Driverless Cars

[Jstpssng]

You were doing so good, without the last sentence.

You beat me to it. :thumbsup: I'm not sure what people are supposed to do for a living, as more and more jobs are replaced by automation.

 

[LD48750]

You beat me to it. :thumbsup: I'm not sure what people are supposed to do for a living, as more and more jobs are replaced by automation.

Program & repair robots.

 

[Raspy]

Woke up today to freezing rain/ hail. The roads were white so there's no way to see any markings. I dread the day of self driving cars being the norm as I can't see how they will ever be able to drive in these conditions. GPS would need to take a major step forward in accuracy.

When you are driving, if all you ever base your position on is the obscured lines on the road, you are dangerous. It means you cannot tell where the road is either. But don't you look at the traffic ahead, the traffic coming the other way, the general center of the road, the speed limit, snow markers or other objects along the side of the road, or in your path, etc. I do. Just because you, yourself can't imagine driverless cars, based on incomplete knowledge of how they work, does that mean they are actually dangerous? Or does it mean there may be more to the story than you realize? I suggest you may have an irrational fear based on incomplete knowledge and not based on facts or statistics. I'll admit I would be nervous in the back seat of a driverless car, but I'd likely be terrified in the back seat of someone who had never driven if the snow before and didn't understand the increased stopping distance, reduced cornering traction, etc.

 

[buckeyefarmer]

Program & repair robots.

Yes, lots of work there. My company cant get enough engineers and programmers.

 

[aczlan]

When you are driving, if all you ever base your position on is the obscured lines on the road, you are dangerous. It means you cannot tell where the road is either. But don't you look at the traffic ahead, the traffic coming the other way, the general center of the road, the speed limit, snow markers or other objects along the side of the road, or in your path, etc. I do. Just because you, yourself can't imagine driverless cars, based on incomplete knowledge of how they work, does that mean they are actually dangerous? Or does it mean there may be more to the story than you realize? I suggest you may have an irrational fear based on incomplete knowledge and not based on facts or statistics. I'll admit I would be nervous in the back seat of a driverless car, but I'd likely be terrified in the back seat of someone who had never driven if the snow before and didn't understand the increased stopping distance, reduced cornering traction, etc.

Given that a self-driving car has run someone into the back of a fire truck and the end of a concrete barrier due to bad lane markings and a fire truck stopped in the lane of traffic, I would say that that is a valid concern for self-driving cars.

Aaron Z

 

[3930dave]

I suggest you may have an irrational fear based on incomplete knowledge and not based on facts or statistics. I'll admit I would be nervous in the back seat of a driverless car, but I'd likely be terrified in the back seat of someone who had never driven if the snow before and didn't understand the increased stopping distance, reduced cornering traction, etc.

Ignorance can breed Fear. It can also breed Blind Faith. Both can lead to sub-optimal outcomes.

Many rational (recognizing that there are Irrational ones on the road....) human drivers can recognize when the road conditions are dangerous, and adjust their driving accordingly - pulling over and stopping in a safe location if needed. Tech can be programmed to do that too, and perhaps one day it will get there, safely, but I don't believe (given the lethal accidents listed) we are there yet.... the coders are still busy solving higher-priority problems.

Rgds, D.

 

[cqaigy2]

I might be going out on a limb here, but i think pretty much nobody, is ready for the self driving vehicle on public roads at it's current tech level , but i also think that it will be here sooner then most of us think it should be.

When vehicles first started out on public byways, their technology level was quit low and wouldn't be allowed on the roads as a new model vehicle. A number of deaths had to happen for some, if not most, of the tech changes for safety to occur, and i think it will be the same with self driving cars.

I also think that driverless cars will reduce fatalities.

I believe most of us think we are much better drivers than we really are and haven't had a problem because other people on the road made adjustments that kept us out of trouble. I think the skill level of drivers, at least in the US, is getting worse and add in-cabin distractions to the mix, that sell vehicle, isn't helping.

And as 3930dave and others have said, thread drift can be interesting, but hate to see this thread moved to the "Friendly Politics".

 

[crazyal]

When you are driving, if all you ever base your position on is the obscured lines on the road, you are dangerous. It means you cannot tell where the road is either. But don't you look at the traffic ahead, the traffic coming the other way, the general center of the road, the speed limit, snow markers or other objects along the side of the road, or in your path, etc. I do. Just because you, yourself can't imagine driverless cars, based on incomplete knowledge of how they work, does that mean they are actually dangerous? Or does it mean there may be more to the story than you realize? I suggest you may have an irrational fear based on incomplete knowledge and not based on facts or statistics. I'll admit I would be nervous in the back seat of a driverless car, but I'd likely be terrified in the back seat of someone who had never driven if the snow before and didn't understand the increased stopping distance, reduced cornering traction, etc.

I understand exactly how they work. Like humans they gather all the information they can to make the best decision. It starts with GPS. GPS is only accurate down to about 10 feet. Things like other cars and signs on the side of the road only come into play when they are present. In fact driverless cars depend on cars with drivers. If the software can't figure out where it is close enough to stay on the road and drives into a ditch the next driverless car is only going to be as good as it's programing to determine if that car is parked and which side of the car it should be on. The 10 foot accuracy is best case. The current generation of GPS just gets you close to where it thinks you are and then uses the map stored in it's memory and assumes you are on the road.

What this means is driverless cars need other fixed points of reference. Right now that's the lines on the road. Some day we may have cars like Google uses for street view imaging that will log every fixed point (buildings, trees, etc.) so the software will have more data. However that's a massive amount of data which would mean wireless connectivity to download it on the fly. In fact driverless cars of the future will also upload what it's seeing compared to what's already been stored so any changes can be made real time (A tree gets cut down, a new house is built, etc.) Before that happens there needs to be a lot of improvements as well as how do we pay for this network. Do you pay a monthly fee to a company like Verizon? Does the government own it and maintain it? Does the government pay to install sensors under the pavement in the middle of the road that the computer can 'see'? And how would these sensors be powered?

In some places it'll be easy. My point is that there are parts of this country that would require some sort of investment before it's possible. Around here there's roads that are basically the same height as the fields on either side. When the corn is standing high it's real easy to know where the road is. In the middle of winter with blowing snow (or an ice storm) I can't see a computer having the information needed to do the job.

I'm not saying that driverless cars will not happen, I'm saying we are much closer to a system that works 75% to 85% of the time and requires the human to do the rest. It's that 15% to 25% that's going to be the issue. We can see with Tesla's autopilot just how stupid people are (climbing into the back seat so they can get a picture of the car driving itself or not stopping the car from plowing into a fire truck). The less they drive the worse they will be. I'm sure there was the same problem when cars started taking over for horses.

 

[Reg]

Woke up today to freezing rain/ hail. The roads were white so there's no way to see any markings. I dread the day of self driving cars being the norm as I can't see how they will ever be able to drive in these conditions. GPS would need to take a major step forward in accuracy. What that's going to mean is a bunch of people who rarely drive suddenly trying to in the worst conditions. Can't see how that would end in failure. Reminds me of something I read where the younger generations can't read the hands of a clock. All they've seen is digital displays so they had no need.

Yeah, even in my school days kids were no longer taught how to sharpen quill pens, such lost opportunity to teach anachronistic skills.

Not a MAJOR step in GPS accuracy or precision (I'll assume you know the difference).
Centimeter accuracy is already IN USE !
There is a LOT of automation in farm tractors - I almost forgot which forum I am in, you can look it up.

We (our vehicles) won't need centimeter accurate maps of EVERYWHERE preloaded, the next hour or two of the planned route will be more than adequate and that isn't a LOT of data - in a couple of years it will be viewed as even less.
Some huge percentage of our travel is local, it is VERY local, i.e. MOST of the data needed for very precise navigation could/would be cached anyway.

I haven't driven to California (from New England) for almost eleven years, my map base for AZ, CA, NM etc wouldn't need to be refreshed until I go there again, perhaps never.
Despite that I get road updates for ALL STATES on GPS every time I update it.
An "on demand", or "as needed" update for the planned route would be so much more efficient.

I'm optimistic on this, perhaps more so than most.
What is CRAZY is ~2 ton vehicles hurtling toward each other at closing speeds of ~150 MPH with totally independent control.
Just a few "rules" that their operators are supposed to follow.

 

[Raspy]

Given that a self-driving car has run someone into the back of a fire truck and the end of a concrete barrier due to bad lane markings and a fire truck stopped in the lane of traffic, I would say that that is a valid concern for self-driving cars.

Aaron Z

It's easy to be concerned about new technology and new things can have bugs. But if you want to make an argument against something, at least be accurate with your points.

The Tesla is NOT a self driving car. It is a semi-autonomous car. Tesla makes it clear that you must keep your hands on the wheel and stay alert. If you don't, the car will warn you and will eventually pull over and stop if you don't pay attention and keep driving it yourself.

The "fire truck" accident was apparently not a result of not seeing lane markers. It was more a glitch in the car deciding what was a threat and what was not after a car in front of it moved to another lane. The system has to evaluate what is just a roadside stationary thing and a moving car in front of it. This is why you have to be aware and monitor it.

It was a disturbing incident and a situation that has to be refined out, but Teslas are not self driving cars and lane markers may have not been an issue at all.

 

[Gale Hawkins]

It's easy to be concerned about new technology and new things can have bugs. But if you want to make an argument against something, at least be accurate with your points.

The Tesla is NOT a self driving car. It is a semi-autonomous car. Tesla makes it clear that you must keep your hands on the wheel and stay alert. If you don't, the car will warn you and will eventually pull over and stop if you don't pay attention and keep driving it yourself.

The "fire truck" accident was apparently not a result of not seeing lane markers. It was more a glitch in the car deciding what was a threat and what was not after a car in front of it moved to another lane. The system has to evaluate what is just a roadside stationary thing and a moving car in front of it. This is why you have to be aware and monitor it.

It was a disturbing incident and a situation that has to be refined out, but Teslas are not self driving cars and lane markers may have not been an issue at all.

Good points. I expect we are years away from seeing the first self driving cars on the roads. One of my doctors was excited last week about her 2015 Subaru Forester applying the brakes when a car suddenly slowed to make a right turn in a 35 MPH speed zone.

 

[Reg]

It is a PROGRESSION from the driver assist "components" that we now have in Teslas, Mercedes, BMWs, etc., that are already coming down market to Tonka, Nissan, GM, Ford - - lane marker following, cruise control back off if too close, emergency braking, etc.

From those "components" to incorporation into "systems" that include GPS, real time map updates, construction, accident, congestion, even potholes of the day along the planned route.
I don't think this is an unmanageable amount of data for journeys along a planned route and routes can be re-planned en route.

I think all the enabling technologies are already THERE, public acceptance is lagging, regulation is lagging even more.

 

[Gale Hawkins]

Right on Reg. Thanks

 

[3930dave]

It's easy to be concerned about new technology and new things can have bugs. But if you want to make an argument against something, at least be accurate with your points.

The Tesla is NOT a self driving car. It is a semi-autonomous car. Tesla makes it clear that you must keep your hands on the wheel and stay alert. If you don't, the car will warn you and will eventually pull over and stop if you don't pay attention and keep driving it yourself.

The "fire truck" accident was apparently not a result of not seeing lane markers. It was more a glitch in the car deciding what was a threat and what was not after a car in front of it moved to another lane. The system has to evaluate what is just a roadside stationary thing and a moving car in front of it. This is why you have to be aware and monitor it.

It was a disturbing incident and a situation that has to be refined out, but Teslas are not self driving cars and lane markers may have not been an issue at all.

Not piling on Aaron..... but yeah, that's part of my point/concern..... Be an Informed Consumer...... usually a good practice, but esp. when complexity scales to where these products are going.

Human nature..... do most people read some (any?) of a conventional vehicle's Owner's Manual ? Couple that tendency with Elon's force of nature (over)marketing...... yeah, people are going to incorrectly assume that vehicles are All Seeing/Knowing/Driving.....

..... When, as you correctly point out Raspy, they shouldn't....

That fire-truck scenario has been duplicated on a closed-course track..... Same result. I think I've posted that test in here.... British video.... Top Gear ? or maybe just a Brit Eng firm, or govt dept.

Rgds, D.

 

[3930dave]

I think all the enabling technologies are already THERE, public acceptance is lagging, regulation is lagging even more.

As challenging as the tech can be to get right, even just most of the time, some of those top-level macro issues are tougher....

Liability is a big one. If I'm ever going to "own" a self-driving car (fully autonomous), Why would I need personal insurance ?

My personal views/preferences aside, THAT ^ issue is the acid test for me...... re....... "Are We Really There Yet ?".

Trust the Tech, Trust the Tech, Trust the Tech........ question otherwise, and get brushed aside as a Luddite :laughing:

Where does the accountability lie, When Things Go Wrong ?

Automation, Transport....... some things can be learned from the aviation world......

Plowhog recently posted an interesting ILS accident.....

https://www.tractorbynet.com/forums/front-porch/404639-commercial-airline-accidents-8.html#post5304436

..... ^ described in Post #72.

Guess who got blamed, when the autonomous tech put them (literally) into the weeds.....

It will be some time before driverless cars have all the validation hours locked down, that those ILS systems do.

Rgds, D.

 

[Reg]

@3930dave,
Holding forth a pathological case as a "concern" is OK, but it isn't representative.
It is being addressed (as are very large bobbing animals that were NOT anticipated in early trials, but showed up anyway in Australia (of all place - who'd a thunk it ?)).

The GOOD news is that once solved the fire truck problem will have been solved for THAT case and many cases that have same/similar attributes - and of course it will be solved for every vehicle on every journey once the "patch" is installed.

For human drivers the story is worse, MUCH worse.
Every student driver has to anticipate anything/everything that might happen - and we have well over a hundred years of data to support the statement that they don't (can't ?).
SO MANY near misses form the experience base of those of us who have (so far) got away with it.
Some of my accidents could have turned out very much worse, but for another factor or two - I'll never know, but it is easy to imagine the same accident (with just one more factor) resulting in a fatality.

 

[sd455dan]

Thought some readers of this thread may find this interesting .

Autonomous Driving Future - YouTube

Zak and Jesse are big Tesla supporters - for sure,

Zak has done an interesting series on

(Our Autonomous driving Future)

It is a series of ten Videos ( 5-13 minutes each) - some interesting statistics, info and perspectives in them.

Some of the ideas- future projections/ conclusions make sense-

some not so much for Rural dwellers as always YMMV

 

[Reg]

OK, so a handful of "golly gee technology is a killer" cases - gottit.
Against that - I haven't yet seen the results of last night's human drivers' death toll while intoxicated, distracted, unable to handle prevailing conditions (visibility, ice, etc.)
I'd take bets that it is many times more than the semi autonomous vehicle incidents though.
How many thousands (millions ?) do we have to have before we acknowledge that the existing system of human (not)controlled vehicles is so much worse ?

 

[3930dave]

Thought some readers of this thread may find this interesting .

Autonomous Driving Future - YouTube

Zak and Jesse are big Tesla supporters - for sure,

Zak has done an interesting series on

(Our Autonomous driving Future)

It is a series of ten Videos ( 5-13 minutes each) - some interesting statistics, info and perspectives in them.

Some of the ideas- future projections/ conclusions make sense-

some not so much for Rural dwellers as always YMMV

Thanks for posting that, I will go through them.

Rgds, D.

 

[3930dave]

How many thousands (millions ?) do we have to have before we acknowledge that the existing system of human (not)controlled vehicles is so much worse ?

I'm not big on treating a symptom, while ignoring root cause.

No question, many parts of North America need better drivers..... starts with lack of training and meaningful standards and enforcement. As has been stated on here, you don't need much more than a pulse, to get a driver's license in some jurisdictions. Why was that allowed ?

If we are going to remove dangerous items from individual control, why stop at cars ?

It will be interesting to see how this plays out...... I'm betting that insurance will be a key factor.

Rgds, D.