[zenmaster]

Originally Posted by Efthreeoh

If you want an electric vehicle with autopilot, take the train....

Oh wait, they crash too.

BMW will - must - have a fleet of them - electric and semi autonomous - in 5 years. Which is shortsightedly behind the curve by a few years, but sometimes the latecomers wind up owning new markets. Electric range should be near 400mi/650km by then.

[Efthreeoh]

Originally Posted by zenmaster

BMW will - must - have a fleet of them - electric and semi autonomous - in 5 years. Which is shortsightedly behind the curve by a few years, but sometimes the latecomers wind up owning new markets. Electric range should be near 400mi/650km by then.

So here is alternate perspective regarding autonomous driving I’d like to put forth. We all agree that Commercial airflight is the most highly automated and safe form of transportation currently on the planet. Air traffic is highly automated, from automated flight scheduling, planes using autopilot during enroute flight, to the automated routing of air traffic over designated air routes, to the automated tracking of aircraft in terminal areas, to the human-controlled orchestrated landing and takeoff of planes at the surface level. For just the US, all of this to safely control around 10,000 daily flights. It is a highly complex system built over the last 60 years since RADAR was implemented to track aircraft in flight. It takes hundreds of thousands of people every day to make this automation system work and maintain it. And that is just for 10,000 flights all kept at 5-mile separation and thousands of feet in elevation separation, with numerous redundant systems to prevent intersections of planes, systems such as ACARS, FANS, just to name a few. Again all just for 10,000 flights a day and all extraneous to the actual aircraft itself, except for the control hardware and software for the auto pilot system.

Tesla has a nice little “autopilot” system in the vehicle, which prevents it from hitting stuff in its intended path. It works well in some situations, but doesn’t when the traffic situation gets complicated, such as we’ve seen when a tractor trailer pulls across a roadway and Tesla’s autopilot determines the truck is a billboard and someone gets decapitated. This is “semiautonomous” control of the vehicle. The more semiautonomous vehicles on the road, controlled only by what each vehicle’s sensor suite “sees” and processes in the software, there will be more accidents, not less. Google’s cars have hit stuff and have a probable top speed of 35 MPH. Five year’s time is not going to make Tesla’s nor Google’s semiautonomous any safer. Until such time the control of ground automobile traffic is controlled by a third-party (i.e. not drivers in their cars) control system will autonomous driving be safer than human-driving. And then I even doubt that. The traffic automation control systems now used for the safest form of travel (lowest deaths per millions of miles traveled) - commercial airline travel, can’t even begin to cope with automotive traffic and it’s millions of daily “flights”.

One Tesla owner here loves his autopilot function and believes Musk's proposition that it is actually safer than human-controlled driving. It is a complete misrepresentation of Tesla's capability of safety. Take a Tesla X or S and put it in autopilot mode in New York City and see what happens.

Automated automobile traffic is many decades away. A five-year advancement of what’s currently on the table is getting nowhere. A cell phone app is not going to solve this problem.

[zenmaster]

Originally Posted by Efthreeoh

So here is alternate perspective regarding autonomous driving I’d like to put forth. We all agree that Commercial airflight is the most highly automated and safe form of transportation currently on the planet. Air traffic is highly automated, from automated flight scheduling, planes using autopilot during enroute flight, to the automated routing of air traffic over designated air routes, to the automated tracking of aircraft in terminal areas, to the human-controlled orchestrated landing and takeoff of planes at the surface level. For just the US, all of this to safely control around 10,000 daily flights. It is a highly complex system built over the last 60 years since RADAR was implemented to track aircraft in flight. It takes hundreds of thousands of people every day to make this automation system work and maintain it. And that is just for 10,000 flights all kept at 5-mile separation and thousands of feet in elevation separation, with numerous redundant systems to prevent intersections of planes, systems such as ACARS, FANS, just to name a few. Again all just for 10,000 flights a day and all extraneous to the actual aircraft itself, except for the control hardware and software for the auto pilot system.

Apples and oranges.

Originally Posted by Efthreeoh

Tesla has a nice little “autopilot” system in the vehicle, which prevents it from hitting stuff in its intended path. It works well in some situations, but doesn’t when the traffic situation gets complicated, such as we’ve seen when a tractor trailer pulls across a roadway and Tesla’s autopilot determines the truck is a billboard and someone gets decapitated.

It may not work how you want it to work, but it works as advertised in all situations.

Originally Posted by Efthreeoh

This is “semiautonomous” control of the vehicle. The more semiautonomous vehicles on the road, controlled only by what each vehicle’s sensor suite “sees” and processes in the software, there will be more accidents, not less.

You are clearly confused. By definition, semi autonomous does not mean controlled only by the sensor suite and software. There is the crucial human driver requirement.

Originally Posted by Efthreeoh

Google’s cars have hit stuff and have a probable top speed of 35 MPH. Five year’s time is not going to make Tesla’s nor Google’s semiautonomous any safer. Until such time the control of ground automobile traffic is controlled by a third-party (i.e. not drivers in their cars) control system will autonomous driving be safer than human-driving.

Nah, not good enough. If 3rd party is a requirement, then some other manufacturer will make a better one which does not have such a limitation.

Originally Posted by Efthreeoh

And then I even doubt that. The traffic automation control systems now used for the safest form of travel (lowest deaths per millions of miles traveled) - commercial airline travel, can’t even begin to cope with automotive traffic and it’s millions of daily “flights”.

I'd drop the air traffic control analogy. It just doesn't fit.

Originally Posted by Efthreeoh

One Tesla owner here loves his autopilot function and believes Musk's proposition that it is actually safer than human-controlled driving. It is a complete misrepresentation of Tesla's capability of safety. Take a Tesla X or S and put it in autopilot mode in New York City and see what happens.

That is a ignorant idea, simply because the system was not designed to handle nor intended to operate in city traffic.

Originally Posted by Efthreeoh

Automated automobile traffic is many decades away.

Fully autonomous driving is probably decades away where it would be able to replace driver in all possible situations, but the tech's utility incrementally improves each year.

Originally Posted by Efthreeoh

A five-year advancement of what’s currently on the table is getting nowhere.

Wrong and willfully ignorant.

Originally Posted by Efthreeoh

A cell phone app is not going to solve this problem.

Straw man.

[Efthreeoh]

Originally Posted by zenmaster

Apples and oranges.

It may not work how you want it to work, but it works as advertised in all situations.

You are clearly confused. By definition, semi autonomous does not mean controlled only by the sensor suite and software. There is the crucial human driver requirement.


Nah, not good enough. If 3rd party is a requirement, then some other manufacturer will make a better one which does not have such a limitation.

I'd drop the air traffic control analogy. It just doesn't fit.

That is a ignorant idea, simply because the system was not designed to handle nor intended to operate in city traffic.

Fully autonomous driving is probably decades away where it would be able to replace driver in all possible situations, but the tech's utility incrementally improves each year.

Wrong and willfully ignorant.

Straw man.

The cell phone app was tongue in cheek, so relax. But the point is that people think technology advances so fast because a cool new app comes out everyday to solve a simple problem; autonomous vehicle traffic requires some serious and expensive technology and is far from simple. Using the air traffic control system of the US to develop standards for ground vehicular traffic is what the US DOT (who will have overview of autonomous vehicular traffic) is going to start with it because that is the best model they have, so it is apples to apples. I work in the DOT industry, I know how the people at the DOT think. The US Government is not going to allow the automotive industry develop autonomous driving on its own because people in the Government are taught to not trust commercial industry.

Getting cars to drive themselves based off of sensor data is just the first step, and the cars are not very good at it. There is no redundancy built into the system for errors and the traffic situation is far too dynamic to allow for errors and even if redundancy is available it is probably not quick enough to counteract. The event of the Tesla running into the tractor trailer is the perfect example. Per Mr. Musk, the Tesla sensor suite is programmed to not see billboards as traffic, so the car didn't slow down and stop. The driver of the truck either (a) didn't see the Tesla, or (b) saw the Tesla and thought "I'm a big ass truck and I'm turning left in plenty of time for the Tesla to recognize I'm cutting him off so he'll slow down and avoid hitting me." What the truck driver didn't realize was the Tesla was in autopilot mode and was programmed to not recognize billboards (or semi-tractor trailers perpendicular to traffic). The Tesla pilot was trusting Musk's autopilot to steer clear of traffic hazards. The result... dead Tesla Pilot. The redundancy in that situation would have been the placement of the truck and the Tesla would have been controlled by a Vehicular Traffic Control System, which would prevented the truck from crossing paths with the Tesla; when the DOT report comes out with its findings, I guarantee you that will be one of the findings.

Controlling traffic by means of car-to-car communication is not going to work because the data processing to do so is not economically available, and may never be; especially with the need to have double or triple redundancy. When the day comes humans write software code without bugs, will be the day the sun burns the planet to a crisp. Software bugs and car-to-car controlled traffic traveling at 60 MPH does not mix. Controlling cars with GPS-based timing is completely insane because GPS is highly vulnerable to local spoofing and jamming. Some butthead with a $80 GPS jammer can wreak havoc.

Finally, Musk is the one who is ignorant. It was his statement that his Autopilot app has operated in Beta-mode to the point thus far that the accidents per million miles for his cars in autopilot mode is less than human-driven vehicles. What is ignorant and false about his statement is the level of traffic complexity his cars have operated in Autopilot mode vs. the overall accounting of human-driven traffic, which accounts for all miles driven regardless of traffic complexity. So that means that humans drive in far more complex traffic situations and have about the same level of accidents per million miles than Tesla's Autopilot used on a narrow section of a low complexity traffic environment. So his comments are apples to oranges.

[Fundguy1]

Cafe standards. Part of EPA. One ore reason to completely dismantle EPA. Caused death of the station wagon. Was crated because of fuel shortages. Entire purpose was to lower gasoline consumption because we were dependent of foreign oil. We aren't any more. Now its used as a weapon of the left to fight large corporations and force the bogus carbon emissions crap. It's resulted in many people buying much smaller cars. Less safe. Deaths. Less resale value. Lost wealth. Also resulted in people buying large SUVs instead of station wagons. More fuel consumption and easier to kill prius drivers.

Fuel economy should be driven by fuel prices, not a stupid law that kills and makes you loose your shirt financially. People should have the choice to buy what they want. Not be forced by the 9th circuit court of liberal bs.

[Efthreeoh]

Originally Posted by Fundguy1

Cafe standards. Part of EPA. One ore reason to completely dismantle EPA. Caused death of the station wagon. Was crated because of fuel shortages. Entire purpose was to lower gasoline consumption because we were dependent of foreign oil. We aren't any more. Now its used as a weapon of the left to fight large corporations and force the bogus carbon emissions crap. It's resulted in many people buying much smaller cars. Less safe. Deaths. Less resale value. Lost wealth. Also resulted in people buying large SUVs instead of station wagons. More fuel consumption and easier to kill prius drivers.

Fuel economy should be driven by fuel prices, not a stupid law that kills and makes you loose your shirt financially. People should have the choice to buy what they want. Not be forced by the 9th circuit court of liberal bs.

Just had this conversation yesterday with a colleague at work. He's looking to replace a 14-year old Hyundai Elantra. He supports our logistics side and has far less than a 6-figure salary. He was complaining that the average working man can't afford a decent-sized new car because the prices are too high. I told him he's right and it is because of the zealous chase by the left to curb carbon exhaust and maximize fuel conservation. The cars today are engineered to pass high-MPG EPA tests in the lab (re: VW) without regard to longevity and total lifecycle costs. A balance needs to be struck between capability, economy, and lifecycle cost. The focus seems to be on economy only. We can only wonder what a non-electric vehicle will be that meets the EPA 54.5 MPG mandate in 2025.

[zenmaster]

Originally Posted by Efthreeoh

The cell phone app was tongue in cheek, so relax. But the point is that people think technology advances so fast because a cool new app comes out everyday to solve a simple problem; autonomous vehicle traffic requires some serious and expensive technology and is far from simple.

Yes, it is a technical challenge involving several domains of research but one which is yielding surprising progress nonetheless.

Originally Posted by Efthreeoh

Using the air traffic control system of the US to develop standards for ground vehicular traffic is what the US DOT (who will have overview of autonomous vehicular traffic) is going to start with it because that is the best model they have, so it is apples to apples. I work in the DOT industry, I know how the people at the DOT think. The US Government is not going to allow the automotive industry develop autonomous driving on its own because people in the Government are taught to not trust commercial industry.

That may be the case for their current mindset and positioning. But research on these systems is along the lines of a completely new paradigm, without actual analogy to to air traffic control.

Originally Posted by Efthreeoh

Getting cars to drive themselves based off of sensor data is just the first step, and the cars are not very good at it. There is no redundancy built into the system for errors and the traffic situation is far too dynamic to allow for errors and even if redundancy is available it is probably not quick enough to counteract.

That is correct, but the current systems are only supposed to support a low level of semi-autonomous driving capability. If the system works as specified, then there is no problem, there is no flaw. It is that simple. You are totally confusing what you want it to potentially do with what it is designed to do and does quite well.

Originally Posted by Efthreeoh

The event of the Tesla running into the tractor trailer is the perfect example. Per Mr. Musk, the Tesla sensor suite is programmed to not see billboards as traffic, so the car didn't slow down and stop. The driver of the truck either (a) didn't see the Tesla, or (b) saw the Tesla and thought "I'm a big ass truck and I'm turning left in plenty of time for the Tesla to recognize I'm cutting him off so he'll slow down and avoid hitting me." What the truck driver didn't realize was the Tesla was in autopilot mode and was programmed to not recognize billboards (or semi-tractor trailers perpendicular to traffic). The Tesla pilot was trusting Musk's autopilot to steer clear of traffic hazards. The result... dead Tesla Pilot.

That is a shamefully false characterization of cause and effect. Unless shown otherwise, the actual cause was driver not paying attention as is the stated requirement for the driver using this level of assisted driving tech.

Originally Posted by Efthreeoh

The redundancy in that situation would have been the placement of the truck and the Tesla would have been controlled by a Vehicular Traffic Control System, which would prevented the truck from crossing paths with the Tesla; when the DOT report comes out with its findings, I guarantee you that will be one of the findings.

This is silly. The investigation *should* be focused on whether or not the car somehow took control away from the driver - that is either it steered into or accelerated toward the truck. Those two things are basically the only possible liability. The only liability for not stopping would be if the driver applied brakes and the car did not comply.

Originally Posted by Efthreeoh

Controlling traffic by means of car-to-car communication is not going to work because the data processing to do so is not economically available, and may never be; especially with the need to have double or triple redundancy. When the day comes humans write software code without bugs, will be the day the sun burns the planet to a crisp. Software bugs and car-to-car controlled traffic traveling at 60 MPH does not mix. Controlling cars with GPS-based timing is completely insane because GPS is highly vulnerable to local spoofing and jamming. Some butthead with a $80 GPS jammer can wreak havoc.

All these things may be true, but these are more straw men from the standpoint of actual research in the area. Understanding is going to come from the full-time research by scientists and engineers, not FUD and uninformed, casual speculation.

Originally Posted by Efthreeoh

Finally, Musk is the one who is ignorant. It was his statement that his Autopilot app has operated in Beta-mode to the point thus far that the accidents per million miles for his cars in autopilot mode is less than human-driven vehicles. What is ignorant and false about his statement is the level of traffic complexity his cars have operated in Autopilot mode vs. the overall accounting of human-driven traffic, which accounts for all miles driven regardless of traffic complexity. So that means that humans drive in far more complex traffic situations and have about the same level of accidents per million miles than Tesla's Autopilot used on a narrow section of a low complexity traffic environment. So his comments are apples to oranges.

I don't think we know enough at this point to assert higher accident rates for semi autonomous vs human-driven only, regardless of factoring out total miles driven in more complex environments.

The current level of tech has always required a non-distracted driver, just like non-assisted driving does. There is no difference in driver responsibilities. This is where you seem to be confused.

The only potential problem with assisted driving tech is if it somehow could take control away from the driver in such a manner as to contribute to an accident or otherwise impede the flow of traffic. But there has been no evidence of such a problem.

[Efthreeoh]

Originally Posted by zenmaster

Yes, it is a technical challenge involving several domains of research but one which is yielding surprising progress nonetheless.

That may be the case for their current mindset and positioning. But research on these systems is along the lines of a completely new paradigm, without actual analogy to to air traffic control.

That is correct, but the current systems are only supposed to support a low level of semi-autonomous driving capability. If the system works as specified, then there is no problem, there is no flaw. It is that simple. You are totally confusing what you want it to potentially do with what it is designed to do and does quite well.

That is a shamefully false characterization of cause and effect. Unless shown otherwise, the actual cause was driver not paying attention as is the stated requirement for the driver using this level of assisted driving tech.

This is silly. The investigation *should* be focused on whether or not the car somehow took control away from the driver - that is either it steered into or accelerated toward the truck. Those two things are basically the only possible liability. The only liability for not stopping would be if the driver applied brakes and the car did not comply.

All these things may be true, but these are more straw men from the standpoint of actual research in the area. Understanding is going to come from the full-time research by scientists and engineers, not FUD and uninformed, casual speculation.

I don't think we know enough at this point to assert higher accident rates for semi autonomous vs human-driven only, regardless of factoring out total miles driven in more complex environments.

The current level of tech has always required a non-distracted driver, just like non-assisted driving does. There is no difference in driver responsibilities. This is where you seem to be confused.

The only potential problem with assisted driving tech is if it somehow could take control away from the driver in such a manner as to contribute to an accident or otherwise impede the flow of traffic. But there has been no evidence of such a problem.

Trust me, I'm not confused about anything.

Nope, it's an exact cause and effect and the point of the argument. The Tesla Pilot assumed the "tech" was going to trace the Model S across the roadscape and avoid obstacles (i.e. drive itself regardless of what the language is in the Tesla owners manual). The Tesla Autopilot lead to the driver's distraction rather than improving (assisting) his poor driving skills. If the Tesla Autopilot system requires full driver attention, then what is the point of it. If the car is going to semi-autonomously drive itself, then by human nature the human is going to become disinterested in the activity of driving. In all actuality all the Tesla can do is in limited fashion keep the car on the road in a well defined lane (environment); one could train a chimpanzee to do the very same thing. However, when the cognitive part of driving comes into play, such as when a tractor trailer driver decides to place his vehicle perpendicular to traffic and block it, the car and the chimp fail to execute what humans do extremely well (when motivated).

I'll have to watch the videos in the morning as I'm bandwidth limited during the day, but I did look at the first 15 minutes of the first video and didn't hear any new paradigm information; I heard sensors and 3D mapping, all current stuff. So anyone interested in the topic and who is pretty well read in it, and has a background in transportation and traffic control may get a bit bored in the beginning of the talk. But I'll give it a view tomorrow.

What I do know is it might be better to spend all the hundreds of billions of research money that will take to possibly implement an autonomous system, on simply teaching humans how to drive better. Maybe an educational course a few steps beyond a high school football coach teaching driver's ed during summer break. And perhaps the audacity to have real requirements for students/drivers to pass and keep passing to maintain their license and knock off the idea the driving is a "privilege". Perhaps developing systems that provide drivers of more information to make better choices. Perhaps use the money to design and build better roads that don't lead to accidents.

But back to the tech... the whole idea of taking the human out of the act of driving is to condense traffic so it can flow better on the limited amount of lane space available (exactly what the FAA is currently doing with its NextGen Air traffic Control System). The concept behind NextGen is to get more precise aircraft position data and give pilots more situational awareness. The DOT is not going to think any different about ground vehicular traffic.

Google's idea is to allow people to text and surf the internet while driving (err... travelling).

[zenmaster]

Originally Posted by Efthreeoh

Trust me, I'm not confused about anything.

Nope, it's an exact cause and effect and the point of the argument. The Tesla Pilot assumed the "tech" was going to trace the Model S across the roadscape and avoid obstacles (i.e. drive itself regardless of what the language is in the Tesla owners manual). The Tesla Autopilot lead to the driver's distraction rather than improving (assisting) his poor driving skills. If the Tesla Autopilot system requires full driver attention, then what is the point of it.

Such a pilot is making a very bad assumption. The point of the tech is to assist the driver in keeping the lane, and a safe distance from other vehicles. It will catch things like blind spot occupancy and cars changing lanes into your car. This is a very useful feature and one I'd like my car to have. If semi-autonomous is not useful to you personally, then you can wait a while for fully autonomous tech.

Originally Posted by Efthreeoh

If the car is going to semi-autonomously drive itself, then by human nature the human is going to become disinterested in the activity of driving. In all actuality all the Tesla can do is in limited fashion keep the car on the road in a well defined lane (environment); one could train a chimpanzee to do the very same thing. However, when the cognitive part of driving comes into play, such as when a tractor trailer driver decides to place his vehicle perpendicular to traffic and block it, the car and the chimp fail to execute what humans do extremely well (when motivated).

It's "human nature" to smoke cigarettes, play with fire, and eat like a pig too. There's no culpability here on the part of Tesla due to a driver not following extremely simple requirements. If you insist on shifting blame for clearly irresponsible actions then you are part of a thought pattern that serves to reduce freedom (and dignity) for all. Most sane people would probably agree that the system is 100% safe if used correctly. Many Tesla drivers really like the system, as limited is it may be at this point.

Originally Posted by Efthreeoh

I'll have to watch the videos in the morning as I'm bandwidth limited during the day, but I did look at the first 15 minutes of the first video and didn't hear any new paradigm information; I heard sensors and 3D mapping, all current stuff. So anyone interested in the topic and who is pretty well read in it, and has a background in transportation and traffic control may get a bit bored in the beginning of the talk. But I'll give it a view tomorrow.

In my view Mobileye's system is the right approach. But it's a competitive field and many corps don't want to spill the beans on their secret sauce, rather only provide evidence that they are successfully working through the current challenges.

Originally Posted by Efthreeoh

What I do know is it might be better to spend all the hundreds of billions of research money that will take to possibly implement an autonomous system, on simply teaching humans how to drive better. Maybe an educational course a few steps beyond a high school football coach teaching driver's ed during summer break. And perhaps the audacity to have real requirements for students/drivers to pass and keep passing to maintain their license and knock off the idea the driving is a "privilege". Perhaps developing systems that provide drivers of more information to make better choices. Perhaps use the money to design and build better roads that don't lead to accidents.

With autonomous driving *millions* of people may not need to even own a car in the first place, so teaching them to drive better would be worthless. Cab fees would not have to include driver wages. Pretty much all goods would cost less due to the reduced embedded transportation costs. The disabled and elderly could have the ability to travel by only needing to know where they were headed.

Originally Posted by Efthreeoh

But back to the tech... the whole idea of taking the human out of the act of driving is to condense traffic so it can flow better on the limited amount of lane space available (exactly what the FAA is currently doing with its NextGen Air traffic Control System). The concept behind NextGen is to get more precise aircraft position data and give pilots more situational awareness. The DOT is not going to think any different about ground vehicular traffic.

Google's idea is to allow people to text and surf the internet while driving (err... travelling).

Yes, taking human out of driving can allow for better synchronization of traffic and determination of a commute duration. But it can also allow for ad hoc vehicle pooling with potentially less total traffic (for mass-transit options), and greatly reduced commuting costs.

[Efthreeoh]

Originally Posted by zenmaster

Such a pilot is making a very bad assumption. The point of the tech is to assist the driver in keeping the lane, and a safe distance from other vehicles. It will catch things like blind spot occupancy and cars changing lanes into your car. This is a very useful feature and one I'd like my car to have. If semi-autonomous is not useful to you personally, then you can wait a while for fully autonomous tech.

It's "human nature" to smoke cigarettes, play with fire, and eat like a pig too. There's no culpability here on the part of Tesla due to a driver not following extremely simple requirements. If you insist on shifting blame for clearly irresponsible actions then you are part of a thought pattern that serves to reduce freedom (and dignity) for all. Most sane people would probably agree that the system is 100% safe if used correctly. Many Tesla drivers really like the system, as limited is it may be at this point.

In my view Mobileye's system is the right approach. But it's a competitive field and many corps don't want to spill the beans on their secret sauce, rather only provide evidence that they are successfully working through the current challenges.

With autonomous driving *millions* of people may not need to even own a car in the first place, so teaching them to drive better would be worthless. Cab fees would not have to include driver wages. Pretty much all goods would cost less due to the reduced embedded transportation costs. The disabled and elderly could have the ability to travel by only needing to know where they were headed.

Yes, taking human out of driving can allow for better synchronization of traffic and determination of a commute duration. But it can also allow for ad hoc vehicle pooling with potentially less total traffic (for mass-transit options), and greatly reduced commuting costs.

I went through most of the Mobileye video up to about 43 minutes in and all the video spoke about was the company's artificial vision technology, which is completely opposite of what I'm discussing. All I got from the video was how the car drives in space and Mobileye's tech is using sensor data integrated with 3D map technology to recognize "free space" and keep the car placed there. All fine, but that is not autonomous driving in the sense - the end game - of what I am discussing, which is how is all traffic is managed to prevent collisions.

This is why I make an analogy to the US (and the World's) air traffic control system, because the air traffic control system is designed to prevent aircraft collisions. The system uses many different technologies as redundancy to each other, so if one tech fails, another tech is sitting in the background to take over. The air traffic control system works extremely well. It works well because air traffic is managed within the boundaries the technology allows (i.e. 5-mile separation, route planning, altitude separation). The system is extremely robust and can handle unpredictable situations, which mainly is weather (there are others); the system can reroute and manage transitional flight changes excellently, but all because the traffic is limited, meaning each flight is controlled where it can be and when it can be.

This is opposite to the purpose of personal mobility via personal automobile ownership, which is what most people enjoy about it, they can move about the country on their own free will. Even if the technology is developed to economically manage ground vehicular traffic through the combination of in-car autonomous control systems and a vehicular traffic control system, the public (well, Americans at least) may not accept that their travel is known to, and controlled by, the Government. This is what reduces freedom and dignity.

You've made statements above that Tesla's Autopilot works perfectly if used correctly, but isn't that the root cause of why some people want to take the human out of the act of driving and create autonomous driving, because humans don't "work perfectly". If humans did work perfectly then there would be no need to create autonomous driving. We all agree that if humans worked perfectly, the current ground transportation system would be 100% safe; most sane people would agree at least...

What I think is insane is advocating the driving responsibility from humans to a machine. A machine developed by humans who have no real first-person investment in the safety of the vehicle passengers. You think otherwise. Had the Tesla S not had Autopilot and Mr. Autopilot had been paying attention because he was actually and actively driving the car, I'm sure Mr. Autopilot would still be around to read this discussion. We do agree that Mr. Autopilot was not taking responsibility for his safe keeping, whether by not using Tesla's Autopilot application correctly, or just being a complete dumbass.

[Chihuahua]

Originally Posted by Fundguy1

.

Fuel economy should be driven by fuel prices.

Originally Posted by Efthreeoh

The cars today are engineered to pass high-MPG EPA tests in the lab (re: VW) without regard to longevity and total lifecycle costs.

So on point guys, these are the same points I bring up with friends/family when discussing cars and the environment.

[zenmaster]

Originally Posted by Efthreeoh

I went through most of the Mobileye video up to about 43 minutes in and all the video spoke about was the company's artificial vision technology, which is completely opposite of what I'm discussing.

I know, their tech doesn't depend on govt traffic monitoring, but rather crowd sourcing for map generation. However, you might need the govt to provide a means to indicate problems with govt operated traffic signals and situations where govt (police, construction, emergency workers) have taken control of traffic flow.

Originally Posted by Efthreeoh

All I got from the video was how the car drives in space and Mobileye's tech is using sensor data integrated with 3D map technology to recognize "free space" and keep the car placed there. All fine, but that is not autonomous driving in the sense - the end game - of what I am discussing, which is how is all traffic is managed to prevent collisions.

They are discussing the "end game". That is what they intend to provide and why they are in business. The method you are discussing is thankfully, totally unnecessary.

Originally Posted by Efthreeoh

This is why I make an analogy to the US (and the World's) air traffic control system, because the air traffic control system is designed to prevent aircraft collisions. The system uses many different technologies as redundancy to each other, so if one tech fails, another tech is sitting in the background to take over. The air traffic control system works extremely well. It works well because air traffic is managed within the boundaries the technology allows (i.e. 5-mile separation, route planning, altitude separation). The system is extremely robust and can handle unpredictable situations, which mainly is weather (there are others); the system can reroute and manage transitional flight changes excellently, but all because the traffic is limited, meaning each flight is controlled where it can be and when it can be.

The thing is, we don't need such an analogous system to improve safety for the highly localized, and much more independent automobile traffic.

Originally Posted by Efthreeoh

This is opposite to the purpose of personal mobility via personal automobile ownership, which is what most people enjoy about it, they can move about the country on their own free will. Even if the technology is developed to economically manage ground vehicular traffic through the combination of in-car autonomous control systems and a vehicular traffic control system, the public (well, Americans at least) may not accept that their travel is known to, and controlled by, the Government. This is what reduces freedom and dignity.

That's right and using Mobileye's approach, you don't need govt tracking (or any tracking). So no issue. Oh and you get to own your car too.

Originally Posted by Efthreeoh

You've made statements above that Tesla's Autopilot works perfectly if used correctly, but isn't that the root cause of why some people want to take the human out of the act of driving and create autonomous driving, because humans don't "work perfectly".

There is the obvious implication that, in the future, autonomous driving can be potentially safer than humans driving. BTW, the "working perfectly" is within the constraints of the stated limitations which of course requires human attention and will for some time.

Originally Posted by Efthreeoh

If humans did work perfectly then there would be no need to create autonomous driving.

According to you maybe. But many people want a chauffeur. That statement is like saying, "if humans washed dishes perfectly, there would be no need to create dishwashers."

Originally Posted by Efthreeoh

We all agree that if humans worked perfectly, the current ground transportation system would be 100% safe; most sane people would agree at least...

That's as absurd as saying,
if humans worked perfectly, there would be no need for cars in the first place." Because anything that a human could not do would mean they are not perfect. If the human did not anticipate the landslide, they are not perfect. If the human did not see through the fog, or swerve to avoid the tree, they are not perfect and so on.

Originally Posted by Efthreeoh

What I think is insane is advocating the driving responsibility from humans to a machine. A machine developed by humans who have no real first-person investment in the safety of the vehicle passengers.

Whether people advocate it or not is irrelevant. You'd better get used to this paradigm as it is coming eventually whether or not you like it. Too much money making potential and too many people wanting the feature. Of course there will be whining luddites with their irrational FUD, blabbing away about irrelevant matters.

Any assisted driving or full autonomy features have to be introduced with proof of safety. The primary safety concern should be "can we adequately determine safety criteria". This would seem to be difficult to do comprehensively, considering the variety of driving conditions possible. I bet auto manufacturers wind up sticking with the level 3 label even if full autonomy works well in 99% of driving, or the industry creates new sub levels between 3 and 4 which deal with more granular capabilities rather than overall safety.

[Efthreeoh]

Originally Posted by zenmaster

I know, their tech doesn't depend on govt traffic monitoring, but rather crowd sourcing for map generation. However, you might need the govt to provide a means to indicate problems with govt operated traffic signals and situations where govt (police, construction, emergency workers) have taken control of traffic flow.

They are discussing the "end game". That is what they intend to provide and why they are in business. The method you are discussing is thankfully, totally unnecessary.

The thing is, we don't need such an analogous system to improve safety for the highly localized, and much more independent automobile traffic.

That's right and using Mobileye's approach, you don't need govt tracking (or any tracking). So no issue. Oh and you get to own your car too.


There is the obvious implication that, in the future, autonomous driving can be potentially safer than humans driving. BTW, the "working perfectly" is within the constraints of the stated limitations which of course requires human attention and will for some time.

According to you maybe. But many people want a chauffeur. That statement is like saying, "if humans washed dishes perfectly, there would be no need to create dishwashers."

That's as absurd as saying,
if humans worked perfectly, there would be no need for cars in the first place." Because anything that a human could not do would mean they are not perfect. If the human did not anticipate the landslide, they are not perfect. If the human did not see through the fog, or swerve to avoid the tree, they are not perfect and so on.

Whether people advocate it or not is irrelevant. You'd better get used to this paradigm as it is coming eventually whether or not you like it. Too much money making potential and too many people wanting the feature. Of course there will be whining luddites with their irrational FUD, blabbing away about irrelevant matters.

Any assisted driving or full autonomy features have to be introduced with proof of safety. The primary safety concern should be "can we adequately determine safety criteria". This would seem to be difficult to do comprehensively, considering the variety of driving conditions possible. I bet auto manufacturers wind up sticking with the level 3 label even if full autonomy works well in 99% of driving, or the industry creates new sub levels between 3 and 4 which deal with more granular capabilities rather than overall safety.

Having worked in the engineering field all my professional life I've witnessed countless times were engineers have considered their solution to be perfect, even five-9's perfect. Algorithms are not perfect (designed by humans), chipsets are not perfect (designed and manufactured by humans), software is not perfect (written by humans). All of it is subject to defect. Mobileye is no different and will be no different. Nothing on their website shows me any paradigm change in collision avoidance. I'm not going to waste time looking at the video again past 45 minutes. If Mobileye didn't present a paradigm changing traffic control system within 45 minutes, I can't see where the last 10 minutes will introduce the paradigm change. What I see is a company that believes it can create a machine that drives better than a human. What happens when the local county government deploys an excavation team to repair a sink hole caused by a broken watermain? What happens if the county forgets update the crowd-sourced map? What happens if the Mobileye doesn't for some reason recognize the construction site; defect, remember? If people are riding around in their autonomous automobiles not paying attention to traffic and are fortunate enough to not collide with the construction site, would they then update the crowed-sourced map? Would they care even if they happened to notice the construction zone?

I could write thousands of these scenarios, but never all of them; and that's the point, Mobileye will never create artificial vision and AI that will prevent all accidents. I don't see their redundancy. Stop throwing insults and explain it, I'm all ears (eyes). It's laughable that you insinuate I'm a Luddite when I've been working for the last 9 years on the NextGen program. And before that on the Navy's first automated machinery control system. I'm way fucking the opposite of a Luddite, so drop it.

Sorry, but I'll trust a responsible driver's skills, experience, and intelligence over anyone's AI. The paradigm to change is the one that driving is a privilege. No, it's a Right. You get taxed to drive, therefore under the Constitution you have a right to drive. With Rights come Responsibilities. Threaten to take people's right to drive away because they are not responsible enough to do so, and see how much better they'll drive. Lowering accident rates is a behavior problem, not a tech problem. Mobileye et al. are making it a tech issue so it can solve an engineering problem and make a profit (all for profit making BTW). I'm all for technology that can enhance the driver's situational awareness. Not in all instances does technology solve behavioral problems.

[Fundguy1]

Hey, happy electronic guys. I do not see electric vehicles passing 1%of worldwide sales. Now or any time soon. I do not see any reason they will or should economically or mechanically. I also do not see the operators of cars being replaced by computers in a majoritive any time soon, nor do I feel they should ever totally be replaced. Until these things happen, both are a novelty, nothing more. And hopefully it stays that way for at least the next 50+ yrs.

[Whyte901]

As someone who's tried (two weeks) and ordered a 330e I realise they aren't to everyone requirement but for some they make a very compelling case.

My commute to work is around 6 miles each way, I have to have a company car as my job sometimes requires onsite work at client premises.
I can therefore get to and from work on a single charge (I found the charge to be nearer 17-20 mile in range using Auto E) I also have ability to charge at work but don't require it although I did use it to test.
I live in a hilly area of England (Yorkshire) and for me the range suits my needs, Would I love more range yes (I would also like faster charging) but like everything in life it requires some compromise.

So what was the attraction of the car? (For Me..) Honestly.. the tax savings, The BIK (Benefit in Kind) rate with this vehicle due to the advertised emissions means its much cheaper even if the MPG is overall lower for private mileage (Which it definetly will be than a 320/330d)

To put that into perspective;
A 318d ED would calculate as the following
BMW 320d Touring 163 Efficient Dynamics - 22% BIK Monthly Car Tax Liability £118 (Or £237 for 40% Tax Payer)
BMW 330d Touring M Sport Auto - 27% BIK Monthly Car Tax Liability £176 (Or £352 for 40% Tax Payer)
BMW 330e M Sport BIK 7% Monthly Car Tax Liability £42 (Or £85 for 40% Tax Payer)

So in my case a fully loaded 3 Series but saving money each month, More than an ED edition but with more power. Yes the MPG will be lower on longer journey but over the year the savings on private mileage in £££s in.

[Darkfiber]

Originally Posted by Whyte901

As someone who's tried (two weeks) and ordered a 330e I realise they aren't to everyone requirement but for some they make a very compelling case.

My commute to work is around 6 miles each way, I have to have a company car as my job sometimes requires onsite work at client premises.
I can therefore get to and from work on a single charge (I found the charge to be nearer 17-20 mile in range using Auto E) I also have ability to charge at work but don't require it although I did use it to test.
I live in a hilly area of England (Yorkshire) and for me the range suits my needs, Would I love more range yes (I would also like faster charging) but like everything in life it requires some compromise.

So what was the attraction of the car? (For Me..) Honestly.. the tax savings, The BIK (Benefit in Kind) rate with this vehicle due to the advertised emissions means its much cheaper even if the MPG is overall lower for private mileage (Which it definetly will be than a 320/330d)

To put that into perspective;
A 318d ED would calculate as the following
BMW 320d Touring 163 Efficient Dynamics - 22% BIK Monthly Car Tax Liability £118 (Or £237 for 40% Tax Payer)
BMW 330d Touring M Sport Auto - 27% BIK Monthly Car Tax Liability £176 (Or £352 for 40% Tax Payer)
BMW 330e M Sport BIK 7% Monthly Car Tax Liability £42 (Or £85 for 40% Tax Payer)

So in my case a fully loaded 3 Series but saving money each month, More than an ED edition but with more power. Yes the MPG will be lower on longer journey but over the year the savings on private mileage in £££s in.

Not to mention the fact it's a lot more fun to drive than a 320d ...

[Whyte901]

Originally Posted by Datake

Not to mention the fact it's a lot more fun to drive than a 320d ...

Amen to that, And alot more goooooo!