California DMV will miss its deadline for driverless car regulations

State Senate initially wanted rules for public use by January 1.

Read the whole story

 

[Deleted member 441963]

But, what are the consequences? Will companies like Tesla and Google have to change their plans?

Google seems ready to take to the road, completing a prototype days before regulations should have come into place. Do they need to postpone testing it 'outside'?

 

[robert.walter]

As this is something that will, eventually, have to be regulated by NHTSA via the FMVSS, I don't know why the Feds are not taking a more active role.

 

[robert.walter]

burne_[/url]":1107mkpl]But, what are the consequences? Will companies like Tesla and Google have to change their plans?

Google seems ready to take to the road, completing a prototype days before regulations should have come into place. Do they need to postpone testing it 'outside'?

Laguna Seca.

 

[NathanMattise]

burne_[/url]":24oofxjy]But, what are the consequences? Will companies like Tesla and Google have to change their plans?

Google seems ready to take to the road, completing a prototype days before regulations should have come into place. Do they need to postpone testing it 'outside'?

My understanding: the testing these companies do is already regulated, so Google's prototype can take the roads as intended. The "testing regulations" were put in place this fall.

However, this deadline was for public regulations... and self-driverless cars for public consumption are still a bit away: http://meincmagazine.com/cars/2014/10/pre ... iving-car/

 

[aiken_d]

Well, it *is* the California DMV. This is an organization that has raised bureaucratic inefficiency to an internationally recognized art form.

For all we know, the regulations are done and brilliant, and the guy delivering them is still stuck waiting in the line for window #4 so he can be told whether he needs to wait in the line for window #11 (which provides a blank form needed at window #8) or window #14 (which provides a worksheet needed at window #4 in order to get the right referral).

 

[Volante]

Headline is slightly misleading...the only thing the DMV is missing is on regulations to sell cars to the public.

"the state passed a law that required the DMV to do two things: set standards for autonomous vehicle testing on California roads and publish rules for automakers to follow when selling self-driving cars in the state"

The DMV has done the first (as evident by Google, VW, et. al.). But since we don't know how safe these cars are, they literally cannot do the second. You can't regulate them for sale until they're safe, and you can't determine they're safe until there's loads of testing. Once the cars are ready for prime time, I'm sure the DMV will have things ready by then.

http://jalopnik.com/california-dmv-will ... 1674215369
http://bigstory.ap.org/article/10d2f6cb ... rless-cars

 

[Boskone]

While I'm all for self-driving cars (it's one of the few features that would cause me to replace my perfectly-functional vehicle), I'm not really eager to see rules released too early.

As one example: when should the driver be able to override the car, and vice-versa? My gut instinct would be to say "any time and totally", but at the same time there's a distinct benefit to automated collision avoidance...but what if the collision avoidance is more dangerous than just crashing? (Say, the driver recognizes a small patch of ice and oncoming traffic.) Humans are still really good at aggregating a lot of data and instinctively responding in a reasonable manner. So there's some degree of tension between "computers are faster and more rational than humans" and "humans can react more quickly on less data than computers".

And there's also the issue of how much drivers should be allowed to depend on self-driving cars. I mean, ideally, I'd like to be able to sleep while the car takes care of things, but is that necessarily reasonable? maybe conditionally (e.g. most of my commute is a long, straight highway). Are the conditions that could be coded so that the car under some instances (the above ice example, or maybe bad traffic conditions) requires the driver be ready to assume control by having hands on the steering wheel and feet on the pedals?

Personally, I'd like to see them phased in. Maybe they're first released for highway use (long, fairly predictable traffic), then allowed to run autonomous in low-traffic conditions on regular streets, then high-traffic highways, etc. Maybe 2-5 years between "updates" to allow for new data to be crunched, driving routines to be updated and changed, and so on.

tl;dr: I'd love to see self-driving cars, but I'm also all for releasing them in a controlled and tested manner.

 

[Svip]

Who is to blame if an autonomous car runs over a person?

 

[Frodo Douchebaggins]

Svip[/url]":2h69viv0]Who is to blame if an autonomous car runs over a person?

Given how so many of them act these days, probably the pedestrian.

 

[Sobad]

Boskone[/url]":2z61likn]While I'm all for self-driving cars (it's one of the few features that would cause me to replace my perfectly-functional vehicle), I'm not really eager to see rules released too early.

As one example: when should the driver be able to override the car, and vice-versa? My gut instinct would be to say "any time and totally", but at the same time there's a distinct benefit to automated collision avoidance...but what if the collision avoidance is more dangerous than just crashing? (Say, the driver recognizes a small patch of ice and oncoming traffic.) Humans are still really good at aggregating a lot of data and instinctively responding in a reasonable manner. So there's some degree of tension between "computers are faster and more rational than humans" and "humans can react more quickly on less data than computers".

And there's also the issue of how much drivers should be allowed to depend on self-driving cars. I mean, ideally, I'd like to be able to sleep while the car takes care of things, but is that necessarily reasonable? maybe conditionally (e.g. most of my commute is a long, straight highway). Are the conditions that could be coded so that the car under some instances (the above ice example, or maybe bad traffic conditions) requires the driver be ready to assume control by having hands on the steering wheel and feet on the pedals?

Personally, I'd like to see them phased in. Maybe they're first released for highway use (long, fairly predictable traffic), then allowed to run autonomous in low-traffic conditions on regular streets, then high-traffic highways, etc. Maybe 2-5 years between "updates" to allow for new data to be crunched, driving routines to be updated and changed, and so on.

tl;dr: I'd love to see self-driving cars, but I'm also all for releasing them in a controlled and tested manner.

I'd rather see the driver able to take over control only with the autonomous features failing. While humans may be able to process things better, there are so many terrible drivers on the road that them having control are not a good idea.

On Christmas day for example, one of my very good lady friends was in a car accident because she perceived someone merging into her so she jerked the wheel to the right, spun out and flipped her car. I'd wager she was on the phone as people frequently are or was distracted by something else.

I'll stick with the robots. Then I can retire my tank like truck and won't feel morally obligated to T-bone people that run red lights.

 

[Deleted member 441963]

NathanMattise[/url]":2p6jqb9q]

burne_[/url]":2p6jqb9q]But, what are the consequences? Will companies like Tesla and Google have to change their plans?

Google seems ready to take to the road, completing a prototype days before regulations should have come into place. Do they need to postpone testing it 'outside'?

My understanding: the testing these companies do is already regulated, so Google's prototype can take the roads as intended. The "testing regulations" were put in place this fall.

However, this deadline was for public regulations... and self-driverless cars for public consumption are still a bit away: http://meincmagazine.com/cars/2014/10/pre ... iving-car/

Google wants a few out there in the next year, and that could be considered 'testing' but they want a few hundred on the roads in 2016 and that makes them much more common than Hyuaras or Huracáns.

If Google puts an employee who's not in engineering inside a car and lets him or her use it to commute, that's no longer testing, right? So the DMV is getting close to hindering the actual use of DC's.

 

[benwiggy]

It's 2015. Where's my driverless flying car?

 

[Deleted member 441963]

benwiggy[/url]":2j7o7ons]It's 2015. Where's my driverless flying car?

You dunked your Hummer H2 for it's gas milage. A flying car is worse. 11 mpg versus 8 mpg. And: add on top that you can spend as much as $30,000 a year on mandatory service checks by FAA-certified mechanics.

 

[Derecho Imminent]

Svip[/url]":3eydk0vm]Who is to blame if an autonomous car runs over a person?

Same person responsible if your parked car starts rolling down the hill and runs over someone. You are responsible.

edit: unless you can prove that the manufacturer was negligent to the point where they could be held responsible.

 

[Derecho Imminent]

aiken_d[/url]":rbr2go68]Well, it *is* the California DMV. This is an organization that has raised bureaucratic inefficiency to an internationally recognized art form.

Since you are knocking the Ca DMV why dont you enlighten us on how many other states are ahead of them in the process of regulation of driverless cars?

 

[CraigJ ✅]

aiken_d[/url]":3cimtw0n]Well, it *is* the California DMV. This is an organization that has raised bureaucratic inefficiency to an internationally recognized art form.

For all we know, the regulations are done and brilliant, and the guy delivering them is still stuck waiting in the line for window #4 so he can be told whether he needs to wait in the line for window #11 (which provides a blank form needed at window #8) or window #14 (which provides a worksheet needed at window #4 in order to get the right referral).

The people down-voting this have clearly never waited in line for two hours at the Laguna Hills DMV for their "appointment", only to have the clerk at the window take their break at the exact instant it;e your turn... And that's on a good day.

 

[CraigJ ✅]

adamrussell[/url]":20wisgf0]

aiken_d[/url]":20wisgf0]Well, it *is* the California DMV. This is an organization that has raised bureaucratic inefficiency to an internationally recognized art form.

Since you are knocking the Ca DMV why dont you enlighten us on how many other states are ahead of them in the process of regulation of driverless cars?

DOn't know about driverless cars, but the DMV offices in AZ are highly efficient and speedy, and the smog inspection is hugely streamlined with state run stations. Light years ahead of California. And cheaper too...

 

[DStaal]

Sobad[/url]":2npnq31k]

Boskone[/url]":2npnq31k]While I'm all for self-driving cars (it's one of the few features that would cause me to replace my perfectly-functional vehicle), I'm not really eager to see rules released too early.

As one example: when should the driver be able to override the car, and vice-versa? My gut instinct would be to say "any time and totally", but at the same time there's a distinct benefit to automated collision avoidance...but what if the collision avoidance is more dangerous than just crashing? (Say, the driver recognizes a small patch of ice and oncoming traffic.) Humans are still really good at aggregating a lot of data and instinctively responding in a reasonable manner. So there's some degree of tension between "computers are faster and more rational than humans" and "humans can react more quickly on less data than computers".

And there's also the issue of how much drivers should be allowed to depend on self-driving cars. I mean, ideally, I'd like to be able to sleep while the car takes care of things, but is that necessarily reasonable? maybe conditionally (e.g. most of my commute is a long, straight highway). Are the conditions that could be coded so that the car under some instances (the above ice example, or maybe bad traffic conditions) requires the driver be ready to assume control by having hands on the steering wheel and feet on the pedals?

Personally, I'd like to see them phased in. Maybe they're first released for highway use (long, fairly predictable traffic), then allowed to run autonomous in low-traffic conditions on regular streets, then high-traffic highways, etc. Maybe 2-5 years between "updates" to allow for new data to be crunched, driving routines to be updated and changed, and so on.

tl;dr: I'd love to see self-driving cars, but I'm also all for releasing them in a controlled and tested manner.

I'd rather see the driver able to take over control only with the autonomous features failing. While humans may be able to process things better, there are so many terrible drivers on the road that them having control are not a good idea.

On Christmas day for example, one of my very good lady friends was in a car accident because she perceived someone merging into her so she jerked the wheel to the right, spun out and flipped her car. I'd wager she was on the phone as people frequently are or was distracted by something else.

I'll stick with the robots. Then I can retire my tank like truck and won't feel morally obligated to T-bone people that run red lights.

Long-term that's probably what we'll move towards, but in the short term we know the risks, results, and error rates of having human drivers, while we don't with computer drivers. I'd rather err on the the side of to much human control until we've proven the tech thoroughly.

(And to any California legislators out there reading this: I would advise third-party testing at this point. Outsource it so that you don't have to cover the expense directly, but don't let the same people who programmed the driver test it - they'll only think of tests they've already thought of the solutions to. You need a different set of minds on the testing, so they come up with different conditions for the test.)

 

[SinclairZX81]

aiken_d[/url]":2fovobkr]Well, it *is* the California DMV. This is an organization that has raised bureaucratic inefficiency to an internationally recognized art form.

For all we know, the regulations are done and brilliant, and the guy delivering them is still stuck waiting in the line for window #4 so he can be told whether he needs to wait in the line for window #11 (which provides a blank form needed at window #8) or window #14 (which provides a worksheet needed at window #4 in order to get the right referral).

I see you've spent some time at the California DMV lately. If any of you downvoters think this is exaggeration, it's not.

 

[Carewolf]

burne_[/url]":36h0z80f]

benwiggy[/url]":36h0z80f]It's 2015. Where's my driverless flying car?

You dunked your Hummer H2 for it's gas milage. A flying car is worse. 11 mpg versus 8 mpg. And: add on top that you can spend as much as $30,000 a year on mandatory service checks by FAA-certified mechanics.

A flying car isn't 8mpg it is 8gpm.

 

[Deleted member 441963]

Carewolf[/url]":37xtptca]

burne_[/url]":37xtptca]

benwiggy[/url]":37xtptca]It's 2015. Where's my driverless flying car?

You dunked your Hummer H2 for it's gas milage. A flying car is worse. 11 mpg versus 8 mpg. And: add on top that you can spend as much as $30,000 a year on mandatory service checks by FAA-certified mechanics.

A flying car isn't 8mpg it is 8gpm.

A Cesna 172 consumes 7 gallons per hour and travels 107 nautical miles in that time. That comes to 11 mpg. But you need to speed up and climb, and slow down while descending, and that eats into your mpg, so for a four hour flight 8 miles per gallon is a reasonable estimate.

A Cesna 172 is quite efficient, and it's unlikely a car would do much worse or better, unless you choose for moronic features like VTOL.

I wasn't kidding when I said 8mpg...

(One thing I forgot the mention is the price of your fuel. 100LL goes for $8 a gallon.)

 

[mewmew]

I just hope the regs are sensible. I remember reading that they were considering requiring a person to be in the driver seat at all times with the ability to override the automated controls at any time, which is a terrible idea I think. All it will take is a jumpy person who doesn't trust his car to cause a wreck, and I'm confident that THAT would be a far larger cause of auto accidents involving these cars than anything the car itself does.

 

[Martin Blank]

burne_[/url]":1bmgjlve]

Carewolf[/url]":1bmgjlve]

burne_[/url]":1bmgjlve]

benwiggy[/url]":1bmgjlve]It's 2015. Where's my driverless flying car?

You dunked your Hummer H2 for it's gas milage. A flying car is worse. 11 mpg versus 8 mpg. And: add on top that you can spend as much as $30,000 a year on mandatory service checks by FAA-certified mechanics.

A flying car isn't 8mpg it is 8gpm.

A Cesna 172 consumes 7 gallons per hour and travels 107 nautical miles in that time. That comes to 11 mpg. But you need to speed up and climb, and slow down while descending, and that eats into your mpg, so for a four hour flight 8 miles per gallon is a reasonable estimate.

A Cesna 172 is quite efficient, and it's unlikely a car would do much worse or better, unless you choose for moronic features like VTOL.

I wasn't kidding when I said 8mpg...

(One thing I forgot the mention is the price of your fuel. 100LL goes for $8 a gallon.)

You're off on a few points.

A C172 at 107 knots (which is actually a little slow--I usually cruise around 115-120 knots) at 7GPH is 15 nautical miles per gallon, not 11. It's more than 17 statute miles per gallon. It's not as good as you'll get in a decent sedan (and your payload will be lower), but it is faster and you can usually go in straighter lines between points.

Climbing costs extra fuel, but descent does not. We cut the throttle way back on descent; it's not hard to maintain 105 knots in a 500fpm descent with the engine at very low RPM. Slowing is done by cutting the throttle even further (back to idle if necessary) and pitching up a bit, or by deploying the flaps (appropriate to indicated airspeed, of course).

And annuals are, for most small planes, nowhere near $30K. You can buy some small planes for that much. Annuals for a C172 or a C182 are a few hundred dollars, while even a Beech Bonanza should be around a thousand or so (both cost sets presuming nothing major needs to be fixed).

Finally, fuel prices for 100LL are down to around the $5 mark. It's more expensive because of the additional testing, and because only a handful of refineries still produce it. Hopefully, the work on removing lead from avgas will see some good success soon, and the Part 21 rewrite will allow some new approaches on the aircraft design front.

 

[Derecho Imminent]

mewmew[/url]":3gp5kh0x]I just hope the regs are sensible. I remember reading that they were considering requiring a person to be in the driver seat at all times with the ability to override the automated controls at any time, which is a terrible idea I think. All it will take is a jumpy person who doesn't trust his car to cause a wreck, and I'm confident that THAT would be a far larger cause of auto accidents involving these cars than anything the car itself does.

That may be true. As well, its just human nature that if the car drives itself people will stop monitoring it and I guarantee you 90% of all people will NOT be ready to jump in if it actually became necessary.

 

[Deleted member 441963]

Martin Blank[/url]":rc4qv8ht]
And annuals are, for most small planes, nowhere near $30K.

Thanks for the additions and corrections. I do have one remaining niggle: annuals etc.

That figure was quoted by a Cessna pilot in a discussion about flying cars and using them for daily commuting. Flying a plane three hours per day, five days a week will massively increase wear and the service your plane needs compared to occasional weekend use. Engine rebuild every four years. Fresh pair of tyres twice a year. Have a mechanic replace your oil every 35 hours. That's twice a month. One of the airports you use will charge you for landings. With 440 landings per year that will add up.

(I'm pretty sure we won't see aircar/planes landing on the parking lot behind your company any time soon.)

 

[bb-15]

mewmew[/url]":2c4z9v5q]I just hope the regs are sensible. I remember reading that they were considering requiring a person to be in the driver seat at all times with the ability to override the automated controls at any time, which is a terrible idea I think. All it will take is a jumpy person who doesn't trust his car to cause a wreck, and I'm confident that THAT would be a far larger cause of auto accidents involving these cars than anything the car itself does.

I think the requirement that the driver be able to override automatic controls at anytime is a good one.
I spoke with an Audi mechanic who told me about ways that this could be done (such as the need for the driver to be in the driver's seat for the automatic controls to work).

Responsibility for an automatically controlled vehicle is still with the driver.
The driver still has to have the ability to override the automatic system if something goes very wrong with automated controls.
- Think of all the problems with GPS/map navigation. Every mapping service sometimes can make a big mistake.
The user still has to be in charge to monitor and correct the automatic system.

 

[Martin Blank]

burne_[/url]":15ev4h3f]

Martin Blank[/url]":15ev4h3f]
And annuals are, for most small planes, nowhere near $30K.

Thanks for the additions and corrections. I do have one remaining niggle: annuals etc.

That figure was quoted by a Cessna pilot in a discussion about flying cars and using them for daily commuting. Flying a plane three hours per day, five days a week will massively increase wear and the service your plane needs compared to occasional weekend use. Engine rebuild every four years. Fresh pair of tyres twice a year. Have a mechanic replace your oil every 35 hours. That's twice a month. One of the airports you use will charge you for landings. With 440 landings per year that will add up.

(I'm pretty sure we won't see aircar/planes landing on the parking lot behind your company any time soon.)

Annual inspections are a single part of annual costs. You can't take all of the costs of a plane over the course of a year and refer to it as "annuals" because that confuses things. As long as the inspector doesn't find anything major, you're not going to pay much for it. That's separate from the things you may notice need fixing, like tire replacements.

Pilots can do minor maintenance like replacing the oil, spark plugs, and a few other things. And not all airports you fly into will charge for landings. During my time in California, only one airport charged for landings, and that was Catalina. (Airports I landed at in California included FUL, AJO, CCB, CNO, DAG, EMT, HMT, ONT, POC, PSP, VCV, F70, L35, and L65, and in Texas included ADS, DTO, and TKI.) I landed several times at Ontario International Airport, and didn't get charged. Most small airports make their money off of parking and hangar fees, not landing fees, especially since so many are untowered and have no reliable way of tracking landings.

Engine rebuilds are typically every 2000-2500 hours, so your rebuild times will vary significantly based on the amount you fly. At 70 hours per month, you could see a rebuild every 2-3 years. But who, able to fly largely a straight line, would put in that much time? Realistically, you're looking at much shorter travel times, in many cases flying at 1000' AGL or even less. A 100-mile commute becomes a 60-70 mile commute, so the two-hour trip becomes maybe an hour, factoring in time in the pattern and waiting for taxiway access. Now you're down to 35 hours per month, or an annual every 5-6 years.

But that's also all under the old rules. The Part 21 rewrite may significantly change all of that, and we could see 5000--6000 hours between rebuilds as general aviation engine technology catches up. (We're still operating largely on 1970s to early-1980s technology in many cases, even when it comes to fuel injection.) With FADEC and other management technologies, engine life could get a lot better, and flying a lot cheaper.

 

[Deleted member 441963]

bb-15[/url]":1fvenmei]
- Think of all the problems with GPS/map navigation. Every mapping service sometimes can make a big mistake.

The thing that sets Google (and all similar attempts by others since) apart from previous attempts is that they tried to integrate data provided by all sensors. Google published a video about it. It shows how it uses LIDAR and optical cameras to orient itself on the road and its surroundings, how is uses LIDAR and camera's to read traffic signs, how it uses GPS to navigate but camera's to stop it turning right into the path of cyclists etc.

I'm pretty sure that a glaring mistake on the map won't make your google car drive over the edge of a cliff or into a wall. It will notice the problem, stop safely and if the problem doesn't go away in 30 seconds ask you if you want to try another route.

Having good sensors is key. Allow me to explain how TomTom uses sensors to enrich GPS data. GPS is only moderately accurate behind a windscreen with metal coatings on it. When you drive away your position on the map might be off by a hundred yards. Two corners later, it's spot on. How do they do that?

You need an accelerometer and a gyroscope and two 90º corners to find the error between the map and what GPS is reporting and correct it to within a few feet. The assumption is that you will be on the center of the lane of the road you're driving on, and a 90% turn will have you point out the center of the righthand lane to the system within a few feet. The next corner will calibrate the other axis. DPGS accuracy for a few dollars and some clever thinking.

 

[mewmew]

adamrussell[/url]":2qnql8ls]

mewmew[/url]":2qnql8ls]I just hope the regs are sensible. I remember reading that they were considering requiring a person to be in the driver seat at all times with the ability to override the automated controls at any time, which is a terrible idea I think. All it will take is a jumpy person who doesn't trust his car to cause a wreck, and I'm confident that THAT would be a far larger cause of auto accidents involving these cars than anything the car itself does.

That may be true. As well, its just human nature that if the car drives itself people will stop monitoring it and I guarantee you 90% of all people will NOT be ready to jump in if it actually became necessary.

This is exactly the problem. The driver will be sitting there being driven around staring down at his phone or something, the car will make some maneuver to avoid a pothole or rapidly decelerate to avoid rearending a suddenly stopped car, the driver will look up with no idea what's going on and think the car is malfunctioning, yank the steering wheel to avoid what he thinks will be a crash, and end up causing the actual crash.

At the very least a person should need a separate license to be allowed to be the person who can take over controls on an autodriving car, requiring a course that pounds into his head that the car will know exactly what it's doing 999,999 times out of a million.

 

[new2mac]

We need to start making the distinction:

Self-driving car: automated driving, with a human driver present.

Driverless car: automated driving, without a human driver present.

The ramification of each is hugely different. If humanity adopts the driverless car, it will have massive ramifications to civilization as we know it. A self-driving car, is merely a more convenient and safer car.

 

[D.Becker]

I can understand the initial concern about safety, responsibility and liability. At first it seems an overwhelming barrier to self-driving cars ever being allowed on the road.

But once you start thinking about the possibility, watch other drivers, and objective observe your own behavior, you start to realize that even an imperfect system will be an improvement.

I'm frequently in traffic so erratic that it's not safe to take my eyes off the car in front of me, even for long enough to check my mirrors and turn my head before changing lanes. Leaving a two second space is unreasonable, and is unsafe because it encourages other drivers to change lanes in front of you (and often immediately brake, because they want a 1.0 to 1.2 second space, which is the norm in heavy traffic around here). I could be 100% alert, and a self-driving car would still be safer.

Now consider all of the times where people aren't 100% alert. Or just plain bad drivers. Even though it's illegal for years, I still regularly see people holding cell phones to their faces. It seems like a good fraction of drivers, although I suspect it's just because I'm more likely to look at drivers of cars that can't stay in their lane, or are driving below the speed of other traffic in that lane.

Now further consider drunk drivers, or people that are a drink short of being legally drunk. Self-driving cars would eliminate enough accidents that they could be somewhat flawed, yet still be a net win for society. That doesn't help the liability/blame aspect when a specific, known person is a victim, and we don't know the names of the people that never got into an accident because of the technology.

My expectation is that it will only take a few years after self-driving cars are introduced before they are accepted by most people as dramatically safer than manual driving. There will always be a few hold-outs, just as there are still people that believe they can out-brake a 4 channel ABS system.

 

[bb-15]

burne_[/url]":7ukuc1en]

bb-15[/url]":7ukuc1en]
- Think of all the problems with GPS/map navigation. Every mapping service sometimes can make a big mistake.

The thing that sets Google (and all similar attempts by others since) apart from previous attempts is that they tried to integrate data provided by all sensors. Google published a video about it. It shows how it uses LIDAR and optical cameras to orient itself on the road and its surroundings, how is uses LIDAR and camera's to read traffic signs, how it uses GPS to navigate but camera's to stop it turning right into the path of cyclists etc.

I'm pretty sure that a glaring mistake on the map won't make your google car drive over the edge of a cliff or into a wall. It will notice the problem, stop safely and if the problem doesn't go away in 30 seconds ask you if you want to try another route.

You are discussing safety.
I was writing about navigation and the need for the driver to take over the car when the automatic driviing system makes a mistake.

If the automatic system doesn't get the driver where he/she needs to go because of some error, then the driver needs to take over.
There are certain parts of the world where using GPS navigation can be problematic.
For instance when I was in Puerto Rico all my mapping programs (Google, Apple and Mapquest) had problems and I was switching between them.
- A self driving car in those conditions would need to be alert and ready to take over driving the car.

 

[bb-15]

burne_[/url]":w4v636j0]

bb-15[/url]":w4v636j0]
- Think of all the problems with GPS/map navigation. Every mapping service sometimes can make a big mistake.

The thing that sets Google (and all similar attempts by others since) apart from previous attempts is that they tried to integrate data provided by all sensors. Google published a video about it. It shows how it uses LIDAR and optical cameras to orient itself on the road and its surroundings, how is uses LIDAR and camera's to read traffic signs, how it uses GPS to navigate but camera's to stop it turning right into the path of cyclists etc.

I'm pretty sure that a glaring mistake on the map won't make your google car drive over the edge of a cliff or into a wall. It will notice the problem, stop safely and if the problem doesn't go away in 30 seconds ask you if you want to try another route.

You are only focusing on safety.
When GPS navigation fails, the driver needs to take over the car.
If the automatic system doesn't get the driver where he/she needs to go because of some error, then the driver needs to do something.

There are certain parts of the world where using GPS navigation can be problematic.
For instance when I was in Puerto Rico all my mapping programs (Google, Apple and Mapquest) had problems and I was switching between them.
- The "driver" in a self driving car in those conditions would need to be alert and ready to take over the vehicle.

 

[Deleted member 441963]

bb-15[/url]":alv67xf8]There are certain parts of the world where using GPS navigation can be problematic.
For instance when I was in Puerto Rico all my mapping programs (Google, Apple and Mapquest) had problems and I was switching between them.
- The "driver" in a self driving car in those conditions would need to be alert and ready to take over the vehicle.

That will all depend on the quality of the map.

 

[Carewolf]

burne_[/url]":3lu1f60w]

Carewolf[/url]":3lu1f60w]

burne_[/url]":3lu1f60w]

benwiggy[/url]":3lu1f60w]It's 2015. Where's my driverless flying car?

You dunked your Hummer H2 for it's gas milage. A flying car is worse. 11 mpg versus 8 mpg. And: add on top that you can spend as much as $30,000 a year on mandatory service checks by FAA-certified mechanics.

A flying car isn't 8mpg it is 8gpm.

A Cesna 172 consumes 7 gallons per hour and travels 107 nautical miles in that time. That comes to 11 mpg. But you need to speed up and climb, and slow down while descending, and that eats into your mpg, so for a four hour flight 8 miles per gallon is a reasonable estimate.

A Cesna 172 is quite efficient, and it's unlikely a car would do much worse or better, unless you choose for moronic features like VTOL.

I wasn't kidding when I said 8mpg...

(One thing I forgot the mention is the price of your fuel. 100LL goes for $8 a gallon.)

An airplane is not a flying car. A flying car is something resembling a car flying. The aerodynamics of that is just crazy. The only way they can even get off the ground is using VTOL, which they will need everywhere. Flying car prototypes do exist, they are just not compelling. I was not kidding when I said 8gpm either.

 

[lee_machine]

robert.walter[/url]":3kfwncak]

burne_[/url]":3kfwncak]But, what are the consequences? Will companies like Tesla and Google have to change their plans?

Google seems ready to take to the road, completing a prototype days before regulations should have come into place. Do they need to postpone testing it 'outside'?

Laguna Seca.

As someone that lives 5 minutes from there I second this

 

[Derecho Imminent]

Have any insurance companies weighed in on this?

1. No car will be legal to drive without insurance.
2. If anyone understands the science of statistical safety its them. I may not trust them to pay out after and earthquake, but I do trust them to say when something is safe enough to insure.

 

[Deleted member 441963]

Carewolf[/url]":2wjo48g5]
The only way they can even get off the ground is using VTOL, which they will need everywhere.

You'll like the idea of VTOL in front of your driveway until you realise your neighbours car makes almost as much noise as yours and you can't tell him to get home around the same time you do.

Then you realise he has teenage kids with flying cars as well, and they arrive back from the pub at 2 or 3 am.

VTOL flying cars in residential neighbourhoods? It ain't going to happen, ever. Period.

(Noise is one thing. How are you going to make certain no VTOL flying car ever lands on your kid, grandma, dog? The first accident will wipe out the manufacturer with claims unless he holds you on site for a week while forcing you to sign reams of disclaimers. You're sure you're willing to do that?)