[Efthreeoh]

Originally Posted by antzcrashing

I get you. I can see a natural implementation of smart driving as one in which highway driving is autonomous and entering and exiting highways is human controlled. In that way it would be very similar to airplane methods. Even the highway self driving would be extremely risky with other normal drivers (unless they control he situation with dedicated lanes etc). But lets take that for a second, if one dedicated lane, how do you handle a flat tire or a breakdown? Add a breakdown lane to it? Its hard enough to build the infrastructure and get approvals for extra lanes. Maybe thats why Elon is digging tunnels...

Think of what the cost would be to make those kinds of changes to the existing infrastructure. It would take 50 years and hundreds of trillions of dollars. No one has that kind of money or that amount of time for investment payback. And what you speak of is a transitional situation just to integrate machine driving with human driving. It's not financially possible, which is why the proposed engineering solution is car-to-car communication for collision avoidance. That takes multiple layers of redundancy and common platform mechanical and software systems, which leads to uncompetitive development of the vehicles.

There's no capitalistic model for it.

[MisterF80M3]

Originally Posted by Alfisti

Been very, very bearish about this since day 1. I have always maintained that for it to work, drivers need to be banned so the cars can all communicate. This idea that Level 4 is acceptable or Level 5 with a mix of Ai and real drivers on the road will work is hogwash. It's astoundingly difficult and some of the world's largest companies have walked away from it in recent years basically saying "this is just too fucking hard".

5 years ago I said 25 years before this works, and i stand by it.

20 more years to be fully autonomous? Nah.

Tesla is already 80% there. They released a full self-driving beta code this week. Waymo has already launched driverless rides in Phoenix and is continuing to expand to other states.

The technology is there. There just needs to be more testing.

[Alfisti]

Originally Posted by MisterF80M3

20 more years to be fully autonomous? Nah.

Tesla is already 80% there. They released a full self-driving beta code this week. Waymo has already launched driverless rides in Phoenix and is continuing to expand to other states.

The technology is there. There just needs to be more testing.

It really isn't there. It does not work, this IMHO is the toughest thing human beings have EVER tried to do outside of a cure for the common cold.

To be clear, I think we could launch tomorrow where vehicles talk to each other and drive themselves, no issue but we are talking mixing with regular drivers. Absolutely not ready in any way, shape or form.

[antzcrashing]

Originally Posted by Efthreeoh

Think of what the cost would be to make those kinds of changes to the existing infrastructure. It would take 50 years and hundreds of trillions of dollars. No one has that kind of money or that amount of time for investment payback. And what you speak of is a transitional situation just to integrate machine driving with human driving. It's not financially possible, which is why the proposed engineering solution is car-to-car communication for collision avoidance. That takes multiple layers of redundancy and common platform mechanical and software systems, which leads to uncompetitive development of the vehicles.

There's no capitalistic model for it.

I agree that car-to-car messaging is key, but it will take time to develop that communication protocol. Think about this, you may need to discuss who has a right to a lane change while car B waits - that's not an easy answer to sort out

[antzcrashing]

Originally Posted by tranquility

Won't convince me, the infrastructure's still there, all they have to do is tamper w your car and hook it up for remote control.

Valid point. The insider engineers of these companies are themselves a risk for hacking bc they have such a powerful knowledge of the cars

[W///]

Originally Posted by Efthreeoh

Yes, I was reinforcing your point. People connect airplanes and automobiles together via the "autopilot" link. Airplanes are not flown autonomously under autopilot. People equate the fantastic safety record of air travel to autonomous driving because there is a commonly misunderstood purpose of aviation autopilot systems. The main reason air travel is safe is because of mandated separation standards. Humans keep airplanes separated by design with large margins of safety. Automobiles have virtually zero separation standards. They will get introduced with autonomy, and people will not like it.

I suppose, technically, there are separation standards, whether they're followed or not is a different story, especially in the US.

I don't remember what the fine is for tailing someone in Germany, but it's not insignificant.

Not saying this is anyway comparable to air travel standards, you do make a great point.

[MisterF80M3]

Originally Posted by Alfisti

It really isn't there. It does not work, this IMHO is the toughest thing human beings have EVER tried to do outside of a cure for the common cold.

To be clear, I think we could launch tomorrow where vehicles talk to each other and drive themselves, no issue but we are talking mixing with regular drivers. Absolutely not ready in any way, shape or form.

What do you mean it does not work?

Waymo has already proven that it can drive itself and pick up passengers.

[dreamingat30fps]

Originally Posted by MisterF80M3

What do you mean it does not work?

Waymo has already proven that it can drive itself and pick up passengers.

Was it just the video or did those damn things sound like a fucking jet? Loudest minivan I've ever heard!

I would imagine all the extra sensors and gadgets on the waymo are just to make up for the deficiencies in software. Because technically all humans use to drive is one stereoscopic camera on a swivel. So with advanced enough software there should be no reason it couldn't be done with just vision.

[5.M0NSTER]

Originally Posted by dreamingat30fps

Was it just the video or did those damn things sound like a fucking jet? Loudest minivan I've ever heard!

I would imagine all the extra sensors and gadgets on the waymo are just to make up for the deficiencies in software. Because technically all humans use to drive is one stereoscopic camera on a swivel. So with advanced enough software there should be no reason it couldn't be done with just vision.

This is the problem though. To do it just with vision you’d have to re-create the part of the human mind responsible for spatial awareness, reasoning and our ability to extrapolate future location of an object relative to us. Base on a stereo cameras which are our 2 eyes. If you can do that, than cameras are all we need. But we are nowhere there right now. And the Waymo cars already drive with a liquid cooled super computer in the back.

[Alfisti]

IT DOES NOT WORK. If it did it would be released already.

[mkoesel]

Originally Posted by Alfisti

IT DOES NOT WORK. If it did it would be released already.

It's true that fully generalized autonomous vehicle travel is not possible yet. Today's technology allows for vehicles that drive by themselves in a very limited set of circumstances or on a limited number of routes where conditions are tightly controlled within a specific set of parameters.

So it's in its infancy. How much time is needed to get from where we are today to the holy grail of a totally automated fleet? Likely decades. However, that eventuality is not necessary for a preponderance of people to begin to see real life changing benefits. Exactly when we reach the tipping point is difficult to pin down right now. As with any transformative innovation, this will be "doomed for failure" for the naysayers, "years and years away" for the realists, and "on the cusp of breakthrough" for the visionaries. The reality of the future falls somewhere on that spectrum.

[Alfisti]

I've been saying "years and years away" since day one. As I said, called it at 25 years 5 years ago and seen nothing since to change my mind, if anything that may be brave.

[dreamingat30fps]

Originally Posted by 5.M0NSTER

This is the problem though. To do it just with vision you’d have to re-create the part of the human mind responsible for spatial awareness, reasoning and our ability to extrapolate future location of an object relative to us. Base on a stereo cameras which are our 2 eyes. If you can do that, than cameras are all we need. But we are nowhere there right now. And the Waymo cars already drive with a liquid cooled super computer in the back.

I agree. However I'm not sure if self driving cars will really take off unless it can be done with just vision. How much does one of those vans with a super computer cost? Now if you could get a device like the comma.ai to do FSD that would be a game changer. Now you can retrofit existing cars to have FSD. I think it WILL happen the question is when. Not when Elon says because that guy says it's going to happen every year.

Originally Posted by Alfisti

IT DOES NOT WORK. If it did it would be released already.

Why you so mad bro? Of course it doesn't work, that's why apart from waymo in a small area you don't see any autonomous cars. Does that mean they should just stop working on it? We haven't cured cancer so let's stop because IT DOES NOT WORK. Maybe all the people researching this stuff should quit and get into the spam business because we sure need more spammers.

[Alfisti]

I never said stop working on it, I'm saying it is going to take a lot longer than most people realise.

[Efthreeoh]

Originally Posted by mkoesel

It's true that fully generalized autonomous vehicle travel is not possible yet. Today's technology allows for vehicles that drive by themselves in a very limited set of circumstances or on a limited number of routes where conditions are tightly controlled within a specific set of parameters.

So it's in its infancy. How much time is needed to get from where we are today to the holy grail of a totally automated fleet? Likely decades. However, that eventuality is not necessary for a preponderance of people to begin to see real life changing benefits. Exactly when we reach the tipping point is difficult to pin down right now. As with any transformative innovation, this will be "doomed for failure" for the naysayers, "years and years away" for the realists, and "on the cusp of breakthrough" for the visionaries. The reality of the future falls somewhere on that spectrum.

When sanctioning bodies from Government, Law, and Implementation can walk through a cohesive plan on the integrated collaboration of these three elements critical to making a step change in the private transportation system then I'll believe it will happen. Thinking just the technology is going to drive a change to an autonomous driving environment is just shortsighted.

[mkoesel]

Originally Posted by Efthreeoh

Thinking just the technology is going to drive a change to an autonomous driving environment is just shortsighted.

Sure, while the technologists push the limits of what's possible, the bureaucrats will do their thing too. That's how evolution happens. It's not just one entity operating in isolation.

Today's product development tends to occur in an agile fashion, not the old waterfall approach of yesteryear. Not everything is planned out up front. I'm not not necessarily the guy who likes to see all of this bleeding edge stuff proven out while using public spaces as a laboratory, so I certainly do appreciate the legislative forces. For example, I'm glad that at least the Germans have called shenanigans on the "Autopilot" nonsense. Noise and drama like that aside though, I don't have a problem with the see-what-sticks approach. Self parking and adaptive cruise control were witchcraft pre-2000. Folks would have run screaming. Now your economy cars can do it. Autonomy isn't going to land in our laps like the astronauts landed on the moon. It's already happening and it's going to keep happening more and more until one day our great grandkids are debating whether flying cars can realistically replace land vehicles because who can even remember when those didn't drive themselves anymore.

[Efthreeoh]

Originally Posted by mkoesel

Sure, while the technologists push the limits of what's possible, the bureaucrats will do their thing too. That's how evolution happens. It's not just one entity operating in isolation.

Today's product development tends to happens in an agile fashion, not the old waterfall approach of yesteryear. Not everything is planned out up front. I'm not not necessarily the guy who likes to see all of this bleeding edge stuff proven out while using public spaces as a laboratory, so I certainly do appreciate the legislative forces. For example, I'm glad that at least the Germans have called shenanigans on the "Autopilot" nonsense. Noise and drama like that aside though, I don't have a problem with the see-what-sticks approach. Self parking and adaptive cruise control were witchcraft pre-2000. Folks would have run screaming. Now your economy cars can do it. Autonomy isn't going to land in our laps like the astronauts landed on the moon. It's already happening and it's going to keep happening more and more until one day our great grandkids are debating whether flying cars can realistically replace land vehicles because who can even remember when those didn't drive themselves anymore.

The FAA (read DOT) can't even yet figure out how to integrate flying UAV into the national airspace. Drone technology is highly advanced, watch the DRL (drone racing league) as an example and you'll be amazed. I'll not even mention Predator. So if there is any place in the transportation sector that the technology "is here" it is drones. Yet the FAA is fumbling about and can't get out of the way. Letting millions of cars drive themselves with humans inside? Good luck with overcoming that obstacle at the Department of Trans.

[M5theonlyone]

If it goes wrong and there's a smash then who's at fault. There have been two Tesla self drive deaths already?

[dreamingat30fps]

Originally Posted by M5theonlyone

If it goes wrong and there's a smash then who's at fault. There have been two Tesla self drive deaths already?

To be clear.... Teslas do not drive themselves. There have been deaths of people using fancy adaptive cruise control and lane keep assist who were eating shit.

But I do see what you mean when we do get to a point of FSD. I think for the near future there will always be a driver required and they would be responsible. Obviously if we get robo taxis like waymo where the vehicles are owned and operated by the company then they would be responsible if their vehicle is at fault. Personally I think we will see more waymo style FSD where you basically rent the vehicle way before we see FSD for personal vehicles.

[M5theonlyone]

Originally Posted by dreamingat30fps

To be clear.... Teslas do not drive themselves. There have been deaths of people using fancy adaptive cruise control and lane keep assist who were eating shit.

But I do see what you mean when we do get to a point of FSD. I think for the near future there will always be a driver required and they would be responsible. Obviously if we get robo taxis like waymo where the vehicles are owned and operated by the company then they would be responsible if their vehicle is at fault. Personally I think we will see more waymo style FSD where you basically rent the vehicle way before we see FSD for personal vehicles.

Aha right and I agree that FSD should be tried out on taxis first before being implemented on owner driver cars, if at all.

[Dog Face Pony Soldier]

I've said before that the logical progression would be as follows...

First automated trains. This is the easiest to accomplish.

Next automated sea vessels. Big, wide open spaces.... plus we already use Harbor Pilots to handle the last mile(s).

After we gain confidence with the reliability and safety of trains and ships, we start automating inner-city busses. At least these vehicles run a very defined route, but dealing with weather conditions alone will present a major challenge.

THEN we start attempting to automate 'free range' cars, trucks, and busses. These present all the challenges as above, plus they don't stick to pre-defined routes. They also have to navigate private roads and surfaces like parking lots and driveways.

Seems stupid to attempt the most challenging automation first.

[mkoesel]

Originally Posted by Dog Face Pony Soldier

Seems stupid to attempt the most challenging automation first.

I'm no entrepreneur, but I'd probably start wherever the money is if it were me.

The nice thing about a free market - you can plunk down your $$$ and bet against the corporate players who you think are doing it wrong. Short 'em and get rich.