[Rondec]

There simply is no safety reason to track all the vehicles and none of the autonomous vehicles on the road today or in development depend on such tracking. They can't depend on tracking because none of the other vehicles on the road have it so none of the deisgn depend on tracking.

But even if tracking were implemented for other reasons it would not be that burdensome. The average car in the US only drives 33 miles per day -- probably only about 1 hour per day and not 50%. In contrast, there are 327 million mobile phones in the US and most phones are on 24 hours a day. So the bandwidth requirements would only add about 4% to the device tracking data load.

And as for car-to-car communication standards, it would seem that turn-signals, brake lights, and the occasional flash of the headlights suffice for humans and would suffice for autonomous vehicles, too. There's no reason to make this so complicated.

My insurance company offered me a discount (5 percent, I think) if I let them install a box on my vehicles that will watch my driving and upload details to them. This would include number of miles traveled, my driving patterns, and obviously any accidents or near misses.

I suppose this isn't surprising, but I think it is probably where things are headed, much more than government control. For the most part, the government steps in when something goes wrong -- if there is an accident or something like that, then they will get a download from your car's computer to see what happened.

[photoptimist]

My insurance company offered me a discount (5 percent, I think) if I let them install a box on my vehicles that will watch my driving and upload details to them. This would include number of miles traveled, my driving patterns, and obviously any accidents or near misses.

I suppose this isn't surprising, but I think it is probably where things are headed, much more than government control. For the most part, the government steps in when something goes wrong -- if there is an accident or something like that, then they will get a download from your car's computer to see what happened.

Interesting! Did you accept the tracking box?

There's not need for more government control. Both insurance companies and "safe" drivers have financial incentives to adopt these types of tracking technologies. As time goes on, unsafe drivers will simply see their premiums rise and rise because so many of the safe drivers will abandon the unmonitored high-risk pool.

[monochrome]

1. The essential element of personal liberty is that others may allow tracking (Waze) if they so choose, so long as I am allowed not to be tracked and not forced to suffer any dimunition of Liberty as a result of that decision.

2. Cars will use collective location positioning technology (static, fixed, position indicators like cell towers) so constant GPS monitoring is not necessary, and use RFID signals with each other to position themselves relative to each other and the last position indicator. Parked cars will actually assist this process in dense urban areas. The flock of birds metaphor is not only apt, it is the metaphor used to model the collective behavior algorithms.

ATM it appears the social cost of a System robust enough to respond to random external inputs is prohibitively expensive to society. As I wrote upthread, the current operating solution being discussed is to legislatively mandate confiscation of individual human-driven cars for the general welfare of all.

[ThorSanchez]

I have a co-worker who accepted the tracking box. They only wanted to sample his driving behavior, and he mailed it back to the company after some period like six weeks that they used to set rates. So he drove very carefully for six weeks, then didn't care quite so much.

---------- Post added 07-05-18 at 03:43 PM ----------

As I wrote upthread, the current operating solution being discussed is to legislatively mandate confiscation of individual human-driven cars for the general welfare of all.

By whom? It's vastly more likely that autonomy will be stunted and driven into tiny niche use cases by litigation and legislation.

[monochrome]

By whom? It's vastly more likely that autonomy will be stunted and driven into tiny niche use cases by litigation and legislation.

That will depend upon who has (it is deemed) standing to sue, and who are the legislators at the time.

.:

[PDL]

Aren't there already something like five billion cell phones in the world? They seem to operate pretty well in their allocated bandwidth without cooking us or our eggs.

Also (warning, very back-of-the-napkin calculation here) the sun irradiates the ground with something like 30 million times as much power as all the cell phones in the world transmitting at max power continuously.

Add into that the potential of say 300 million autonomous vehicles, several more 100's of millions IoT devices and other things. Yup that is a lot of bandwidth. Oh and your napkin needs a upgrade as microwave energy from the sun is a minute fraction of the total energy hitting the surface of the earth. ;-)

---------- Post added 07-05-18 at 03:02 PM ----------

And as for car-to-car communication standards, it would seem that turn-signals, brake lights, and the occasional flash of the headlights suffice for humans and would suffice for autonomous vehicles, too. There's no reason to make this so complicated.

So now autonomous vehicles are going to use turn signals to coordinate their movements. I think this is a bit more complicated than what you imagine. Just what signals would be used for an emergency stop, or coming up to a stop sign, or a normal right/left turn etc. They could use a flashing code sort of like when you change the channel on your TV (if you have one). I think that now is the time for all of these diverse companies to come together and build an equivalent to the IATA in order to share/develop protocols that allow for these vehicles to communicate with each other. Heaven help you if the emitter/sensor gets dirty (a la #36) or your particular vehicle is the only one that uses EBCDIC rather than Unicode-ASCII ;-)

[MarkJerling]

Here in Arizona - a couple of months ago in Tempe, one of Uber's cars ran over a lady that it didn't "see". It turns out that the development engineers disabled the automagic brake system - as it was apparently somewhat annoying - if I remember correctly. So, there is an urgency to get these things out on the street. There will be causalities (or collateral damages) - too bad for the lady, however her relatives will probably hit the lottery after the court case. Kind of reminds me of the Ford Pinto fiasco. Crispy critter a few folks and life goes on (the cost benefit analysis only shows it will cost us $xxxxxx - which is substantially less than installing a shim under the gas tank). Too bad if you are involved, though. ... and what of Uber? Life was too hot to handle here in Phoenix, so they pitched their tent - fired everyone, and moved to a city back east - where the political temperature was more to their liking.

I posted the video of that accident earlier in the thread. Unfortunately, she chose to cross a 45mph road at night, far away from any pedestrian crossing, in the dark and chose to do so right into the path of the Uber car. It's pretty hard to blame the technology when the Uber driver was supposed to be on the lookout for this sort of thing - but was not. The automatic braking system is (rightly or wrongly) disabled when there's a driver in the car. Uber and her family have, already, settled out of court. It was an unfortunate accident.

In 2017, 5,984 pedestrians died in the USA. About 60,000 were injured. That's around 180 pedestrians a day hurt or killed in the USA, almost all of them by ordinary motorists in ordinary cars. I very much doubt you can compare this 1/5984 deaths to a Ford Pinto situation.

[MarkJerling]

Will someone hack into the communications system and create occasional havoc? Probably. But we're currently staking our lives on 17-year-olds with a Big Gulp in one hand, an iPhone in the other, and the steering wheel being controlled by their knees.

Very true!

[MarkJerling]

Add into that the potential of say 300 million autonomous vehicles, several more 100's of millions IoT devices and other things. Yup that is a lot of bandwidth. Oh and your napkin needs a upgrade as microwave energy from the sun is a minute fraction of the total energy hitting the surface of the earth. ;-)

---------- Post added 07-05-18 at 03:02 PM ----------



So now autonomous vehicles are going to use turn signals to coordinate their movements. I think this is a bit more complicated than what you imagine. Just what signals would be used for an emergency stop, or coming up to a stop sign, or a normal right/left turn etc. They could use a flashing code sort of like when you change the channel on your TV (if you have one). I think that now is the time for all of these diverse companies to come together and build an equivalent to the IATA in order to share/develop protocols that allow for these vehicles to communicate with each other. Heaven help you if the emitter/sensor gets dirty (a la #36) or your particular vehicle is the only one that uses EBCDIC rather than Unicode-ASCII ;-)

The communication systems currently in development allow many modern cars to communicate wirelessly with cars nearby. (Many new cars in Japan do this already) The purpose is to reduce the risk accidents by having the car be aware of the traffic stream slowing ahead and allowing the car to slow and avoid a nose-to-tail pileup. In time, they think this technology could be used to also prevent many common intersection accidents. There is no intent , at this stage, to have this data somewhere in a "control centre". The idea is simply that the cars will communicate wirelessly and encrypted with each other - a hive intelligence reducing risk of accidents.

There's no need of turn signals or brake lights being "read" by the other cars with optical sensors. Autonomous cars will, of course, need optical and radar / lidar sensors to navigate the roads. How they keep those sensors clean I don't know!

[interested_observer]

I posted the video of that accident earlier in the thread. Unfortunately, she chose to cross a 45mph road at night, far away from any pedestrian crossing, in the dark and chose to do so right into the path of the Uber car. It's pretty hard to blame the technology when the Uber driver was supposed to be on the lookout for this sort of thing - but was not. The automatic braking system is (rightly or wrongly) disabled when there's a driver in the car. Uber and her family have, already, settled out of court. It was an unfortunate accident.

In 2017, 5,984 pedestrians died in the USA. About 60,000 were injured. That's around 180 pedestrians a day hurt or killed in the USA, almost all of them by ordinary motorists in ordinary cars. I very much doubt you can compare this 1/5984 deaths to a Ford Pinto situation.

If this was just any old run of the mill vehicle, I would absolutely agree with you. Having said that, this was just no ordinary vehicle. The National Transportation and Safety Board (NTSB), the same folks who review aircraft crashes are examining this accident and have not released a report of its findings. It will probably be at lease a few more months.

According to this article, Uber drivers were familiar with this area with jaywalkers and others, where the autonomous systems detected the folks on the street and acted appropriately or the driver having to take over due to the vehicle's confusion.

• Uber technology failure in fatal Tempe accident shocks experts

There are a few things about this crash which is very concerning to me.

• "The automatic braking system is (rightly or wrongly) disabled when there's a driver in the car." - If I would have known that, I personally would have gone down to the state capital and lobbied against Uber being allowed to test their vehicles on the public streets. Humans are good operators, but lousy monitors. If in fact true, having the driver be solely responsible for braking is absurd. One of the reasons autonomous system are in the vehicles is to address specific situations like that. The LIDAR system should have detected the lady and taken appropriate action - automatically. The area of the accident is just across the bridge from ASU (Arizona State University) - where you have students and folks there in the area doing stupid things all the time. This is not the area in which to be testing experimental technology at night. Given the above information - this was just an accident waiting to happen.

• I'm very aware of autonomous vehicles. That's as much as I can and will ever say - given the various non disclosure agreements (NDAs), neatly filed in the cabinet right next to me. I'm enjoying my retirement way too much to have it abruptly interrupted.

• After nearly 50 years of embedded real time system experience - this particular situation with the technology on the vehicle should have been easy to handle.

• Automated Collision Avoidance Systems - are a recent but reasonably mature technology currently available on vehicles in production from Ford, Chevrolet, Honda, Volkswagen, Nissan, Chrysler, etc. to name a few, and it's affordable. There is absolutely no reason for an Uber vehicle not to have this capability. These systems are designed to prevent just this type of accident. The 2016 Chevrolet Malibu has the available Front Pedestrian Alert Technology that detects and can brake. Why would anyone expect anything less from Uber if they are permitted to be testing on the public streets.

• 10 Cars with Automatic Emergency Braking Systems | Autobytel.com

I would indeed like to read the NTSB report on this. And, I am still of the opinion that this can be compared to the Ford Pinto. To me, at this time, it appears to be a matter of poor system design. Just my opinion.

[MarkJerling]

If this was just any old run of the mill vehicle, I would absolutely agree with you. Having said that, this was just no ordinary vehicle. The National Transportation and Safety Board (NTSB), the same folks who review aircraft crashes are examining this accident and have not released a report of its findings. It will probably be at lease a few more months.

According to this article, Uber drivers were familiar with this area with jaywalkers and others, where the autonomous systems detected the folks on the street and acted appropriately or the driver having to take over due to the vehicle's confusion.

• Uber technology failure in fatal Tempe accident shocks experts

There are a few things about this crash which is very concerning to me.

• "The automatic braking system is (rightly or wrongly) disabled when there's a driver in the car." - If I would have known that, I personally would have gone down to the state capital and lobbied against Uber being allowed to test their vehicles on the public streets. Humans are good operators, but lousy monitors. If in fact true, having the driver be solely responsible for braking is absurd. One of the reasons autonomous system are in the vehicles is to address specific situations like that. The LIDAR system should have detected the lady and taken appropriate action - automatically. The area of the accident is just across the bridge from ASU (Arizona State University) - where you have students and folks there in the area doing stupid things all the time. This is not the area in which to be testing experimental technology at night. Given the above information - this was just an accident waiting to happen.

• I'm very aware of autonomous vehicles. That's as much as I can and will ever say - given the various non disclosure agreements (NDAs), neatly filed in the cabinet right next to me. I'm enjoying my retirement way too much to have it abruptly interrupted.

• After nearly 50 years of embedded real time system experience - this particular situation with the technology on the vehicle should have been easy to handle.

• Automated Collision Avoidance Systems - are a recent but reasonably mature technology currently available on vehicles in production from Ford, Chevrolet, Honda, Volkswagen, Nissan, Chrysler, etc. to name a few, and it's affordable. There is absolutely no reason for an Uber vehicle not to have this capability. These systems are designed to prevent just this type of accident. The 2016 Chevrolet Malibu has the available Front Pedestrian Alert Technology that detects and can brake. Why would anyone expect anything less from Uber if they are permitted to be testing on the public streets.

• 10 Cars with Automatic Emergency Braking Systems | Autobytel.com

I would indeed like to read the NTSB report on this. And, I am still of the opinion that this can be compared to the Ford Pinto. To me, at this time, it appears to be a matter of poor system design. Just my opinion.

A prelim report has already been issued. The car recognised the person as a pedestrian 1.3 seconds before impact. At that time, the car was travelling at 39 miles per hour. That means that, had the automatic breaking not been disabled, the car may have braked as early as 22.6m before impact, but most certainly 17m before impact. But, the driver in the vehicle - who were supposed to have their eyes on the road - did not. The driver was either looking at the centre console or their phone, according to the in-car cameras. If the human driver had seen the person at 1.3 seconds prior to impact, their reaction time would likely have meant that they would not have braked in time anyway. So, it's likely that the outcome, if this was an ordinary car with a human driver looking at the road, would have been no different.

Like you, I don;t believe the automated braking system should have been disabled. It may have given the pedestrian a better chance as the car may have slowed - even slightly - and that may have reduced the severity of the accident.

[onlineflyer]

A prelim report has already been issued. The car recognised the person as a pedestrian 1.3 seconds before impact. At that time, the car was travelling at 39 miles per hour. That means that, had the automatic breaking not been disabled, the car may have braked as early as 22.6m before impact, but most certainly 17m before impact. But, the driver in the vehicle - who were supposed to have their eyes on the road - did not. The driver was either looking at the centre console or their phone, according to the in-car cameras. If the human driver had seen the person at 1.3 seconds prior to impact, their reaction time would likely have meant that they would not have braked in time anyway. So, it's likely that the outcome, if this was an ordinary car with a human driver looking at the road, would have been no different.

Like you, I don;t believe the automated braking system should have been disabled. It may have given the pedestrian a better chance as the car may have slowed - even slightly - and that may have reduced the severity of the accident.

That is IF the human driver was not paying attention. However, if the driver was paying attention, it would have been a different outcome.

[photoptimist]

So now autonomous vehicles are going to use turn signals to coordinate their movements. I think this is a bit more complicated than what you imagine. Just what signals would be used for an emergency stop, or coming up to a stop sign, or a normal right/left turn etc. They could use a flashing code sort of like when you change the channel on your TV (if you have one). I think that now is the time for all of these diverse companies to come together and build an equivalent to the IATA in order to share/develop protocols that allow for these vehicles to communicate with each other. Heaven help you if the emitter/sensor gets dirty (a la #36) or your particular vehicle is the only one that uses EBCDIC rather than Unicode-ASCII ;-)

Why complicate it? Human drivers get along fine without complex flashing codes.

Autonomous vehicles can (and should) use the exact same signals that human drivers use: a blinking left red tail light means turning left or shifting one lane to the left; a blinking right red tail light means turning right or shifting one lane to the right; solid red tail lights means braking for whatever reason (human drivers don't have special brake lights for different breaking situations), blinking hazard lights means driving slowly due to some problem.

There's no need for any special code and the standards for these signals are already industry wide.

[MarkJerling]

Why complicate it? Human drivers get along fine without complex flashing codes.

Autonomous vehicles can (and should) use the exact same signals that human drivers use: a blinking left red tail light means turning left or shifting one lane to the left; a blinking right red tail light means turning right or shifting one lane to the right; solid red tail lights means braking for whatever reason (human drivers don't have special brake lights for different breaking situations), blinking hazard lights means driving slowly due to some problem.

There's no need for any special code and the standards for these signals are already industry wide.

No autonomous vehicle will use "flashing codes" for information. Whether they can or will read flashing indicator lights I don't know. With most people using their indicators erratically anyway, I'm not sure it would be wise to consider a system that works that way. So far as I'm aware, hazard lights are only to be used when a vehicle is stationary, so the "driving slowly" use is not something likely to be programmed into any automated system.

Autonomous vehicles don't need brake lights to know that a car is slowing or has stopped. Their radar and/or lidar systems know the relative speeds and locations of all objects around the car. The "learning" difficulty is to know - always - what to do with that information.

[Racer X 69]

How they keep those sensors clean I don't know!

Tiny little wipers.