Self Driving Cars?

[NW-Bound]

Truth is not the same as engineering progress.

One would hope that a truth is perpetual. Its acceptance may indeed take time.

On the other hand, a progress means it takes time to develop, as well as acceptance.

And when someone says more progress is still needed, that's also the truth. No?

 

[ERD50]

Looking forward to SDC whenever and however they rollout. And they don’t need to be “flawless” - they only need to be substantially better than the status quo with humans. The people actually working on SDC are light years beyond the concerns expressed here, but that won’t stop anyone from thinking otherwise and endlessly repeating their pet speculative talking points. Some people love to criticize things they don’t know much about...

It's really not clear (at least not to me) just what you are trying to say here.

If the "people actually working on SDC are light years beyond the concerns expressed here", then why did the Tesla perform so very poorly in the video that NW-Bound posted?

Or are you talking about Waymo or others? Are there videos showing their abilities in these situations?

You appear to be setting up a straw-man, saying that we accept only flawless. Where does that come from? Legal issues and perception aside, clearly if it's better than a reasonable benchmark (which might not be the 'average driver', but the driver of that class/cost of vehicles), then it is better. Who says other-wise?

And if you leave the timeline completely open, well, anything can happen, but I think people here are talking about the next 10~15 years. Many of us won't be driving much past that.

All truth passes through three stages. First, it is ridiculed. Second, it is violently opposed. Third, it is accepted as being self-evident. — Arthur Schopenhauer, German philosopher (1788 – 1860)

Humans haven’t changed in 200 years...

And you also think no one ever changes their mind on a forum

IMO, that quote is overused, just like "but, the Wright brothers!", or "we landed on the moon!", or "computers used to be the size of a room!". It's just a trite/cliche response to criticism, in order to make it sound like the criticism is not valid.

Just because something is ridiculed, doesn't make it true! Very poor logic to apply that quote to every case. We only know if it applies after we see the new idea to actually be a truth.

-ERD50

 

[NW-Bound]

Speaking of the Wright Brothers' invention...

Was it wrong when people back then said that the following contraptions could not fly?

If they said "flying machines are never possible", then of course they would be wrong.

Some people confuse the two statements. One is an observation of the current state of the art. The other is a prediction of the future.

What did the Wright Brothers think about the above two statements?

 

[ERD50]

.... But imagine if there were only level 5 cars on the road, they wouldn’t act unpredictably, and they’d probably all be “talking” to one another, working out issues far faster than humans. I don’t expect to live to see only level 5 cars everywhere, but isn’t that where this ends? ...
...

I don't think we know. Level 5 cars everywhere may be so far out, that things change so much, that Level 5 cars are no longer needed/wanted.

... In 1900 I am sure there would have been threads here with a handful of people explaining over and over and over why humans would never fly...

The year 1900 is good to illustrate my above point. In 1900, one f the biggest concerns in urban areas was horse manure in the streets, and even dead horses rotting by the side of the road, unclaimed. Mucky mud when it rained, nasty dust when it was dry. Many people were looking to solutions to this problem.

But none of those solutions were needed, because the car/truck took over so fast. Some breeds of then common work horses almost became extinct. For all we know, SDC may end up that way, alternative solutions we can't even foresee today.

-ERD50

 

[pj.mask]

If I recall, the reason Tesla’s run into parked trucks, medians, and semi trailers crossing to road is because the alternative risk is slamming on the brakes in the middle of the highway for a false reading - and I think statistically is more likely to cause an accident.

Imagine a plastic bag blowing across the road results in being rear ended.

We will get there but complaint on every failure seems to be ignoring the many successes.

As Elon has provided, auto pilot is much safer than humans along. There are lots of trucks on the road but I think there are many less trucks cutting across the highway where you have to significantly brake when the lighting is such that the trailer blends in with the surrounding sky (contrast issue).

I think we’re a long way off but I think we should be able to get reliable highway autopilot soon enough

 

[ERD50]

...

We will get there but complaint on every failure seems to be ignoring the many successes. ...

? No, the failures are what we must focus on, and compare to a benchmark.

I haven't had an at-fault accident in over 40 years (and that was a minor fender-bender). That's a lot of succeses. So what?

... As Elon has provided, auto pilot is much safer than humans along. There are lots of trucks on the road but I think there are many less trucks cutting across the highway where you have to significantly brake when the lighting is such that the trailer blends in with the surrounding sky (contrast issue).

...

Do you have any data that is up-to-date on that? IIRC, the data Elon provided was difficult to benchmark.

...

I think we’re a long way off but I think we should be able to get reliable highway autopilot soon enough

'soon enough'? What does that mean?

-ERD50

 

[NW-Bound]

If I recall, the reason Tesla’s run into parked trucks, medians, and semi trailers crossing to road is because the alternative risk is slamming on the brakes in the middle of the highway for a false reading - and I think statistically is more likely to cause an accident.

Imagine a plastic bag blowing across the road results in being rear ended...

Yes. That shows the difficulties of the problem. The first step in solving a problem is to recognize it.

We will get there but complaint on every failure seems to be ignoring the many successes...

There are amazing things done with this technology. There's a guy who builds an autopilot kit to be retrofitted into a few cars, and he is selling it for $1000.

On the surface, his test drive shows about the same capability as the Tesla autopilot (not FSD). However, he stressed that his system was only Level 2, and the driver must stay alert. His system even watches the driver's face, to be sure that his eyes are on the road. Of course I am impressed.

But if we talk about completely autonomous, no-steering-wheel cars, then I just watched a video interviewing several different researchers/developers in this field. They said that it would be "several decades away", without being more specific about the timeline.

... As Elon has provided, auto pilot is much safer than humans along...

Again, that's "Man+Machine" combination. They can complement each other, if the man does not fall asleep.

I am waiting to see the performance of his "Machine" without assistance from the driver.

 

[Midpack]

See you again in a month or so, when a small group of people will be saying the same things once again - going on two years now...

 

[NW-Bound]

Saying the same things?

We did not have ADAS and pedestrian avoidance systems to talk about, two years ago.

As time goes on, there will be more developments to talk about.

Perhaps Musk will deliver his fully autonomous FSD with the steering wheel removed. There would be a lot to talk about.

 

[explanade]

Yeah I'd give up driving myself, though I don't know if being a passenger in rush hour traffic would necessarily be more relaxing than having to drive it.

 

[NW-Bound]

^^^ Smartphones are great toys with which to spend the time. Same as most people, I reach for mine all the time when waiting at the airport, in doctor's offices.

And by the way, lane-keeping and following the car ahead is a very basic and safe form of self-driving. I would not mind having that feature at all, if I were still working and had to drive during rush hour.

I wonder how common the above feature is. It is not as involved as a full-scale autonomous car, and should be available in many new cars, but I don't know.

 

[NW-Bound]

Forgot to reiterate that I just talked about a $1000 add-on kit that provides a Level 2 autopilot that has about the same capability as the Tesla AP (without FSD).

And it is complained that I keep talking about the same thing.

 

[Chuckanut]

And by the way, lane-keeping and following the car ahead is a very basic and safe form of self-driving. I would not mind having that feature at all, if I were still working and had to drive during rush hour.

I wonder how common the above feature is. It is not as involved as a full-scale autonomous car, and should be available in many new cars, but I don't know.

FWIW, the Toyota basic safety package includes these features. I think it is on all new Toyota cars along with a few more features. I find the lane-keeping in conjunction with adaptive cruise control to be quite interesting. The car steers itself and if the car ahead slows down so does my car. When the car ahead speeds up so does my car, but very very slowly. But, it is NOT self driving. It does not read signs or detect the color of a traffic light. Still, it's better than not having it. And side striping that follows an off ramp seems to confuse it. It does make driving long distances on freeways a bit less stressful. But, I would never trust the safety of myself or my loved ones to it. And Toyota makes it clear it is not self-driving and the driver must still be alert.

Edited to add: The car will beep annoyingly at me if it detects my hands are not on the wheel for more than a few seconds.

Blind Spot monitoring is the #1 safety feature. It's a great aid when backing out of a parking space in a lot and some [-]idiot[/-] inexperienced driver comes tearing down the lane in the wrong direction at twice the safe speed.

 

[NW-Bound]

The above mode is exactly what I was talking about to alleviate the weariness of driving in bumper-to-bumper stop-and-go rush hour traffic.

And yes, in stop-and-go traffic, I think I would even allow myself to be distracted on/off with such mode. Repeat, only in stop-and-go traffic such as on LA freeways during rush hours, at 5 mph average speed.


PS. The problem I can see is I may get used to it, and allow myself to slowly expand my tolerance to the point of abusing it in more hazardous conditions. That tendency is real, and I may just fall into it.

 

[pj.mask]

? No, the failures are what we must focus on, and compare to a benchmark.

I haven't had an at-fault accident in over 40 years (and that was a minor fender-bender). That's a lot of succeses. So what?

What does “at fault” mean; it sounds like you have been involved in accidents. There are many documented cases of Tesla’s system avoiding accidents where they would have been Not-at-fault. Not to mention Tesla removing you from the driving that could have been a contributing factor (speeding, aggression, not paying attention).

I disagree that we need to focus on failures and ignore the success - as mentioned if they get the 88% part better by 50% but still fail (50% worse) at the 12% the. They would still reduce events by 32%. Sure, there is always room for improvement but you can’t discount/ignore what they have already improved.

Do you have any data that is up-to-date on that? IIRC, the data Elon provided was difficult to benchmark.

Tesla releases this every quarter, you can see a summary here: https://www.tesla.com/VehicleSafetyReport

While they don’t get into details, I’m not sure how you would be able to evaluate all the parameters.

Non-autopilot active safety is 1.28x as safe as regular drivers

Autopilot is 1.87x as safe as regular drivers

The real gem here is using autopilot is 1.46x safer than other Tesla’s - that to me is the most comparable statistic.

'soon enough'? What does that mean?

You can see where they are currently at:

https://youtu.be/fKXztwtXaGo

And I’ll bet it is a lot further than the $1000 system talked about above. I think we should have relative complete highway driving in good conditions within 5 years. (Good meaning no snow, heavy rain, with decent road markings or traffic) once they have enough examples of how to avoid stopped cars.

 

[Texas Proud]

But while SDC pedestrian avoidance needs to improve, this is only one aspect of automobile-related accidents, injuries, and deaths. I don't think I have ever needed to brake sharply to avoid a pedestrian, but I have had to brake sharply to avoid a serious accident with another vehicle. And while I realize that that is anecdotal, not data, according to the NHTSA, pedestrian fatalities are only 12% of all traffic fatalities. If SDCs can make a huge difference in the other 88% (https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/810968, p.11), I think we need to consider how many occupant lives we're basically willing to trade in order to save one pedestrian. I do think you are making a very important and often ignored point when it comes to SDC safety, of course, but it's not that simple.

Do not know if someone else posted on this or not....


The problem with this thinking is that pedestrian death would probably go up if a SDC cannot stop... I can say living in NY and London that cars have to brake for pedestrians all the time, especially in London where there are many crosswalks without lights....


So, you save lives in auto crashes but who knows if the total number of deaths go up or down... I would bet up...




BTW, a co-worker was hit by a car in NYC and it was a surprise to see that there is a preprinted form the cops have.... so, there are SO MANY pedestrian hits they made a form for it... how many of these would result in death instead of minor or major injury as happens now?

 

[NW-Bound]

My state has the infamous claim of having twice the per capita rate of pedestrian deaths of the US average. In 2017, Arizona had a total of 1736 pedestrian accidents, of 226 deaths plus 1510 injuries, mostly happening in the Phoenix metropolitan area.

I am glad to hear a poster saying ADAS is now standard on some cars priced as low as $20K. Apparently, it does not take expensive hardware to implement this function of pedestrian detection and emergency braking. It does not work really well yet, but I am sure it will improve with time. Even if it fails 1/3 of the time to catch driver's mistake, that's still 2/3 of the accidents that will be avoided.

AZ is also the place of the first SDC fatal accident. Some of you may remember this. We talked about that here. In March 2018, an experimental SDC by Uber struck and killed a woman jaywalking at night, in Tempe and at a location not too far from Arizona State University. It caused quite a national stir.

PS. In 2017, AZ had 32 cyclist fatal accidents, plus 1371 injuries. This makes us the 5th dangerous state for cyclists.

 

[Chuckanut]

PS. The problem I can see is I may get used to it, and allow myself to slowly expand my tolerance to the point of abusing it in more hazardous conditions. That tendency is real, and I may just fall into it.

FWIW, my Rav4 will beep annoyingly at me if it detects my hands are not on the wheel. So, there is a limit to one's expanding tolerance of the vehicle's abilities.

 

[NW-Bound]

FWIW, my Rav4 will beep annoyingly at me if it detects my hands are not on the wheel. So, there is a limit to one's expanding tolerance of the vehicle's abilities.

OK, that's good. No texting or surfin' on the smartphone then. How long does it let you take your hands off the wheel?

Tesla initially did not have any requirement. Then, they put on some requirements, but drivers quickly found a way to defeat it using the "orange trick". I am not a Tesla owner, and even I knew about this. I don't know if this trick still works.

 

[Chuckanut]

OK, that's good. No texting or surfin' on the smartphone then. How long does it let you take your hands off the wheel?

Maybe 10 seconds at the most?? I have to time it one day. If I don't crash, I'll report a more accurate number.

 

[ERD50]

What does “at fault” mean; it sounds like you have been involved in accidents. There are many documented cases of Tesla’s system avoiding accidents where they would have been Not-at-fault. ...

This makes no sense. Of course I have avoided many accidents.

... I disagree that we need to focus on failures and ignore the success - as mentioned if they get the 88% part better by 50% but still fail (50% worse) at the 12% the. They would still reduce events by 32%. Sure, there is always room for improvement but you can’t discount/ignore what they have already improved.
...

But that's just an "if". I can throw out any scenario I want, that doesn't mean it reflects reality. The question is, are they improving enough to reduce accidents overall.

We need to put failures into context - accidents per miles driven with a comparable benchmark (same profile of cars/drivers/conditions).

...


Tesla releases this every quarter, you can see a summary here: https://www.tesla.com/VehicleSafetyReport

While they don’t get into details, I’m not sure how you would be able to evaluate all the parameters.

Non-autopilot active safety is 1.28x as safe as regular drivers

Autopilot is 1.87x as safe as regular drivers

The real gem here is using autopilot is 1.46x safer than other Tesla’s - that to me is the most comparable statistic. ...

That lacks context, and you take it out of context. The higher accident rate w/o Autopilot engaged could be because Autopilot can't be engaged in that scenario. We need more info.

From Tesla's site:

On Model 3, Navigate on Autopilot can be engaged on most highways by moving the gear lever twice downwards, in quick succession.

On Model S and Model X, Navigate on Autopilot can be engaged on most highways by pulling the cruise stalk toward you, twice in quick succession.

So only on highways? IIRC, highways are ~ 4x safer than other roads. This all needs context and bench-marking to be meaningful.

...

You can see where they are currently at:

https://youtu.be/fKXztwtXaGo

... I think we should have relative complete highway driving in good conditions within 5 years. (Good meaning no snow, heavy rain, with decent road markings or traffic) once they have enough examples of how to avoid stopped cars.

If you can define "relative complete highway driving in good conditions", we can set our calendars to revisit this in 5 years. While we certainly will see improvements, I remain very skeptical that a sub $100,000 (un-subsidized) car could be relied on to be fully autonomous, even limited to a highway, and even limited to good conditions. Maybe on special segments of roads dedicated to testing the technology, but not just for any given road trip, where the driver can just read, take a nap, whatever, and let the car take over for hours. I could be proven to be wrong, but I highly doubt it.

There is a heck of a lot more to it than avoiding stopped cars.

-ERD50

 

[NW-Bound]

Elon promised that in 2020, the Tesla Model 3 FSD (Full Self-Driving) will be good enough to allow it to be used in a taxi fleet. Moreover, he said Model 3 owners could sign up for their cars to join the service to have some income. I watched the video of his presentation.

It's a very interesting idea. When you do not need your car, you tell Tesla it is available for hire. Then, while parked on your driveway, your Tesla suddenly backs out to the street, runs a taxi trip for hire, and comes back to your driveway until the next call. Or it can run one trip after another if there's demand, until it needs to come back to you for a recharge. Your car makes money for you. Nice income. Think of it as Uber, except you do not have to drive and stay home.

Interesting business model aside, this full self-driving capability would be amazing. This is something Waymo is still testing and testing in my town. I encounter Waymo cars out in the streets every time I run an errand. They are all over the place. When are they going to go into revenue mode? What is Waymo afraid of?

So, of course I am waiting for Tesla to turn on this FSD capability on the Tesla 3 for the whole world to see. It's this year.

 

[Midpack]

^^^ Musk is famous for overpromising on timing, but is that what’s most important? Or are the accomplishments themselves what are of most value whenever they come to pass? If Tesla achieves FSD in 2025, is that negated by not doing it by 2020? Same with Waymo timing, I’m interested in seeing what they achieve far more than when. All the “experts” here tell us how much more difficult SDC is than anyone realizes, but then you want to judge them by when they reach level 4 or 5?

 

[NW-Bound]

Then, do we not talk about the progress they are making along the way? Do we not talk about the deficiencies that still have to be corrected?

As an engineer, I want to understand these points, and even if I do not work on this technology, I am very interested in different approaches to the same problem.

Take the Boeing 737 Max problem. I read about it, and tried to understand the basic root cause (failure of an alpha vane), and read about a similar problem that caused a crash of an Airbus.

Again, I am an engineer, and many of the posters here have the same interest. The more we understand a problem, the more we appreciate the solution. The tougher the problem, the more I admire a brilliant solution. If it were easy, then a caveman could have done it.

And we do not believe everything salesmen tell us. Not regarding any product, or political issues for that matter. Engineers think differently. I was told that in school, they now call it "critical thinking". I simply call it thinking.

When several developers/researchers said a true SDC of Level 4 or 5 would be "several decades away", and someone said he would deliver it in this year, you are telling me that should not catch my attention?



PS. I occasionally watch some Youtube videos, in order to learn about the real-life performance of these systems. I usually do not read the comments, because they are made by ignoramuses. But one time, I found something that was hilarious. The commenter said Tesla autopilot was a "sentient" thing.

 

[NW-Bound]

Here's another thing. Waymo cars have a lot more sensors than Tesla cars. A lot more.

Why is Waymo so paranoid? Sensors cost money, add complexities, and add more downtime to the vehicle when they fail (a Level 4 or 5 SDC would detect a failure, and pull over).

So, I studied the location of Waymo sensors on their test car, trying to guess what they are trying to detect.

And then, I watched the performance of the other car, trying to catch the deficiency that I suspect it would have for the lack of that sensor. A Tesla owner reported something on Youtube that confirmed my thinking. Bingo!

But I guess I will not bore you with it.