Self Driving Cars?

[Amethyst]

True - I have two relatives with amblyopia ("lazy eye" - one eye has only peripheral vision) and they passed their driving tests at the usual age.

More - In my mid-20's, I lost most of my sense of smell (and taste) due to a virus which damaged my olfactory nerve. (At the time, I thought I just had a bad cold). The neurologist who diagnosed my trouble way back when, assured me that my brain would adapt to my new level of sensing, and he was right. To this day, I have only about 30% normal smell and taste; cannot detect ambient odors unless they are really strong; but if you blindfold me and give me a plate of food, I can tell a) what each food is and b) if it was made with good ingredients and was cooked properly.

I mentioned stereoscopic computer vision earlier and the difficulty in using it. However, I forgot to bring up something else. That is, a human can learn to do well with just one eye.

Yes. Think about it. People who are blind in one eye do not stumble all over the place.

So, how do they do it? Part of it is from memory when they navigate familiar surroundings. Part of it is from cognitive reasoning. They know from experience how high a step is, how wide and long an object is, etc... Part of it is that the brain and its memory can determine the distance to objects when the line of vision changes. It's the same as having multiple eyes, but the brain combines the vision at one instant to that a second later but at a different viewpoint when one moves. That's how one can get the depth perception with one eye.

Similar to the combining of vision from different viewpoints is the SAR (Synthetic Aperture Radar). An aircraft looking down with a radar while flying can combine the images later to have an extremely fine view of the ground. It takes special equipment for this.

Anyway, while we are in awe of the digital computer, we should not forget to pat ourselves on the back for the awesome cognitive power of our brain.

 

[NW-Bound]

I've not read through every single post here but with all the concern for safety and accidents one would think that the current intelligence behind the wheel (us) was foolproof!

Not fool proof. But at 500K miles per accident, an autonomously-driving computer has to do better than that to claim an improvement. It is tougher than most people imagine. Google's cars currently need human assistance once every 5,000 miles.

 

[Midpack]

I've not read through every single post here but with all the concern for safety and accidents one would think that the current intelligence behind the wheel (us) was foolproof!

No, that's not it at all. We posted some stats on current drivers. As I posted recently (I'll just repeat here, this is a long thread)...

Since we are getting better and better driver assistance features on cars, the relevant question is not whether a self-driving/autonomous car will have fewer accidents per mile than a current car with an average driver. The relevant question is whether a self-driving/autonomous car will have fewer accidents per mile than a car with driver assistance features, and especially one that keeps the driver involved (which I don't think we have yet - monitoring the head/eye movements as I mentioned before).

It's not that black and white either. Many driver assistance features are being developed as precursors to fully autonomous cars - there not completely unrelated. The evolution of self-driving cars has been (for now) defined as a series of steps. There's level 0 (total manual) and level 5 (fully autonomous), and the progression from level 1-4 arguably falls into the driving assistance category. Each step will have to improve safety, at reasonable cost, to move on - I can't imagine people will pay a premium for more convenience and/or safety, if they believe they're actually less safe?

I believe we'll reach widespread adoption of level 5 and significantly reduce accidents, within 50 years if I had to guess. Not trying to put words in anyone's mouth, but it appears you believe, level 3 or 4 may prove to be the safest attainable condition. I can't discount that outcome, and it IS definitely what I expect in the decade ahead.

Re-posted

-ERD50[/QUOTE]

 

[NW-Bound]

I have been thinking about the dual Lidars on top of the Cruise Automation (GM now) car. Why have two if one is already expensive? A single lidar can determine distances so it is not for stereoscopic view.

I recall this car passing a parked delivery truck in a narrow street. Could it be so that the car can see better around another car, like in this situation? Google's car single lidar cannot see past the edge of that parked truck until the center of the car is in the opposite lane.

If that is the case, one can see that the human driver has a very simple solution that costs nothing: he sticks his head out of the window.

 

[NW-Bound]

As an aside, does anyone have links to any basic explanations of how this kind of software is developed?

It's obviously not procedural based, you just can't write enough lines of codes that cover all the possibilities:
...

Their not AI based, but it does appear to us as an intelligence of sorts.

-ERD50

I think the processing of lidar point-cloud data, grouping the dots into "solid" objects can be done with non-AI software. Same with keeping track of these "objects", and projecting their trajectories. Same with making sure your planned trajectory does not intercept them.

Some kind of AI will be needed to interpret policeman's and flagman's gestures, etc... AI is a very broad term. Speech recognition is considered AI. Same as reading street signs, traffic signs, etc...

 

[ERD50]

It's not that black and white either. Many driver assistance features are being developed as precursors to fully autonomous cars - there not completely unrelated. The evolution of self-driving cars has been (for now) defined as a series of steps. There's level 0 (total manual) and level 5 (fully autonomous), and the progression from level 1-4 arguably falls into the driving assistance category. Each step will have to improve safety, at reasonable cost, to move on - I can't imagine people will pay a premium for more convenience and/or safety, if they believe they're actually less safe?

...

That chart seems to skip over what I'm proposing - driver assistance along with driver involvement (detecting a distracted driver, plus actively keeping the driver involved in the activity).

I recall my son saying that in drivers ed, they were taught to keep a mental 'running commentary' as they drove. I try to that (saying to myself - 'now there's a red car behind me, keep track of him moving to the right or left lane, that white car is coming up on my right blind spot, there's a car pulling out on the right ahead....' ). If you do that, you just are so much more aware of everything and are sort of thinking ahead on what to do if one of those objects you are mentally tracking changes from what you visualize as its planned path.

I think driver assistance can work with us in that way, keeping us more involved, and seeing what we might miss. For example, if my blind spot warning flashes, the system could be looking to see if I turned my head or redirected my eyes to notice it. If I didn't respond, it could get more aggressive, eventually warning and slowing the car to a stop. That should also be effective against drunk or otherwise impaired drivers

... I believe we'll reach widespread adoption of level 5 and significantly reduce accidents, within 50 years if I had to guess. Not trying to put words in anyone's mouth, but it appears you believe, level 3 or 4 may prove to be the safest attainable condition. I can't discount that outcome, and it IS definitely what I expect in the decade ahead. ...

All bets are off if you are talking 50 years out! Geez, in 1967, we didn't even have portable calculators!

History of Electronic Calculators

bold mine

1969 -- Impressed with the "Cal-Tech" prototype and Texas Instruments' IC production capability, Canon (Japan) begins work with TI on the electronics for a small, hand-holdable calculator which would be called the "Pocketronic."


1970 -- The first battery-operated "hand-held" calculators are sold. Most are too large to actually be considered "pocket calculators," but they are far smaller than anything seen before. In mid-1970, Sharp begins to sell the QT-8B which, by using rechargeable batteries, is a portable version of their desk-top QT-8. Canon's "Pocketronic" sales begin in the Fall of 1970 in Japan and February 1971 in the USA. Canon used Texas Instruments' ICs and thermal printer. Selling for just under $400, the "Pocketronic" was a four function, hand-held, printing calculator, with the only display being the printed tape running out of the side of the machine.

So I don't doubt the technology will be sufficiently advanced after another 50 years, but then I'll also stick to my previous claim - go out that far, and 'self driving cars' will be a 'solution looking for a problem'. The idea of cars and driving as we know them today probably won't exist.

There's a slim chance we may live long enough to find out, or at least be close enough to that time to have a clearer picture! Probably won't still be posting here though Even advanced computers won't be able to keep track of my post count by then!

-ERD50

 

[NW-Bound]

Maybe REWahoo's asteroid will strike and put us out of misery soon?

 

[dixonge]

Google cars are 1,700 times better. But is it good enough? No. If a driver does not catch the computer screw up in time, he will have an accident and may not live if it happens on a highway. I do not want to have a risk of dying every 5,000 miles.

The point again and again is that I do not hold my breath to wait for a self-driving car. I am not excited about this. When it happens, I will know.

Given that Waymo cars never go faster than 35mph, I think that the driver is relatively safe from a life-ending screw-up. I can't even begin to imagine commuting for 5000 miles before I ever have to touch the steering wheel once.

 

[ERD50]

Given that Waymo cars never go faster than 35mph, I think that the driver is relatively safe from a life-ending screw-up. I can't even begin to imagine commuting for 5000 miles before I ever have to touch the steering wheel once.

But the average for a driver is something like 12,000 miles/year. Now, we don't know what those 'interactions' involved, but if even 1/4 of them might have ended in an accident, that's an accident every 20,000 miles, or more than one every two years. So it's got a long way to go. And most of us want/need to drive faster than 35 mph, so that's a factor.

-ERD50

 

[explanade]

Article on 3D maps being developed for SDCs.

https://arstechnica.com/cars/2017/0...aps-of-the-world-will-be-for-cars-not-humans/

 

[Fedup]

Would one buy a Level 5 car when it becomes available, even if it costs $150-200K? Many posters here would when it gives them mobility that they would not otherwise, such as allowing them to do to a doctor or the hospital.

It is within their means, and when they become so invalid, they will realize that they cannot "take it with them". Besides, such a valuable car is an asset that can be passed down as inheritance, so it is not really wasted money.

Our city has small bus for people who can't drive. $5 per ride, door to door service.
I wouldn't pay $200k. Not worth it.

 

[gerntz]

I recently bought a new car, and was able to convince my old LBYM self to go several trim levels up to get some of the driver assistance features. I don't really drive much, so I'm still getting accustom to these features. But so far:

I absolutely love the blind spot detection. I'm not going to rely on it, but it's nice to have a system helping you out, providing a little more info. It's more like having a passenger with you, helping you out. I'm sort of mentally thinking, when I see the indicator "Yep, I'm aware of that car there, thank you.".

lane departure alarm: It's interesting. I don't think it has ever warned of me something I wasn't really aware of. Sometimes I will intentionally drift a bit, like to avoid something on the road, or maybe cut a corner in a bit more of a straight line, and it will let me know. But it does keep me a little more involved I think, like "hey, I'm watching you, so pay attention!".

Front Collision Detection: This one doesn't actually apply the brakes, it gives a green car symbol on the dash when a car ahead is in it's view, it turns yellow if you approach it too fast, then will beep and flash red light on the window if you get really close. I set it to the most sensitive setting, and was able to get it to flash a couple times when I intentionally 'pushed the envelope' to test it.

I like them all, the added info seems to keep me more engaged, rather than kicking back to let them do their thing (though of course, they aren't anywhere near that level anyhow).

Don't have adaptive cruise control, so can't comment.

And also, rear camera, with side alerts. I love this. I can see so much more when I back up than in any car I've ever owned. The side warning does a good job, almost too good, but that's OK. It will alert me to someone pushing a cart in the parking lot - they are moving slow, so I can back slowly w/o worry, but it's warning me. I kid my DW, that it takes me too long to back out of a parking space now - in the past, I'd move slow in my old car, and people would get out of the way - now I feel guilty backing up if this is beeping at me!

Front and Rear object detection - helps in parking a bit, you know how close you are coming to something. Seems pretty good.

These features are advancing pretty fast, I'll bet that I don't keep this car as long as I have kept cars in the past, I'm going to want the latest features, and won't wait 11 years for them (actually went 16 years in my old Volvo, which was safe for the time in terms of the air-bag/restraint systems).

-ERD50

Have most of these. My MB C300 approaching 3 years old.

- BSD - sure I look, but you can't help from relying on it since it goes off at times you don't expect.

- for lane departure, steering wheel thumps unless crossing center yellow or going off left lane at road side, then takes over steering quickly back to my lane on own. It's so quick & sharp that it makes it look like a lost control.

- I get a beep versus a light for approaching collision. That's better since I can't see but can hear.

- Love ACC. Really great in stop/slow/go traffic vs. repeated foot braking & acceleration.

- Has rear camera, but not sure I have side warning or rear object detection.

- Doesn't have self-parallel parking. Not that I do it much, but sounds/looks excellent.

Now if it could only avoid potholes. Blew out both front tires within 3 weeks this winter on freeways. Recently hit another but not as hard.

 

[dixonge]

But the average for a driver is something like 12,000 miles/year. Now, we don't know what those 'interactions' involved, but if even 1/4 of them might have ended in an accident, that's an accident every 20,000 miles, or more than one every two years. So it's got a long way to go. And most of us want/need to drive faster than 35 mph, so that's a factor.

-ERD50

You are right in that there are still many obstacles to overcome to achieve full 100% autonomy. However, 5,000 miles in-between human interventions is exactly 5,000 more miles than my current car, which requires human intervention for every single inch. In those terms, it's an astounding achievement level. In some senses, it is, for all practical terms, already fully autonomous! By that, I mean that the current state of the technology is already so far ahead of 99.5% of all cars on the road that it will *seem* as if it is already autonomous.

If your frame of mind is "well, it's not perfect yet, so I'm not interested" then yes, you're going to be waiting a long time. But if your frame of mind is "Yes, I would *love* to be a very aware passenger who occasionally has to take over and override the robot car" then that level of driving can be purchased in a few places right now.

I think my own "Aha!" moment with all of this came about a few years ago in Ajijic, a sleepy little hamlet on the shores of Lake Chapala, central western Mexico. We were housesitting for friends and driving their Subaru SUV. It had the rear-view camera and auto-cruise control. Seeing and feeling it automatically braking then starting back up again was pretty amazing. And this is pretty low-tech compared to Tesla, Waymo, etc.

 

[Midpack]

Given that Waymo cars never go faster than 35mph, I think that the driver is relatively safe from a life-ending screw-up. I can't even begin to imagine commuting for 5000 miles before I ever have to touch the steering wheel once.

The Waymo you're probably referring to they call the Firefly and it only goes 25 mph. Waymo has other cars, before and after Firefly, that go at highway speeds every day.

 

[Midpack]

I may be the only autonomous car geek here willing to sit through these long videos, but I found them very interesting and informative. They address most if not all of the issues we've discussed, and more. Some here may be very surprised at how much they've learned already. If I had to pick just one video, it would be the first at SXSW.

https://youtu.be/Uj-rK8V-rik

https://youtu.be/xRGIlwgc--g

And I modify my guess for widespread level 5 adoption to 30 years or less.

 

[ERD50]

dixonge - you are misunderstanding me. Let me break it down.

You are right in that there are still many obstacles to overcome to achieve full 100% autonomy....

Agreed

However, 5,000 miles in-between human interventions is exactly 5,000 more miles than my current car, which requires human intervention for every single inch. In those terms, it's an astounding achievement level.

Yes, it is an outstanding achievement.

In some senses, it is, for all practical terms, already fully autonomous! By that, I mean that the current state of the technology is already so far ahead of 99.5% of all cars on the road that it will *seem* as if it is already autonomous.

Sure, it is far ahead of other cars - but so?

If your frame of mind is "well, it's not perfect yet, so I'm not interested" ...

Not at all. My viewpoint is we should push whatever will save the most lives, in the shortest possible time, and that can be done in parallel with full autonomy (which we may not even need).

I believe that using this tech to enhance the driver is the best route for now. Let them progress, and we will see if the driver isn't needed.

But if your frame of mind is "Yes, I would *love* to be a very aware passenger who occasionally has to take over and override the robot car" then that level of driving can be purchased in a few places right now.

But the problem is, it seems these cars are allowing the driver to become un-involved. If you aren't engaged, no way can you respond if the car suddenly alerts you to 'take over and override the robot car'. So you drop your laptop, look up, and before you can even orient yourself - bammm you crashed.

It really seems straightforward to me - keep the driver engaged, work with the driver, and as the tech advances, we will have data that says - look, we kept the driver engaged, and we only turn things over to them so rarely, that it just doesn't matter if the driver is there or not.

Ahhh, but by what measure will that be? Some people say you can't put a price on human life. So will there be push-back if there is a single fatality a year, and that fatality would have been avoided if the driver was kept engaged? As I told Midpack, the goalposts keep moving. It isn't autonomous car versus average driver - it's autonomous car versus autonomous tech plus engaged driver. It just makes sense that adding a driver to the mix will help, and the tech will need to be super-super-advanced before the driver is nearly totally useless (assuming they are kept engaged).

I think my own "Aha!" moment with all of this came about a few years ago in Ajijic, a sleepy little hamlet on the shores of Lake Chapala, central western Mexico. We were housesitting for friends and driving their Subaru SUV. It had the rear-view camera and auto-cruise control. Seeing and feeling it automatically braking then starting back up again was pretty amazing. And this is pretty low-tech compared to Tesla, Waymo, etc.

Yes, impressive. But that is not even child's play compared to dealing with the various scenarios that arise on the road over the course of years that we experience between accidents. And with every X% improvement, the remaining X% gets tougher and tougher to handle.

And I think it was mentioned way back, if the car is driving 99.9% of the time, us drivers are going to be sooooo out of practice. And then we are supposed to suddenly take over under the toughest situations, on a moments notice? Good luck with that.

-ERD50

 

[Amethyst]

True; but in many of our discussions, the perspective has been "I want to stay mobile even after losing some of my marbles due to age." So that would be a case of someone who is not capable of taking over for the robot car.

s.

If your frame of mind is "well, it's not perfect yet, so I'm not interested" then yes, you're going to be waiting a long time. But if your frame of mind is "Yes, I would *love* to be a very aware passenger who occasionally has to take over and override the robot car" then that level of driving can be purchased in a few places right now.

.

 

[growing_older]

the goalposts keep moving. It isn't autonomous car versus average driver - it's autonomous car versus autonomous tech plus engaged driver. It just makes sense that adding a driver to the mix will help, and the tech will need to be super-super-advanced before the driver is nearly totally useless (assuming they are kept engaged).

But we do not live in a world like that. Even without tech, drivers are NOT kept engaged. There are hundreds of accidents daily from un-engaged drivers. Why do you insist that automation must be better than some theoretical automation plus some theoretical perfect driver, when we already know that drivers do not and cannot be counted on to attain that level of engagement. They won't do it now with ZERO automation.

And then we can start discussing how engaged drivers can make mistakes that the automation might have avoided.

 

[ERD50]

But we do not live in a world like that. Even without tech, drivers are NOT kept engaged. There are hundreds of accidents daily from un-engaged drivers. Why do you insist that automation must be better than some theoretical automation plus some theoretical perfect driver, when we already know that drivers do not and cannot be counted on to attain that level of engagement. They won't do it now with ZERO automation.

And then we can start discussing how engaged drivers can make mistakes that the automation might have avoided.

You missed my point. I'm saying that the tech should keep the driver engaged. It's certainly do-able, and far easier to do than autonomous.

I'm betting it isn't getting the press, because it isn't as 'sexy' as a self-driving car. What, you want me to pay attention!?

-ERD50

 

[ERD50]

Probably deserves it's own thread, but for now, here's one document addressing driver engagement:

http://cvrr.ucsd.edu/ctran/files/BookChapterVOW.pdf

An important real-life application domain of computer vision techniques looking at people is in developing Intelligent Driver Assistance Systems (IDAS’s). By analyzing information from both looking in and looking out of the vehicle, such systems can actively prevent vehicular accidents, improve driver safety as well as driver experience.

....
A 2006 study sponsored by the US Department of Transportation’s National Highway Traffic Safety Administration
concluded that driver inattention contributes to nearly 80 percent of crashes and 65 percent of near crashes.

and specific to driver fatigue:

Clemson Vehicular Electronics Laboratory: Driver Alertness Monitoring

-ERD50

 

[Midpack]

But we do not live in a world like that. Even without tech, drivers are NOT kept engaged. There are hundreds of accidents daily from un-engaged drivers. Why do you insist that automation must be better than some theoretical automation plus some theoretical perfect driver, when we already know that drivers do not and cannot be counted on to attain that level of engagement. They won't do it now with ZERO automation.

And then we can start discussing how engaged drivers can make mistakes that the automation might have avoided.

You missed my point. I'm saying that the tech should keep the driver engaged. It's certainly do-able, and far easier to do than autonomous.

I'm betting it isn't getting the press, because it isn't as 'sexy' as a self-driving car. What, you want me to pay attention!?

-ERD50

I'd say you missed his/her point. 90% of accidents are attributed to human error now, and he noted drivers aren't paying attention now, mobile devices have made it worse. More driving assistance features won't make them more attentive, odds are less so.

The technologies making the most progress are shooting for level 4-5 automous cars like Tesla, Ford, Lyft, Uber, Volvo, Waymo, Mothers Against Drunk Driving, National Federation of the Blind, United Spinal Association, the R Street Institute and Mobility 4 All. Level 5 is the most robust solution, lots of industry people believe it can be done.

Who are the groups making progress trying to stop at add driving assistance and engaging drivers more?

 

[meierlde]

I finally found a chart from Gartner on autonomous vehicles. Gartner divides life cycles of tech into several phases Innovation Trigger, Peak of Inflated Expectations, Trough of disillusionment, slope of enlightenment, and Plateau of Productivity. It shows autonomous vehicles at the top of the Peak of Inflated Expectations, about to fall into the trough of disillusionment, with a 10 year plus development time: https://www.forbes.com/sites/louisc...machine-learning-for-first-time/#406c31cf3f82
Gartner is a service a lot of IT departments use to evaluate technologies and decide when and if to use them.

 

[Texas Proud]

This is going fast... and I just read what was posted today... WHEW.....


But, I have read about one accident per 500,000 miles (or 100,000)... and I would like to question this... maybe I will do a poll...

I have never had a car get to 100,000 miles without it being in an accident... sometimes 2 or 3.... some are very minor, but still an accident... (note, very few are my fault)....

So how many people have had a car go 100,000 without being in any kind of accident?

 

[Ronstar]

I've never had a vehicle get to 100,000 miles without an accident. Most never got to 50k before I hit a deer. I may be wrong, but I don't think it's possible for a self driving car to miss a deer crossing a snowy road in the dark.

 

[gerntz]

So how many people have had a car go 100,000 without being in any kind of accident?

Is blowing out a tire in a pothole an accident?