep?
For example, several years ago there's was the fancy new camera tech that could tell if the person blinked during the photo or not. Except for certain people of Asian descent where it thought their eyes were always blinking. What if your car just thought your eyes were always closed?
Or the neural net was never given training data on black people, so it just doesn't see them. Bam car never moves no driver. Lots of fail states from bad design and garbage in/garbage out.
A non-AI version of this actually happened with hands free taps not activating for PoC.
In this moment, I am euphoric. Not because of any phony god’s blessing. But because, I am enlightened by my intelligence.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
Your care can handle lane changes yes
But how does it recognize the shoulder from just another lane? That is, what would keep it from deciding “this is the shoulder, I stop now” and parking itself in an actively traveled lane
Or how do you keep it from thinking it’s now in a lane should someone pull onto the shoulder behind you?
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
Your care can handle lane changes yes
But how does it recognize the shoulder from just another lane? That is, what would keep it from deciding “this is the shoulder, I stop now” and parking itself in an actively traveled lane
Or how do you keep it from thinking it’s now in a lane should someone pull onto the shoulder behind you?
The same way you do, by seeing a solid line and/or the edge of the road?
While racing light mechs, your Urbanmech comes in second place, but only because it ran out of ammo.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
My car right now brakes me if something kicks up from the road and blocks the sensor or camera. It has run me off the road once nearly killing me (I was in no danger until the car made me in danger) and forced me to stop twice for no discernible reason.
It also suggested I get a cup of coffee once (I was not swerving or anything). I turned that 'feature' off
(which is another fun thing I can't wait for! Pay a premium to not have ads in my car!)
edit: in a perfect human society, I think AVs would be wonderful for the record!
In our human societies, I will bet every cent I have that it will be a massive shit show that everyone can see coming a hundred miles away.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
Your care can handle lane changes yes
But how does it recognize the shoulder from just another lane? That is, what would keep it from deciding “this is the shoulder, I stop now” and parking itself in an actively traveled lane
Or how do you keep it from thinking it’s now in a lane should someone pull onto the shoulder behind you?
I'd assume it would avoid these things via the same way it recognizes the shoulder/ parking lanes currently and doesn't change lanes onto it or try and drive on it? I drive on roads where people park on the shoulder, It doesn't get confused by this. This is very much illustrates the point I was trying to make though, I'm not sure why you'd think these are some sort of gotchas that would need to be reinvented.
People seem to be under the impression that the current systems are like gen 1 TomToms just wired into the steering controls.
Tonnes of people die to cars each year, so make them AI driven and you eliminate a whole class of crashes. However AIs need tonnes and tonnes of training data from many difference circumstances to work in all conditions. How many AVs recognise camels, or a flooded road, or oil slicks etc etc. Some probably do, but how many? and do you know what your AV can't do?
AI is bad at edge cases. Driving is chock full of edge cases, at least the dangerous parts are.
In this moment, I am euphoric. Not because of any phony god’s blessing. But because, I am enlightened by my intelligence.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
Your care can handle lane changes yes
But how does it recognize the shoulder from just another lane? That is, what would keep it from deciding “this is the shoulder, I stop now” and parking itself in an actively traveled lane
Or how do you keep it from thinking it’s now in a lane should someone pull onto the shoulder behind you?
The same way you do, by seeing a solid line and/or the edge of the road?
There’s a lot that goes into seeing a solid line or the edge of the road and determining then that you are no longer on road
It seems simple because we don’t regularly have to think about how we do these things that are extraordinarily complex cognitive tasks
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
Your care can handle lane changes yes
But how does it recognize the shoulder from just another lane? That is, what would keep it from deciding “this is the shoulder, I stop now” and parking itself in an actively traveled lane
Or how do you keep it from thinking it’s now in a lane should someone pull onto the shoulder behind you?
The same way you do, by seeing a solid line and/or the edge of the road?
you mean those things that don't exist on many roads or that once existed but haven't been maintained in decades?
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
My car right now brakes me if something kicks up from the road and blocks the sensor or camera. It has run me off the road once nearly killing me (I was in no danger until the car made me in danger) and forced me to stop twice for no discernible reason.
It also suggested I get a cup of coffee once (I was not swerving or anything). I turned that 'feature' off
(which is another fun thing I can't wait for! Pay a premium to not have ads in my car!)
edit: in a perfect human society, I think AVs would be wonderful for the record!
In our human societies, I will bet every cent I have that it will be a massive shit show that everyone can see coming a hundred miles away.
It’s like having a backseat driver who can wrest control away from you without warning AND they’re super incompetent?
So I think AVs are the wrong solution to the issue of car crashes and traffic fatalities. Or at least they should only be a part of the solution.
Inter and intra city public transit could stand to be massively improved just about anywhere in America, and this would dramatically reduce the need to drive, and by extension traffic fatalities. Less pollution is also a nice bonus. Increased diversity of road options (bike paths / routes, busses, escooter schemes etc) also has a similar effect.
I guess I'm asking is what could we do to limit the role of the car, and would that make AVs easier to build/make safe?
In this moment, I am euphoric. Not because of any phony god’s blessing. But because, I am enlightened by my intelligence.
WASHINGTON (AP) — Congress has created a new requirement for automakers: Find a high-tech way to keep drunken people from driving cars.
The cameras make sure a driver is watching the road, and they look for signs of drowsiness, loss of consciousness or impairment.
If signs are spotted, the cars will warn the driver, and if the behavior persists, the car would turn on its hazard lights, slow down and pull to the side of the road.
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
My car right now brakes me if something kicks up from the road and blocks the sensor or camera. It has run me off the road once nearly killing me (I was in no danger until the car made me in danger) and forced me to stop twice for no discernible reason.
It also suggested I get a cup of coffee once (I was not swerving or anything). I turned that 'feature' off
(which is another fun thing I can't wait for! Pay a premium to not have ads in my car!)
edit: in a perfect human society, I think AVs would be wonderful for the record!
In our human societies, I will bet every cent I have that it will be a massive shit show that everyone can see coming a hundred miles away.
It’s like having a backseat driver who can wrest control away from you without warning AND they’re super incompetent?
that is so much better than I was describing
yes, this is it exactly
edit:
but don't worry, once everyone has some form of the software that's created and implemented by everyone's lowest bidder it'll be fine
Posts
As say they did operate them in a circuit so you can use a bike or walk to the stops it would make the ideal of owning rather inconvenient
There is a youtube channel from someone that lives in the netherlands that talks about bad roads/infrastructure and ways to fix it called Not Just bikes
This depends pretty heavily on country, to be fair. The average age of Car /Light truck in the US is currently around 11 years, while in NZ it's currently around 14. Australia is slightly ahead at 10,, and Canada is at 9. But, even assuming all manufacturers rolled this out in all models, which is wildly optimistic, it still seems like it will take longer to roll through.
Anyways I’m currently working on my second AV project. On both projects, my role was in brake control software development and calibration. So while I don’t directly work on AV control logic (“Should I apply the brakes right now?”) I do spend a lot of time interfacing with it and reacting to its commands (“Computer says decel at 5 m/s2, apply 6 mPa brake pressure.”)
Project 1 was for an startup making a driverless delivery vehicle. It was an utter shitshow, the highlight of which was when we delivered brake software to the client and had them acknowledge that it was for closed course use only and unsafe for public roads. They immediately put it on public roads anyway because fuck you, Silicon Valley don’t care.
Project 2 is for a very respected Japanese OEM building prototypes for closed course testing in a safe manner. It is also a shitshow, has been in development for 5 years so far and will be ready for 40 kph closed course testing this month. Maybe. Mostly it’s notorious around the test track for things like suddenly losing steering control.
Anyway this stuff is a long way off yet, and I sure as shit wouldn’t trust a word of what any techbro company is touting about AV performance. The legacy automotive OEMs are at least attempting to do it safely.
You can't give someone a pirate ship in one game, and then take it back in the next game. It's rude.
Don't care if they are level 2 or level 5 AVs ASAP.
Especially since trains can provably gets to 2-300 km/h. With existing technology. From the '70s.
Here's some interesting stuff
At least part of that will, I assume, understanding which part of the road is the side
How will the car know which lane the driver is in, the locations of other cars, where the shoulder is, what is on the shoulder, how wide the shoulder is, if there is a shoulder, and about 50 other things?
Because if they're relying on existing cameras and tech that are unfortunately implanted in every new car these days, they have a LONG way to go.
edit: and on a more selfish note, I hate paying for 'features' that I don't want, don't use, and don't work. I also have to pay more insurance because they cost more to fix when they break. I'm not a rich man and it totally sucked when my monthly insurance nearly tripled just because I finally bought a new car.
is it?
Perry's commuter bullet between Dallas, Austin, and SA would have been awesome but the amount of eminent domain legal battling just to get the land made it impossible. Everybody in the cities wants high speed rail between them, and they also want all the people in between to give up their property in service to that goal.
It's a hard sell, even here where conditions are perfect.
The problem with rail is always what happens when you get there. Fixing car-based design at the local level kinda has to come before connecting hubs.
I tried googling public transit from my house to downtown Raleigh and google just said, "No". The closest bus stop is 3 miles away at a park-and-ride place for a technical college.
Did you buy a new car or did finance a new car?
I purchased it outright.
edit: I also called several other insurance companies to compare and all were higher than the company I was (and still am) already with (GEICO if it matters).
edit2: and the specific reason for the massive increase was that newer tech is more expensive to replace. (I went from a 2004 Saturn Ion to a 2020 Hyundai Elantra) ($45 a month to $120 a month with the exact same deductible) (100% clean driving record)
You're overlooking the point of the entire thread, which is that those normal cars modded in unsafe manners still have human drivers, which are way better than computers at handling unexpected situations. Modded cars with no drivers have no such safety net. They are much more dangerous. It's not a 'new problem' but it will be magnified in new ways because of a lack of human oversight.
Again, the Tesla results, which are the worst of all AV companies, and have resulted in multiple deaths, are the closest we have to real world conditions. I think that at some point in the future AVs will be safer than human drivers, but as far as how they would do now, it's a crapshoot.
edit:: It's like saying that a child didn't get into trouble while their parent was around, so the child should be just as fine on their own.
If by "interesting" you mean "same old problematic shit."
We already have algorithms deployed to "make sure" kids are paying attention during online classes or employees are not sleeping during Zoom meetings or people are not displaying "suspicious behavior" in stores or in airports.
These algorithms are all flawed and are especially egregious when deployed upon neurodivergent individuals and non-white persons.
To be clear, citing that Tesla has resulted in multiple deaths does not mean it's not better than human drivers. It's entirely possible now that if we were to magically have Tesla Autopilot-equivalent tech in every car on the road, it would be much safer.
Or maybe not. Tesla's data implies it would, but also they aren't doing a good job of sharing all their data apparently? So instead we're left with anecdotes and hypotheticals. The goal is not perfection, it's improvement.
3DS Friend Code: 3110-5393-4113
Steam profile
I gotta tell you that I don't really think we should make exceptions for if you're neurodivergent when it impacts your ability to safely operate a vehicle. Likewise "Person is not conscious" is a pretty low bar to get over when you're talking about a person sitting in the driver's seat and frankly just about anything that looks similar should also trigger stopping the car.
exactly. Each 1% of improvement saves 300+ lives and prevents 55,000 crashes.
I never said they're not better than human drivers. I'm saying that this rosy-cheeked optimism that they're better is based on biased data from biased sources, and a failure to imagine negative scenarios when they're in the wild. I don't know if they're better right now, and neither do you.
It's less that it would impact your ability, and more that the software was probably only tested on white neurotypical men, so it'll be more likely to read anyone else as unable to drive regardless of their actual abilities.
3DS Friend Code: 3110-5393-4113
Steam profile
I must have misinterpreted your post, sorry. We agree that more data is required
3DS Friend Code: 3110-5393-4113
Steam profile
I guess this is just ignorance on my part. In my brain, I'm thinking that "driver is slumped over the wheel or no longer visible" is a problem regardless, but I guess we could come up with some circumstances where a person is awake and alert but looks asleep?
For example, several years ago there's was the fancy new camera tech that could tell if the person blinked during the photo or not. Except for certain people of Asian descent where it thought their eyes were always blinking. What if your car just thought your eyes were always closed?
3DS Friend Code: 3110-5393-4113
Steam profile
Or the neural net was never given training data on black people, so it just doesn't see them. Bam car never moves no driver. Lots of fail states from bad design and garbage in/garbage out.
we know that there are more good drivers than bad right now by a fairly huge margin right? Why do we think that the myriad of problems AVs will introduce that will face all drivers across the entire country will result in less accidents?
Lets say that the country moves to all AVs in 15 years. Heck, lets say 25 years. Has there ever been an effort that would have to be more coordinated (and successfully coordinated!) in american history? Matching software between dozens of car manufacturers from several countries and encompassing hundreds of models across hundreds of thousnads of miles in places that still, in the year of our lord 2021, don't have internet .... I just don't see it happening.
and I could go on with a hundred other factors that would have to be accounted for
I'll admit I haven't been following the various competitors and researchers as closely as I would like, but in the absence of trustworthy external verification, I wouldn't trust a damn thing that Tesla or Uber publishes. These two companies have been cobbling together driver-assist features and releasing cars into the wild with enough automation to lull untrained drivers into a false sense of security but not enough automation to ensure safety when the driver is completely checked out.
To contrast, Waymo/Google spent $texas and years building an entire closed-course simulated city for testing their vehicles, then slowly moved to limited tests with trained drivers in the wild, and then have been incrementally increasing the breadth and depth of those tests. They've also been working alongside UC Berkeley AI researchers who publish papers in peer-reviewed (and other) journals about AI safety, and they've released their city scanning data to the public so anybody who wants to use it in AI research may do so.
There's a reason that the most egregious safety violations (and deaths!) are coming from Tesla and Uber vehicles, while the worst we hear about Waymo are minor fender-benders with no/negligible injuries, mostly in situations where some other vehicle or pedestrian did something like run out in the middle of the road that would have also been difficult for a human driver to react to.
the "no true scotch man" fallacy.
You and other people in this thread keep making these very hard declarative statements like this...and like what the hell are you basing this on?
My car right now:
Passes people on the left lane, if I am in the leftmost lane and not passing moves me one lane right, but doesn't move me right if I am not in the leftmost lane.
It knows if I am in the correct lane(lane 4 out of 6) for the split I am approaching for my programmed course.
It merges me out of lanes adjacent to construction barrels, but not away from standard guard rails, etc.
It parallel parks itself-and doesn't curb its wheels.
This is a 3 year old car, built on 5+ year old sensor hardware. The idea that current/near future cars somehow aren't able to handle pulling off to the side of the road...
I'm struggling to think of another example (The first iPhones maybe?) of generally available tech where people are so confident it can't do...shit it already can do.
That’s not what it means though
What is being pointed out is the potential that the system misreads the physical cues of a neurodivergent person because they will not necessarily cue the same as a neurolotypical person. And that the systems also will not be able to see black people because, again, automated camera systems designed to recognize people instantly fail to understand that a black person is, well, a person.
In other words: the system will generate a lot of false positives and false negatives because they’re designed and tested on a very small subset of people and they’re not particularly reliable.
EDIT: RECOGNIZE WORDS, PHONE
A non-AI version of this actually happened with hands free taps not activating for PoC.
Your care can handle lane changes yes
But how does it recognize the shoulder from just another lane? That is, what would keep it from deciding “this is the shoulder, I stop now” and parking itself in an actively traveled lane
Or how do you keep it from thinking it’s now in a lane should someone pull onto the shoulder behind you?
*gets off the Jersey Turnpike*
*drives a quarter mile on some back road*
*gets back on the Jersey Turnpike*
The same way you do, by seeing a solid line and/or the edge of the road?
My car right now brakes me if something kicks up from the road and blocks the sensor or camera. It has run me off the road once nearly killing me (I was in no danger until the car made me in danger) and forced me to stop twice for no discernible reason.
It also suggested I get a cup of coffee once (I was not swerving or anything). I turned that 'feature' off
(which is another fun thing I can't wait for! Pay a premium to not have ads in my car!)
edit: in a perfect human society, I think AVs would be wonderful for the record!
In our human societies, I will bet every cent I have that it will be a massive shit show that everyone can see coming a hundred miles away.
I'd assume it would avoid these things via the same way it recognizes the shoulder/ parking lanes currently and doesn't change lanes onto it or try and drive on it? I drive on roads where people park on the shoulder, It doesn't get confused by this. This is very much illustrates the point I was trying to make though, I'm not sure why you'd think these are some sort of gotchas that would need to be reinvented.
People seem to be under the impression that the current systems are like gen 1 TomToms just wired into the steering controls.
AI is bad at edge cases. Driving is chock full of edge cases, at least the dangerous parts are.
There’s a lot that goes into seeing a solid line or the edge of the road and determining then that you are no longer on road
It seems simple because we don’t regularly have to think about how we do these things that are extraordinarily complex cognitive tasks
you mean those things that don't exist on many roads or that once existed but haven't been maintained in decades?
It’s like having a backseat driver who can wrest control away from you without warning AND they’re super incompetent?
Inter and intra city public transit could stand to be massively improved just about anywhere in America, and this would dramatically reduce the need to drive, and by extension traffic fatalities. Less pollution is also a nice bonus. Increased diversity of road options (bike paths / routes, busses, escooter schemes etc) also has a similar effect.
I guess I'm asking is what could we do to limit the role of the car, and would that make AVs easier to build/make safe?
that is so much better than I was describing
yes, this is it exactly
edit:
but don't worry, once everyone has some form of the software that's created and implemented by everyone's lowest bidder it'll be fine