Self Driving Cars are Going Nowhere

https://www.bloomberg.com/news/features/2022-10-06/even-after-100-billion-self-driving-cars-are-going-nowhere

Maybe a couple excerpts?

“ They were supposed to be the future. But prominent detractors—including Anthony Levandowski, who pioneered the industry—are getting louder as the losses get bigger.”

And

“ It all sounds great until you encounter an actual robo-taxi in the wild. Which is rare: Six years after companies started offering rides in what they’ve called autonomous cars and almost 20 years after the first self-driving demos, there are vanishingly few such vehicles on the road. And they tend to be confined to a handful of places in the Sun Belt, because they still can’t handle weather patterns trickier than Partly Cloudy. State-of-the-art robot cars also struggle with construction, animals, traffic cones, crossing guards, and what the industry calls “unprotected left turns,” which most of us would call “left turns.” “

They are nowhere close to solving the weather problem: rain, snow, fog, and oh yes, darkness. And they’re even further away from having the AI do anything at all about “edge” situations.

What’s NOT an edge situation? Driving between the lines on a limited access road without such inconveniences as traffic cones, pedestrians, kick balls rolling in the street, animals, drivers doing unexpected things, and lots and lots of others. It’s fairly easy to train on very normal streets, not so much any other place. And, as one proponent points out, even if you get the car to navigate an edge situation once, there’s no guarantee it will do it the next time. Or the next. Or the next.

One edge situation described in the article; a flock of birds walking around on or near the road. A human understands that the birds will fly away, a robot car probably doesn’t, so it brakes. (Most common accident involving self driving cars: rear end collisions.) But suppose the birds aren’t birds but are, oh, say cats, or gophers or squirrels. The AI doesn’t (yet) distinguish because those are rare enough that there isn’t enough training because it’s just too uncommon. (And no, simulations don’t cover all “edge” cases, even if you run them a million times. The “edge” case is the outlier, and you can’t train a bazillion outliers.

Warren Buffett famously said something like investing in airlines has been a net loss for investors since the beginning of time; it would have been better if they simply torched the industry (from an investment point of view) than trying to make a profit from it. It may be that “self driving cars”, along with “flying cars” and a bunch of other gee-willickers things is another of those cases: it just may never come true.

4 Likes

Maybe about 7 - 8 years ago, the local surface streets in the city of Mountain View (near Google’s main campus) used to be a common “training” area for the Google autonomous cars. The cars were also on my work campus for about a year or two. Have not seen the cars (easily recognizable with their “turrets” ) in years. Yup, autonomous definitely is hard.

1 Like

I think the economically feasible use of self-driving vehicles will be long-haul truck transport on interstates between warehouses located in isolated areas. Not passenger cars in cities.

Wendy

2 Likes

I donno recent MIT computer scientist v Bloomberg report. Who would be right about this?

Would the industry blow smoke? Why?

The MIT grad says the problem is the legal framework for liability.

No one building the cars wants the liability. There is no driver to take the blame. The corporations are going to shelve the self driving cars entirely.

But they are safer than human driven cars today as we speak according to the MIT guys. It does not take much to be safer than a human driver. LOL

Note the article found a case of birds v little animals as outliers that do not happen often enough to work it out. If they are not happening all that often that is not a reason to say this does not work. It means it works and they are really digging for any excuse in the outliers to say it doesn’t work.

There are three sensors onboard, radar, infrared and optical. Meaning cloud cover and rain are not a big deal for these cars. I do not know who they are fooling. I will give you the accident rate will rise slightly. The accident rate for human drivers soars under these conditions. D’oh!

It all goes to show you how profitable the auto insurance industry is because in effect people are paying at least twice over for their cars.

But, but but but, the car manufactures are then building two cars more or less to replace the wrecked cars. Those are not totally accurate rates of cars totalled. Not meant to be. In essence that is our world…buy it, insure it…to replace it…all at the driver’s expense. We wouldn’t want to give that up. Seriously a model where the manufacture built two cars for the price of one plus paid out on serious accidents…a total no go.

3 Likes

What Bloomberg published is essentially a human interest piece, written by somebody who knows nothing about autonomy or AI. What that means is that there’s no distinguishing between self-serving whining and technical truth. That piece is full of “I’ve failed repeatedly, so the only way I can preserve my self-respect is by saying it can’t be done.”

Whole lotta failure to go around, but all it shows so far is that endless money and desire doesn’t mean that you’ll win.

That’s just another way of saying that the technical problems of autonomy won’t be solved. The “economically feasible” should be replaced by simply “feasible”, because anywhere self-driving works it is a huge economic win.

I think you’re wrong, but your innate conservatism means you don’t believe anything until you see it working, at least so far as I can tell. Nothing wrong with that, except that you’ll be wrong about everything new until it’s old.

-IGU-

9 Likes

I don’t understand the all-or-nothing view. I was stranded in an airport before COVID with a guy who was on his way home from an agri-tech conference. He said that one farmer was controlling multiple mostly-autonomous tractors, harvestors, what have you, across multiple fields. Getting rid of the last farmer entirely was off in the future but it didn’t matter, they already had 90% of the savings.

The same thing for Wendy’s interstate example. If I can turn on autopilot and watch TV while the car drives from NY to Chicago but I still have to take over at each end, and when the weather turns bad, I’m in. Each partial-use-case provides profit and testbeds for further development.

13 Likes

Legend has it that the Wright brothers’ uncle said in the sermon the Sunday before their first flight that if god had wished for man to fly he would have given us wings. When motor cars first appeared there was a proposal that a man with a red flag walk in front of it so as not to scare the horses. Self driving is in good company.

Talking about nonsense reasoning, one commenter said that if EVs were so good why hadn’t the market adopted it sooner and faster. Of course he ignored the fact that EV makers are selling all they can produce and it is hard to adopt the technology faster than it can be produced.

From horses to cars in under 15 years

The Great Transformation [Part 1] - Patterns of Change, Key Technologies & #PhaseChangeDisruption

The fuller version

Tony Seba: Clean Disruption - Energy & Transportation

The Captain

3 Likes

Not exactly. All technological progress in history has been incremental.

For example, did you know that humans rode horses for hundreds of years with their feet hanging down because the stirrup wasn’t invented until the late 6th century CE?

I think that using self-driving trucks as automatic pilots on interstates would be low-hanging fruit. Low risk, high profitability. This would encourage large-scale manufacturing that would support the R&D for the more challenging task of driving in cities.

Within 50 years, people will be shaking their heads and saying, “How could they ever have allowed humans to drive cars with 50,000 fatalities a year?”

Wendy

8 Likes

That is the exact question I have been asking for a few years. Had a discussion about that with the transportation curator at the Henry Ford Museum.

Right now, all the interested parties have an easy scapegoat: the human behind the wheel. If you tell the human behind the wheel he can take a nap, then who has the liability? The company that built the car? The company that wrote the software? The company that made the sensors? If the supply chain was diversified enough, everyone could point fingers, while the injured party sues one party after another, for years, without resolution. But no-one will write liability insurance for those companies unless the responsibility is fixed, so they can assess the risk. The probabilities that all the insurance companies will agree to keep the liability on the driver are pretty good. So, the companies advertise “autonomous” cars, with a big asterisk, and a footnote that says the driver is responsible for remaining awake and alert, and constantly monitoring the situation. Yeah, that’s the ticket, claim autonomy, but say the driver still bears responsibility for anticipating and pro-actively correcting errors by the system.
/sarcasm

Steve

3 Likes

I’m kinda thinking that now.

5 Likes

It’s pretty clear that Tesla is preparing to take responsibility when their autonomy software is good enough. That’s why they have established an insurance business, which is currently offered in the states of over half their drivers. And expanding.

Elon Musk doesn’t like anybody else in a position to slow things down. So he’s not going to let insurance companies be in a position to refuse to insure autonomous Teslas.

Meanwhile, since they know exactly how you drive, and how much, and where, and will know whether you’re at fault in any crash, they can offer better rates than other insurance companies. A concern, of course, is whether they will abuse this knowledge. So far there’s no evidence of that.

-IGU-

5 Likes

Ray Kurzweil said that not only is progress accelerating but time itself is compressing as systems get more organized (The Age of Spiritual Machines)

Tony Seba says that car adoption went from 10% to 80% in ten years:
The Great Transformation [Part 1] - Patterns of Change, Key Technologies & #PhaseChangeDisruption - YouTube

Smartphone History:

1973: The First Cell Phone Is Invented
1984: The First Portable Mobile Device
1992: The First Smartphone
1994: The Earliest Form of Smartphone Apps
1997: The Start of Mobile Gaming
2000: The First Camera Phone
2001: Cell Phones Access the Internet
2007: Apple Joins the Game
2020: The Modern Smartphone
2022 and Beyond: What’s Next?

Apple disrupted the cellphone industry in less than 15 years

I wonder how relevant your pre Industrial Age example might be…

The Captain

5 Likes

I think your own timeline says a lot: The cell phone was invented in 1973. We used a lot of “brick phones” in the newsroom.

The became somewhat common for consumers in the 90’s - a 20 year adaptation/adoption frame.

The iPhone (which arguably changed everything) came along in another 10+ years. And it’s now been another decade and a half.

Yeah, things change quicker than horses with stirrups, but not nearly as fast as most tech prognosticators think - at least as far as adaption cycles by the general public. Of course techies are wowed by each small change in the technology, so their viewpoint is micro, not macro.

I acknowledge that in 50 years crypto might be a thing. (Or not.) But for the moment it is merely an interesting thing, without a lot of use, or adoption in any meaningful consumer facing transactions.

Likewise, maybe virtual reality, but maybe not. And FSD cars, but not now - and not likely soon. Of course if you can afford to wait 20-50 years, great. Ford decided they couldn’t, and shuttered that division. Others will follow, but eventually, maybe, someone will succeed. It’s still going to take a LOT of time, more than most people think - even the pessimists.

3 Likes

Tesla could inch its way in there to self driving cars.

It is doing it incrementally for now.

It can get very dicy. Insurance fraud could blow up the model overnight. Then Tesla could have millions of vehicles as liabilities. States could issue insurance laws that scr$w up Tesla and then millions of uninsured vehicles where Tesla has the liability.

Agree.

Bad analogy: the first steamships were harbor tugs (moving the sailing ships into/out of anchorages), and riverboats on shallow rivers (e.g. going up the Mississippi). Low-hanging fruit, for niche needs. Lasted a generation, while other beta testing went nowhere (memo to world’s militaries, circa 1850: oceangoing paddlewheel warships = bad idea)

For self-driving, I’m thinking elders who are for another decade or more otherwise independent, but would just as soon not have to drive, nor be reliant on others: grocery store and doctor runs, and the proverbial church-on-Sunday. Short, daylight trips on well-striped routes with the software warning about anything that was now or soon violating autonomous norms wrt weather/daylight and might require self-piloting. We whine about worsening caretaker shortage (maybe there will be robots!) – well, here’s a caretaker for you, almost ready to go, under these limited conditions.

Or, tourists going from the Washington Monument to the far end of the National Mall, or back to their hotel.

Or, those not-lamented 12+ hour days when I drove from home to hospital A to hospital B to office to hospital B to home. (I know many urban specialists whose schedules are worse - hospitals A-D, and maybe even E and F, especially when getting started in their career). The ability to look through medical records [morning labs! EKGs! scans! path reports!] instead of waiting for the light at Broadway and 12th, again, in the middle of a jammed workday – instead of driving, parking, trotting inside the hospital, finding a terminal, logging on to their system, looking up those labs etc while all burning through more irreplaceable minutes away from FTF patient care. (So what if you have to self-drive half the time in winter? That’s still dozens of hours saved per month when averaged over a year)

Or, dispatching the otherwise human-controlled EV sedan to the grocery store autonomously, having the clerk unlock and load with a one-time bar code, then the EV coming home, parking in your carport and letting your hitherto-uninterrupted work-from-home self know it was ready to be unloaded?

I suspect over the next decade or so, limited self-driving will arrive insidiously (or invidiously, depending on your POV) in various corners of all of our lives

–sutton
who remembers a fellow medical student – a very bright woman, now a highly respected subspecialist in a major metropolitan area – asking me in 1983+/- "If you had a computer what would you do with it?"

3 Likes

Goofy:

We tend to focus on the “S” part of the technology adoption cycle but that is about one third of the full cycle. There is a long slow run-up that ends when the technology Crosses the Chasm. That’s when the Tornado happens and producers sell all they can make. That’s where EVs are now. I believe the EV technology Crossed The Chasm around 2019-20.

The sharp upward turn of the “S” curve tends to be around 15% market penetration. With EVs is seems to have happened sooner, it could be on account of the subsidies governments provided. The curve at the top should be at around 85% market penetration. My best current guess is that it will happen around 2030, maybe 2035. Then growth slows dramatically as the old geezers are finally pulled in kicking and screaming. I didn’t get a cellphone/smartphone until April 2019 when I arrived in Madrid, escaping from Venezuela. Quite simply today it’s very hard or maybe impossible to live without one. I dislike the things!

So yes, you are partly right, but do take into account the varying speed of adoption along the technology’s life cycle, it’s NOT steady as she goes. It’s Slow, Fast, Slow. One could say that the EV technology is 120 years old! Few technologies are reborn like the EV technology. Maybe we can date the second coming with the appearance of the six volt golf carts.

The Captain

5 Likes

And sometimes it’s “don’t go at all for years, maybe ever”.

Remember Segue, the scooters that were going to change transportation? Microsoft put out a tablet in 2002, and a watch in 2004. I remember seeing an AT&T video phone at the World’s Fair in 1964.

EVs are clearly a product that has met with success (although I guesstimate the eventual penetration at far less than most,) but we’re talking about “self driving” and I put that one very far out in the future, especially if we’re talking about mass adoption.

Sure, there will be some applications, as Bloomberg reports, one place is in rock mining driving retrofitted trucks that never change course and don’t have intrusions like traffic, kids, cones, etc. And, as we’ve seen, strictly geofenced taxis, maybe long distance trucking. Worthy applications, all, but not on the scale that FSD advocates have been advancing.

I’d like to see it too, but I have my doubts.

2 Likes

Caterpillar already has a 250-ton dump truck that is fully autonomous and in use, and has for several years now. Agriculture has autonomous robots with computer vision to selectively spray fertilizer on plants and herbicides on weeds.

I have a portion of my portofolio in the self-driving arena. That is in DE and in AGCO, both are agriculture plays, not personal transportation.

3 Likes

J6P is driving a Tesla and has it insured with Tesla.

Now Joe decides to go hands free and BBQ in the backseat. There is a major accident. Joe’s family goes to court to sue Tesla and Tesla insurance. The court decides that Tesla insurance is Tesla no matter how many legal hoops were created in its setup. Tesla is on the hook.

Suddenly there are class action suits for whiplash in a Tesla or whatever. Tesla motors is assigned full liability by the courts.

Tesla becomes toast as millions of people drop their Tesla insurance product and drive hands free. The company will pay.

revision this morning, millions do not necessarily drop their tesla insurance at all. Instead Tesla Insurance for whatever reasons goes bankrupt and the parent company is on the hook for all sorts of damages.

1 Like

Except my Grandmother came across the prairie in a covered wagon, My Grandfather had one of the first Ford Garages, and my Grandmother died after the first man was put on the moon and she was flying around the country. It seems advances are happening faster and faster.

Andy

4 Likes