The wamo workers when they get stuck look at all the cameras, then plot a safe route the car drives it’s self out of the confusion via. They never remotely drive them.
It’s even less than that. The act of plotting a course (which we called Augmented Trajectories), was used very sparingly, and mostly just to do “illegal” maneuvers like crossing a double yellow line in order to get around debris in the road. The Waymo won’t (knowingly) break the law on its own, even in those exceptional situations like road obstructions, so we could either tell the car “this road is inaccessible due to the obstruction” and let it try to make a 30-point u-turn and reroute itself 8 miles in another direction, or we can tell the car “let me direct you through this tiny lil one-time crime”. ATs are very limited in scope; we could basically plot out several points in the car’s path, and then tell it to drive itself to those points while ignoring other traffic rules, and then the car will complete that course while steering itself along the path you’ve plotted. You can only set a max distance of like 20 meters or so before having to make the car pull over and plot a new AT again, and the car will only ever go about 3 MPH during an AT. Even during an AT, the car will still refuse to drive over anything it doesn’t recognize as safe and will not collide itself with anything at all.
Almost all of the remote dispatcher’s job is just identifying objects (usually road signs, traffic signals, and other vehicles) that the Waymo isn’t able to immediately identify. No remote driving, and very little manual course-plotting is being done by humans.
Source: was a Waymo remote dispatcher for a year and identified tens of thousands of road objects, and conducted maybe 4 ATs in total. They’re very rare. We mostly had to use them in road work areas, where the workers would leave a very complicated temporary path with road cones that the car just couldn’t figure out, or where the cones had fallen over.
On a completely unrelated note, this type of online conversation is what makes me most pessimistic about LLMs. I’m fairly certain Chozo is a human, but as time goes on it will become much more likely companies will have bots crawling all social media looking for conversations like this as a standard part of their PR team. It will make it impossible to discern truth from hallucination and ultimately just erode trust in everything. It’s such a bleak future.
Almost all of the remote dispatcher’s job is just identifying objects (usually road signs, traffic signals, and other vehicles) that the Waymo isn’t able to immediately identify. No remote driving, and very little manual course-plotting is being done by humans.
So basically Waymo needs humans to do what humans do daily while driving.
I do get your point, but what’s the point of this tech if it cannot actually do what human drivers constantly do? you mention road work, in Canada, basically if it’s not winter (hard to drive because of conditions and almost complete lack of visibility of the road and signs), it’s road construction time…
in Canada, basically if it’s not winter (hard to drive because of conditions and almost complete lack of visibility of the road and signs), it’s road construction time…
You’ll note that Waymo is really only launched in places that tend to have mostly pleasant weather most of the time, because the tech is relatively new (but probably already much better than human drivers most of the time - humans are broadly awful at driving so that doesn’t take much) and they don’t want to try it against the much more difficult conditions in many places yet.
100% and to be fair, that’s how you develop the tech properly. My pet peeve is when they pass the “first hurdle” (overly simplifying here) and then call it “fully autonomous driving”
Also, I cannot imagine a real autonomous driver would be worse than humans because, as you say, humans are broadly awful at driving. However, since there is ZERO accountability with any of these companies in the USA, I need to see some independently ran studies and not just the propaganda the companies peddle and from which their bottom line depends
I get what you’re saying, and I don’t have the data handy to provide the necessary context, but it’s important to keep in mind that these “what am I looking at?” requests from the cars will represent something like ~0.0001% of all scanned objects on the road. Almost all of the time, the car is confidently identifying everything in its path without any human assistance. Almost every request that the dispatchers would receive to identify was already correctly identified by the car by the time the dispatcher presses their buttons. Most of these requests would be something like “Is this a fire hydrant?”, and that request was automatically sent on the first “frame” where the car couldn’t identify the object; more often than not, the car has figured it out by the very next “frame” and has made whatever necessary adjustments are needed to correct its course before dispatchers even respond.
Of course, I have no way to validate any of this, as any numbers I can provide are just made up based on my own experience doing this job around 7 years ago, and I don’t believe Waymo publishes any actual statistics in this regard, so you’ll just have to take the word of a random guy on the internet for however much its worth. :)
For what it’s worth, I will say that self-driving cars are about the only technology that genuinely excite me these days and some of the things they’re capable of are not talked about nearly enough (you hardly ever hear about the fact that Waymo has autonomous semi-trucks on public roads). With the exception of Tesla, who have been seen as mostly a joke by the rest of the industry, pretty much everybody in this particular tech space are doing some very impressive things.
I don’t suck big tech dick that often, but the self-driving car industry has some good techdick right now.
Look, I do understand your points but I will not give any company a pass on this. Are these cars “fully autonomous”? no they are not… if the dude is the Philippines is not around, sooner than later the cars will stop or hurt someone; and this is if we still let them stay in their small, well mapped, cozy area… take them someplace else and the car will be mostly useless
Personally, I am not convinced this tech is any good for anything other than making unwalkable cities even more expensive to navigate. I do of course see the safety aspect (although the stats are very shady at the moment) but again, this is only true and valid if these cars can ever be autonomous and not just ride a virtual track.
Why do you need anything beyond “riding a virtual track”, though? The point of a car is to go from A to B, and a track is literally the most effective way to do that (which is why we need more trains in the first place, but that’s an entirely separate rant I could go on about).
IMO, at a certain point you reach a level of autonomy that is sufficient enough. It doesn’t need to know every nook and cranny of obscure roads because realistically, nobody’s going there to begin with. And even then in those situations where you did need to go to a location that wasn’t pre-mapped, it’s still likely to get you 99.9% of the way to your destination, anyway. Waymos are also constantly scanning and remapping their surroundings in real time (to adjust for things like road work, detours, and other blockages), so it stands to reason that if you had to manually engage the car after reaching the end of its mapped area, it would eventually develop a map of the new areas with a high-enough level of confidence to navigate on its own, over time.
I also think you have the wrong takeaway regarding the remote dispatchers being required. I think it’s a good thing that they exist, because that means that Waymo doesn’t trust their own machines to be 100% accurate. Tesla incorrectly does, which is why their self-driving cars have an exponentially higher rate of at-fault collisions than any of their competitors. Waymo at least understands the realistic limitations of their tech, and has a safety net in place for when those limitations are reached. I think that’s the correct way to do it, as you lead up toward an actual finished product.
Why do you need anything beyond “riding a virtual track”, though? The point of a car is to go from A to B, and a track is literally the most effective way to do that (which is why we need more trains in the first place, but that’s an entirely separate rant I could go on about).
I don’t… but if we have not truly automated literal tracks (trains), I have very little trust in the virtual version being “ready”
IMO, at a certain point you reach a level of autonomy that is sufficient enough.
Agreed, and to me, that means ZERO human intervention (at least to be able to call it “fully autonomous”). The only appropriate intervention is when the car detects it is breaking down and must stop and wait for a tow truck
It doesn’t need to know every nook and cranny of obscure roads because realistically, nobody’s going there to begin with
And I think here is the gist… I do not want or care for robotaxis in my city. That is just the same bullshit with more steps to continue down the line of unwalkable cities where you depend on a car (which is now a rent seeking service). I want truly fully autonomous driving so eventually we can get to the point where a fleet of truly autonomous cars covers the last mile where mass public transit cannot. Therefore, I do need tech that can navigate low traffic neighborhoods, not just go around downtown
Waymo at least understands the realistic limitations of their tech
If they did, they wouldn’t call their tech “Fully autonomous”… I mean, come on, why do you still fight for the lies they try to feed you. It’s the same everywhere, with food “added vitamins” without telling you their processing took out so many nutrients you are basically eating air even with the “added vitamins”
If Waymo had called their tech “minimally assisted driving” I would have the respect for them you seek from me
The point is you can have say 100 people in an office someone manage 5,000 cars instead of having 5,000 drivers operating those 5,000 cars and it’s a way more efficient.
Also well built self driving cars like the Waymo ones are actually significantly safer than human drivers.
For a long time now, humans have been utilising machines and robots to perform tasks that humans would otherwise have had to perform - particularly when it comes to manufacturing. Occasionally, one of those machines will encounter an error or issue, and a human will need to intervene to either fix the machine or guide the process before it can resume its task.
These advancements have allowed a very small number of humans to oversee what would have otherwise required hundreds, potentially thousands of humans to do independently. Even when humans were performing these tasks, they would occasionally need someone with more specialisation or experience to help them with completing such a task.
The point of this tech - the point of all tech that I’ve described above - is to reduce the amount of humans needed to produce a given result. It’s a reduction in the need for labour. It’s a matter of efficiency, not a complete replacement of any need for any human to ever intervene under any circumstances. Under communism, or even well-fleshed out socialism, it would free up humans from having to perform menial labour to instead pursue their passions, work on vital human skills, create and consume art… just live better lives.
Now, would I ride in a self-driving vehicle? Certainly not yet. The tech still needs time to develop before I’m confident in it. As time goes on, the need for human intervention will continue to drop, and hopefully my confidence will grow.
Being able to autonomously drive itself in the vast majority of scenarios is a big fucking deal.
In a controlled limited area, with human intervention… hmmm again, not sure I would use the words “Fully autonomous driving” for this but English is not my first language so maybe it is like the “unlimited” plans from cell phone companies
This is different. It’s like playing a racing video game via trading cards. They’re giving the car directions just like a person might give a driver directions.
The wamo workers when they get stuck look at all the cameras, then plot a safe route the car drives it’s self out of the confusion via. They never remotely drive them.
It’s even less than that. The act of plotting a course (which we called Augmented Trajectories), was used very sparingly, and mostly just to do “illegal” maneuvers like crossing a double yellow line in order to get around debris in the road. The Waymo won’t (knowingly) break the law on its own, even in those exceptional situations like road obstructions, so we could either tell the car “this road is inaccessible due to the obstruction” and let it try to make a 30-point u-turn and reroute itself 8 miles in another direction, or we can tell the car “let me direct you through this tiny lil one-time crime”. ATs are very limited in scope; we could basically plot out several points in the car’s path, and then tell it to drive itself to those points while ignoring other traffic rules, and then the car will complete that course while steering itself along the path you’ve plotted. You can only set a max distance of like 20 meters or so before having to make the car pull over and plot a new AT again, and the car will only ever go about 3 MPH during an AT. Even during an AT, the car will still refuse to drive over anything it doesn’t recognize as safe and will not collide itself with anything at all.
Almost all of the remote dispatcher’s job is just identifying objects (usually road signs, traffic signals, and other vehicles) that the Waymo isn’t able to immediately identify. No remote driving, and very little manual course-plotting is being done by humans.
Source: was a Waymo remote dispatcher for a year and identified tens of thousands of road objects, and conducted maybe 4 ATs in total. They’re very rare. We mostly had to use them in road work areas, where the workers would leave a very complicated temporary path with road cones that the car just couldn’t figure out, or where the cones had fallen over.
Thank you for the insight and write up.
On a completely unrelated note, this type of online conversation is what makes me most pessimistic about LLMs. I’m fairly certain Chozo is a human, but as time goes on it will become much more likely companies will have bots crawling all social media looking for conversations like this as a standard part of their PR team. It will make it impossible to discern truth from hallucination and ultimately just erode trust in everything. It’s such a bleak future.
So basically Waymo needs humans to do what humans do daily while driving.
I do get your point, but what’s the point of this tech if it cannot actually do what human drivers constantly do? you mention road work, in Canada, basically if it’s not winter (hard to drive because of conditions and almost complete lack of visibility of the road and signs), it’s road construction time…
You’ll note that Waymo is really only launched in places that tend to have mostly pleasant weather most of the time, because the tech is relatively new (but probably already much better than human drivers most of the time - humans are broadly awful at driving so that doesn’t take much) and they don’t want to try it against the much more difficult conditions in many places yet.
100% and to be fair, that’s how you develop the tech properly. My pet peeve is when they pass the “first hurdle” (overly simplifying here) and then call it “fully autonomous driving”
Also, I cannot imagine a real autonomous driver would be worse than humans because, as you say, humans are broadly awful at driving. However, since there is ZERO accountability with any of these companies in the USA, I need to see some independently ran studies and not just the propaganda the companies peddle and from which their bottom line depends
I get what you’re saying, and I don’t have the data handy to provide the necessary context, but it’s important to keep in mind that these “what am I looking at?” requests from the cars will represent something like ~0.0001% of all scanned objects on the road. Almost all of the time, the car is confidently identifying everything in its path without any human assistance. Almost every request that the dispatchers would receive to identify was already correctly identified by the car by the time the dispatcher presses their buttons. Most of these requests would be something like “Is this a fire hydrant?”, and that request was automatically sent on the first “frame” where the car couldn’t identify the object; more often than not, the car has figured it out by the very next “frame” and has made whatever necessary adjustments are needed to correct its course before dispatchers even respond.
Of course, I have no way to validate any of this, as any numbers I can provide are just made up based on my own experience doing this job around 7 years ago, and I don’t believe Waymo publishes any actual statistics in this regard, so you’ll just have to take the word of a random guy on the internet for however much its worth. :)
For what it’s worth, I will say that self-driving cars are about the only technology that genuinely excite me these days and some of the things they’re capable of are not talked about nearly enough (you hardly ever hear about the fact that Waymo has autonomous semi-trucks on public roads). With the exception of Tesla, who have been seen as mostly a joke by the rest of the industry, pretty much everybody in this particular tech space are doing some very impressive things.
I don’t suck big tech dick that often, but the self-driving car industry has some good techdick right now.
That’s because they never got anywhere with them and shut down this branch in 2023
https://en.wikipedia.org/wiki/Waymo#Trucking
Look, I do understand your points but I will not give any company a pass on this. Are these cars “fully autonomous”? no they are not… if the dude is the Philippines is not around, sooner than later the cars will stop or hurt someone; and this is if we still let them stay in their small, well mapped, cozy area… take them someplace else and the car will be mostly useless
Personally, I am not convinced this tech is any good for anything other than making unwalkable cities even more expensive to navigate. I do of course see the safety aspect (although the stats are very shady at the moment) but again, this is only true and valid if these cars can ever be autonomous and not just ride a virtual track.
Why do you need anything beyond “riding a virtual track”, though? The point of a car is to go from A to B, and a track is literally the most effective way to do that (which is why we need more trains in the first place, but that’s an entirely separate rant I could go on about).
IMO, at a certain point you reach a level of autonomy that is sufficient enough. It doesn’t need to know every nook and cranny of obscure roads because realistically, nobody’s going there to begin with. And even then in those situations where you did need to go to a location that wasn’t pre-mapped, it’s still likely to get you 99.9% of the way to your destination, anyway. Waymos are also constantly scanning and remapping their surroundings in real time (to adjust for things like road work, detours, and other blockages), so it stands to reason that if you had to manually engage the car after reaching the end of its mapped area, it would eventually develop a map of the new areas with a high-enough level of confidence to navigate on its own, over time.
I also think you have the wrong takeaway regarding the remote dispatchers being required. I think it’s a good thing that they exist, because that means that Waymo doesn’t trust their own machines to be 100% accurate. Tesla incorrectly does, which is why their self-driving cars have an exponentially higher rate of at-fault collisions than any of their competitors. Waymo at least understands the realistic limitations of their tech, and has a safety net in place for when those limitations are reached. I think that’s the correct way to do it, as you lead up toward an actual finished product.
I don’t… but if we have not truly automated literal tracks (trains), I have very little trust in the virtual version being “ready”
Agreed, and to me, that means ZERO human intervention (at least to be able to call it “fully autonomous”). The only appropriate intervention is when the car detects it is breaking down and must stop and wait for a tow truck
And I think here is the gist… I do not want or care for robotaxis in my city. That is just the same bullshit with more steps to continue down the line of unwalkable cities where you depend on a car (which is now a rent seeking service). I want truly fully autonomous driving so eventually we can get to the point where a fleet of truly autonomous cars covers the last mile where mass public transit cannot. Therefore, I do need tech that can navigate low traffic neighborhoods, not just go around downtown
If they did, they wouldn’t call their tech “Fully autonomous”… I mean, come on, why do you still fight for the lies they try to feed you. It’s the same everywhere, with food “added vitamins” without telling you their processing took out so many nutrients you are basically eating air even with the “added vitamins”
If Waymo had called their tech “minimally assisted driving” I would have the respect for them you seek from me
The point is you can have say 100 people in an office someone manage 5,000 cars instead of having 5,000 drivers operating those 5,000 cars and it’s a way more efficient.
Also well built self driving cars like the Waymo ones are actually significantly safer than human drivers.
For a long time now, humans have been utilising machines and robots to perform tasks that humans would otherwise have had to perform - particularly when it comes to manufacturing. Occasionally, one of those machines will encounter an error or issue, and a human will need to intervene to either fix the machine or guide the process before it can resume its task.
These advancements have allowed a very small number of humans to oversee what would have otherwise required hundreds, potentially thousands of humans to do independently. Even when humans were performing these tasks, they would occasionally need someone with more specialisation or experience to help them with completing such a task.
The point of this tech - the point of all tech that I’ve described above - is to reduce the amount of humans needed to produce a given result. It’s a reduction in the need for labour. It’s a matter of efficiency, not a complete replacement of any need for any human to ever intervene under any circumstances. Under communism, or even well-fleshed out socialism, it would free up humans from having to perform menial labour to instead pursue their passions, work on vital human skills, create and consume art… just live better lives.
Now, would I ride in a self-driving vehicle? Certainly not yet. The tech still needs time to develop before I’m confident in it. As time goes on, the need for human intervention will continue to drop, and hopefully my confidence will grow.
You realize the tech is still under development, right?
Being able to autonomously drive itself in the vast majority of scenarios is a big fucking deal.
yes, and I am calling a spade a spade
In a controlled limited area, with human intervention… hmmm again, not sure I would use the words “Fully autonomous driving” for this but English is not my first language so maybe it is like the “unlimited” plans from cell phone companies
That’s like saying someone using the steering wheel isn’t driving because the tires actually turn rhe car.
This is different. It’s like playing a racing video game via trading cards. They’re giving the car directions just like a person might give a driver directions.