If you ask Elon Musk, the future of autonomous driving technology is through robotaxis like the Tesla Cybercab. Up until recently, General Motors might have said the same thing. Not anymore, GM announced today it is no longer going to fund robotaxi development work.
“GM is committed to delivering the best driving experiences to our customers in a disciplined and capital efficient manner,” said Mary Barra, chair and CEO of GM. “Cruise has been an early innovator in autonomy, and the deeper integration of our teams, paired with GM’s strong brands, scale, and manufacturing strength, will help advance our vision for the future of transportation.”
All of this is a little bit of a surprise given how much money GM once projected to make from its Cruise robotaxi business, though less of a surprise when considering how much money GM has spent, which the company’s CFO just said was about $10 billion on a conference call.
In addition to costing money, Cruise has been a hassle for the automaker.
Last October, a Cruise Robotaxi was out driving in San Francisco when a nearby car struck a pedestrian. Unfortunately, the Cruise robotaxi hit the pedestrian and dragged her about 20 feet, trapping her under the car.
The company’s response involved hiding information from regulators, which led to fines. Cruise’s self-driving license was quickly revoked. The service was shut down, and many of the executives were either fired or left. Ex-CEO Kyle Vogt, in particular, was kicked to the curb and most of the company’s mistakes were pinned on him.
[Update: Speaking of Vogt, here’s his recent comment:
Interesting. -DT]
This isn’t to say the company never believed in Cruise’s mission. Before this incident, GM had partnered with Honda to build a robotaxi called the Cruise Origin. In fact, they built more than a few, only to have to stash them at a GM facility in Michigan.
“Robotaxi is not [our] core business today,” Barra pointed out, a few times on the conference call this evening. GM assumes $1 billion in savings every year from not having to try to serve the working capital needed to maintain a fleet.
GM is planning to combine Cruise into GM’s technical team to push forward the company’s SuperCruise autonomous system for personal vehicles. Currently, GM owns about 90% of Cruise. The goal for the company is to get 100% of Cruise so that it can bring the Cruise team and technology into the main company.
When asked directly by an analyst if Barra thought a robotaxi business, with all of the regulatory hurdles and associated costs, was even possible, Barra mostly avoided answering the question.
It’s a good question! A robotaxi business is something that will likely happen in urban areas, at least, in the coming decade. Getting from a small number of cabs to that point is going to cost an incredible amount of money and, since the accident, it’s been clear GM wasn’t interested in spending that kind of money.
Instead, Barra mostly pointed out that moving from Level 3 to Level 4 autonomy for personal vehicles was GM’s main goal since building personal vehicles is GM’s main business.
“Our commitment to driver assistance technology… is unwavering” Barra concluded at the end of her call with analysts today.
Images: Cruise; Iurii/stock.adobe.com
I believe Elon’s idea isn’t to sell RoboTaxis to the public, but to rather unleash them, starting in large urban areas. This would be exactly like Uber, but they wouldn’t have to pay anyone to drive them. In that respect, this is a multi- billion dollar business that, done right, could totally sink Uber and Lyft.
Further proof that C-Suite folks aren’t like me.
If I had an idea that cost the company a ton of money and didn’t pan out, there’s no way in hell I’d still be employed by that company.
C-Suiters? “Oops, my bad! Let’s refocus (feel free to insert another corporate-speak cliche here).”
Didn’t our old friend Roy Wort once worked for them?
What’s with the perception that someone admitting to a mistake, or changing priorities, is a bad thing? “This was a valuable pursuit for us at the inception of the program, and we have made great strides and learned many things from our involvement in the Cruise endeavor. However, we have made the decision to refocus on our core product and improving our SuperCruise driver assistance suite, and see no reason to continue investing in the autonomous taxi market.”
Because being forthright just invites punishment
This should ring alarm bells for investors. They lie about the possible returns.
Full self-driving is one of the better Silly Con Valley dupes. Why you would drop money on a car that you need to insure yourself but it can repossess itself?
The intent for Cruise, Waymo, etc isn’t for you to own the car, it’s for taxi purposes.
Not saying robotaxis make a ton of sense either, but neither Cruise nor Waymo ever included personally owning one of their cars as part of their business model.
Having ridden in a Waymo, it was a completely normal and uneventful Uber-like trip other than the fact I was riding by myself in the backseat with no person at the wheel. The car’s ability to navigate congestion and complex traffic patterns at the same flow of traffic (not overly cautious and timid which is hazardous by itself) was seriously impressive. I don’t think it will change the world per se, but Waymo has some very impressive technology.
she is the one who should be fired. so many typical bad GM choices time and time again
Honestly if has taken $10 billion already and what they’re saying is it’s going to take multiples of that before it operates profitably, then I’d say this is one of her better decisions.
She’s avoiding a sunk cost fallacy. It’s not a bad choice
Development of the taxis may have stopped, but if the ADAS improvements that are going to come through innovations anyway there will come a point that they will be able to throw together existing technology to roll out a robotaxi. Therefore there is no need to have a distinct company or part of your organization that is specifically all-taxi-all-the-time.
They can pivot and just be like, ‘yo what up we’re doing taxis again what it do.
Others working on taxi businesses don’t have a car company they also need to run. GM has no need to be first in that.
You could be right, Technology!!, the eBike is coming for me before I even think of going there.
Full self driving is nothing but Silicon Valley hubris. They think they are so much smarter than everyone, so of course they can program a computer to do it better. Not realizing how smart the average human is and how good even dumb people can do a complex task like driving. Full self driving is a long time off. We were promised it years ago and no closer.
I don’t know why they haven’t just come out with slow moving taxis for local trips that just don’t exceed lower speed limits. If they capped their speed at, say 30 or 35 mph there would still be a huge use for them and they would have to do so much less development. Even if they only went 25 mph there would still be a lot of uses for local travel in cities and other applications.
There is no regulatory framework for a lower speed vehicles in the United States, however.
Too many random things they can’t program in.
At that point why not take the bus?
There is a regulatory category for street-legal LSVs in the US, by the way – Neighborhood Electric Vehicles. I saw some as tour vehicles in Philadelphia a few months ago.
they could put them on rails between high traffic destinations and run at a regular schedule. I bet some could even be underground, or elevated above grade level.
Idk, that sounds a bit too much like socialism to me.
NEURAL NETWORKS
MACHINE LEARNING
SENTIENT ALGORITHMS
They just think if you yell these loud enough at enough people it will just happen.
No, they think if you yell it loud enough they’ll get billions in VC money and all they have to do is release a product with ten percent of the proposed features with 3 times the original development timeline.
True. I wish I could bull shit people out of millions. That seems to be all life is.
The city transit problem has been solved with trains, trolleys, subways, and buses. But our country views mass transit as a handout to poor people, so none of those things get the support it needs.
Well, I was 100% full self driving cars will never happen in my lifetime, just too many uncontrollable variables. But this changes everything! Based on GM’s history, of being right on it, then pulling out. Level 5 within a decade, for sure.
This is bulletproof logic, actually.
It’s just not happening. I’m going back in my den now.
“Last October, a Cruise Robotaxi was out driving in San Francisco when a nearby car struck a pedestrian. Unfortunately, the Cruise robotaxi hit the pedestrian and dragged her about 20 feet, trapping her under the car.”
It’s worth remembering that pedestrian was knocked into the Cruise car by the other car driven by a meatbag, the Cruise didn’t just hit her. The meat bag then SPED OFF and has yet to be caught while the Cruise pulled over right away. There is absolutely no evidence of which I am aware that a meat bag in the place of the Cruise would have done any better and its very clear the other car driven by a meatbag who knowingly hit that pedestrian did far, far worse.
The Cruise car did drag the pedestrian for 20ft after recognizing that an impact occurred, before pulling over and stopping
But I think the main issue wasn’t really that, but more how Cruise management responded to it afterward that wrecked confidence in the company from regulators and GM
Again I don’t see how a human could have done any better. The Cruise registered a thump and saw nothing on its cameras so it pulled over ASAP to assess the situation. 20 ft is about 2 car lengths, about the minimum distance from the left lane to the nearest safe place to exit the flow of traffic. That is exactly what would have been expected of any human driver. Do critics feel it should it have stayed put where it was and been a roadblock instead? Or get out in traffic and risk getting hit themselves?
The company should certainly have responded better. They could have pointed out this whole tragedy was caused by a POS meat bag and kept the outrage focused where it belongs, on that meat bag of terrible decisions. If anything this story is a great example of why AVs are better than meat bags.
It seems to me however many critics, (including many commentators here) aren’t willing to accept that. They are blinded by the threat poised to their livelihoods or freedums by AVs and have been using this to whip up anti AV hysteria by demonizing the AV for doing the right thing while ignoring or trivializing the initial hit and run.
Had the woman been thrown under and dragged by a car driven by a human there would have been no story other than that for the manhunt of the POS meat bag whom as I mentioned is still at large.
Holy shit, this is the level we accept now?
The billion dollar (sometimes using your tax dollars) product being designed to do the same or worse than the ambient shittery of your human neighbors doing it for free is actually part of the problem of modern tech.
Its not “the same or worse at all”. In fact having a pedestrian thrown at you, then pulling over ASAP and waiting for the cops even though you don’t know what happened is EXACTLY the opposite of knowing damn well you hit someone then and ran.
Please redirect your outrage to the meat bag who deserves it.
I remember when this story was 1st posted on the old site. The story had been that the Cruise just stopped and did not attempt to move after coming to rest on top of the person. Commenters were outraged and said the Cruise should have at least attempted to move itself off of the person. Then later it was revealed the Cruise didn’t just sit there and it dragged the person 20ft again commenters angry. People would’ve been upset no matter what the Cruise reaction was, but where they really messed up was not being transparent from the beginning.
The cruise taxi kept driving after the pedestrian was hit. A human driver, most likely would’ve stopped.
Assuming you simply did not read the story and are not trolling I will reiterate:
The human driver involved, the one who hit the pedestrian in the first place, did NOT stop. That meat bag – knowing full well a pedestrian had been hit – took off and remains at large.
That pedestrian was shoved into the Cruse which registered a collision of some type. It did not *see* anything because the pedestrian was not visible. A human driver would not have seen anything either.
The Cruse OTOH IMMEDIATELY pulled over but needed to move 20 feet out of the traffic lane to do so. Unfortunately the pedestrian was caught underneath the car. 20 feet is the minimum distance the car could be expected to move to get to a safe location out of the flow of traffic. So no, the Cruse did NOT “keep on driving”.
I think its quite likely a human driver in that situation would have shrugged and drove on, taking much more time to realize the strange noises were a pedestrian trapped under the car. I think that pedestrian would would have been injured much more grievously or even killed by a human driver.
So now that you know what happened you too can properly focus your outrage at the meat bag driver who remains at large and support the further development of AVs.
I believe the human driver, being sentient, would’ve stopped.
It’s difficult to program elevators, if the elevators in my office building are indication. If computers can’t master elevator programming, how on earth can they perfect Driving automation.
I am 100% on board with driver assist, but for automation, not a chance
I believe the human driver, being sentient, would’ve stopped.
Again the human driver who actually was involved proves otherwise. And again the Cruse DID stop but moved out of the flow of traffic first which was the safest thing to do.
It’s difficult to program elevators, if the elevators in my office building are indication. If computers can’t master elevator programming, how on earth can they perfect Driving automation.
They don’t need to be perfect, just better than humans. Would you prefer elevators be in control of rando humans?
I am 100% on board with driver assist, but for automation, not a chance
You should be. Too many humans are terrible drivers who make terrible decisions like intentionally ignore traffic laws, traffic signs, intimidate others, fleeing the scene, DUI, drowsy driving and road rage. AVs will never weave through crowded traffic to get home to catch a sportsball game or to use the bathroom.
Human vision is also limited to the visible spectrum wheras an AV can see anywhere from ultraviolet to microwave, maybe even more. That means it can see much better in poor conditions like heavy fog. Same with sound, a car can hear sounds imperceptible to humans. And that’s for humans at their peak – human senses deteriorate with age.
Cars are also sometimes driven by half blind, barely self aware octogenarians, mentally ill, and incompetent folks too you know. Funny how the same folks so concerned with the perfection of AVs never seem to mention that.
Furthermore AVs also can communicate not only with the car ahead but all the cars ahead. It can communicate with the road too so it can know the right speed to make all the green lights, even adaptive traffic lights. A human has very limited capacity to do the same and only as an added layer of complexity to the same systems as the AV would use.
Airliners are pretty self-driving. But they still have someone onboard to take over if needed.
If airliners can’t accomplish unsupervised self driving, with a greatly reduced set of variables compared to cars, how will cars?
Of course, the risk of one plane crash is much greater than the risk of the crash of an automated car. But, of course, there will be a lot more of the automated cars…
Trains still need drivers, and they literally only go forwards or backwards. Too many people try to say complex things are simple, when they are not. Take trains. Did you know you can snap the couplings by accelerating too fast after a hill? The back half is still climbing while the front is going downhill. You have to use the throttle just right to keep enough headway to not stall out on a hill while not going so fast you pull things apart. And if the driver does break a coupler, they have to fix it themselves. Which means you carry a spare 120lb coulper from an engine all the way to the break. And that can be quite a distance.
Good point.
Is it really that damn expensive to employ people? People need jobs.
As someone who is a boss, I can testify that it’s extremely hard to get people to do what they are supposed to do.
Shit, it’s hard for me to make me do what I’m supposed to do.
So, you are bad at it then? 🙂 Because literally millions of employees successfully complete their tasks every single day.
I guess option B would be to program a computer to do those jobs successfully 100.0% of the time.
I came here to say basically this. If railroads can’t be automated, how can cars? With Driving there is an infinite number of variables, and it’s not possible to program for an infinite number of variables.
I don’t know about railroads, but modern subway systems don’t have drivers anymore. Of course there are people who can take control of necessary, and cameras everywhere, but they are automated
The autopilot in an airplane is a really simple system, all it can do is keep the plane pointed in the right direction and altitude, at the desired speed. The first autopilots were built without any kind of computer, if I’m not mistaken. Granted, modern planes have a bunch of technology and can do much more than what the autopilot could, but they are far from self-flying. And if you think about it, flying, other than take-off and landing, is a lot simpler than driving. No roads, no road signs, no traffic except in the vicinity of the airport, where there’s a team of humans (ATC) making sure nothing bad happens. Also you usually have a lot of room for errors. It doesn’t matter if you’re deviating from your trajectory (again, except for take-off and landing), you won’t hit a tree, a house or another car just because you’re a few meters too much left or right. Probably the only cases where flying is difficult is in bad meteorological conditions or if something goes horribly wrong, like a total loss of power just after take-off, forcing you to land on a river in NYC…
I believe most airliners today have the capability to “autoland.”. That seems like the most complicated part.
You are on to my point. Flying is pretty simple yet we don’t do it unsupervised. How can we expect to do it with cars.
Going from 0% to around 50% autonomy is something that just about any competent team of computer scientists can do within a year. Getting to 80% starts to separate the teams somewhat and the great teams can get there in another 1-3 years. This is about where most autonomous driving is at now. The last 20% will in my computer science educated/professional mind will take another 10-20 years if I were to estimate how far we are away right now. The point being that the process gets exponentially more difficult and the broad strokes are more or less “easy” to tackle but the millions of edge cases, terrain differences, weather conditions, and combinations of the three(and other things that I’m leaving out and stuff I don’t know that I don’t know) are all still needing to be solved. I don’t agree with Teslas hardware approach at all. It’s like let’s use machine learning multi processing and all the other things that computers do well, but when it comes to the ONLY data source for all of this let’s use something(webcam) that is inferior to eyes in almost every way and not use any other sensors that could give the car super vision at all. Yes I know it also takes a human drivers input into the dataset but splash mud on the cameras while the autonomous cars driver sleeps in the driver seat and expect them to wake up and take control seamlessly- not happening. Cruise seems to have had the better approach with hardware and computer vision, and it would appear the MB system is level 4 in a limited capacity (I don’t know much about their system tbh) Anyway to bring my opinion/rant to a close I am predicting that meaningful widespread autonomous driving is not going to happen until 2035-2040.
I’m not the expert on the topic, but to me it seems like the only chance of getting it sooner that your time frame was if we (and by we, I mean the automakers that have been heavily investing in the tech but also the regulators) had started rolling out V2V communication 5+ years ago. There are too many people out there who aren’t driving predictably for too many reasons (mostly because they suck at driving). The LIDAR systems are really good at picking up everything going on around them and the more data they have about the other vehicles around them the less they need to figure out. Meanwhile Tesla’s system is shitty at seeing things and shitty at parsing the data, but yeah sure they’ll have robotaxis out next year.
To continue the rant and comment on my own comment – while full autonomous self driving is a moonshot requiring huge breakthroughs, V2V and V2I are relatively much easier, cheaper, and can still be massively helpful in improving safety. Even basic emergency braking systems, which they are trying to make mandatory on new cars, would be massively improved. We could have been rolling the tech out for years and seeing benefits, which the autonomous driving could then piggyback on.
Yes, but every company that rolls out V2V will also use the tech to simultaneously roll out Vehicle to Anyone Who Will Pay Us For Data Including Your Insurance Company (V2AWWPUFDIYIC), get caught sharing all of that without consent, cost consumers money, have to backtrack, and ultimately further sew distrust in the whole enterprise.
We always say that self-driving will eventually be better than people b/c people suck (and they often do…), but we tend to forget that sucky people also become CEO’s, computer programmers, etc. I’d rather make my own mistakes than be the victim of someone else’s, especially if it involves no longer having a steering wheel, brake pedal, etc.
The big problem is that there are too many unknown unknowns. The driving variables are infinite.
I recall the DARPA challenge and how awesome it was when a truck made it through the desert. Seems like little progress has been made since then.
Well if they finally see the error of their ways any chance they might go to the GMC core business and continue to make full size vans with the complete range of business bodies? Ambulances, box trucks, etc.
$1bn essentially thrown into an incinerator every year. I don’t need to read out the stats on child poverty in this nation for people to understand how fucked that is.
Uh, that’s not how it works at all. That $10 billion was what the company spent on a failed project. The money didn’t get incinerated, it ended up in the bank accounts of employees, contractors, and suppliers who in turn transferred it to other people’s bank accounts and so on. Spending $10 billion on a failed project that employed a lot of people and bought a lot of stuff probably did far more to alleviate child poverty than just giving it back to shareholders.
If you are a GM shareholder, sure it sucks, because that’s money that could have been yours. But corporations losing money doesn’t magically increase child poverty, frequently the opposite actually.
Would have been better off spending the money on something that actually contributed to the bottom line – new production models, marketing, a strategic acquisition overseas to grow the company and reduce dependence on just North America and Red China, etc.
For sure. But that’s how investment in developing technologies works. Sometimes you hit something, other times you don’t.
Same as with any of the things you suggest. They could all fail. New model could go up in flames and they can’t just axe it, it has a tail on it so now their supporting a failed new model for 10 years along with reputation hit. Strategic acquisition on the heels of their sale of brands and capabilities overseas because they were dragging the company down… sounds risky too. Nothing is certain unless you are looking back with perfect hindisght.
I’m interested to know what technology they got out of the robotaxi development that could be used in other ways in their products. It’s probably not all a loss.
I still think this course of action was pretty obvious between $2 billion and $4 billion ago and GM kept dumping money in beyond the point where there was a good chance of viability
Cruise was run independently like a startup. And those types of organizations are always overly optimistic at the prospects of success. they have no choice otherwise they are out of a job. But instead of hitting a funding wall and folding (which is usually what a true startup would do) they were able to continue to drain money from GM. To your point, probably longer than they should have.
It could also be a Batgirl/Scooby Doo/Coyote vs. Acme situation.
Those were three fully completed movies that actually screened well. The studios, however, preferred to cancel all of these completed movies before they came out so they could write off a loss of $30 million so they would have a lower tax obligation and/or higher profits.
Fun fact: The live action-cartoon hybrid (think Who Framed Roger Rabbit) “Coyote vs. Acme” (starring John Cena and Will Forte) was replaced by “Barbie” for theatrical release on 7/21/23 before later removing it from its release schedule entirely.
I completely agree, but 1B spread across a US with 350M residents doesn’t go a long way if even looking like a token effort.
Wouldn’t the majority of that money be used to pay salaries? It would hardly be burned.
Child poverty? Last time I checked, there isn’t a single kid in the world with their own bank account that has 6 months of living expenses saved.
A tragedy, indeed.
I thought Cruise was doing reasonably well on the operation front. Is this another one of those “GM snatches defeat from the jaws of victory” things they like to do?
Like I’m positive they’ve almost gotten away with several good ideas that take some number of years to pan out, but the execs get cold feet at the last minute.
After the lies/nontransparency they were caught in with the dragging incident, one wonders what lies or things are we not aware of in relation to the business or concept? Cutting bait may be the best option. The good stuff gets publicized, the other stuff not so much.
Well, now we know that self-driving services are going to take off. When GM abandons an idea it usually turns out that it works for other companies. EV1, the Bolt and the Volt all come to mind.
GM even had the EPA cert for the Origin already. With all those mothballed examples, it’s evident their release on the roads was imminent.
If Cruise had just responded to the pedestrian incident quickly and appropriately, I can only imagine where they might be now. Because just from the eyeball test, they were probably the closest of the major players to actually pulling it off. Maybe Zoox, but that remains to be seen.
Waymo regularly ferries passengers around downtown SF (including limited freeway use), and a few other cities. Zoox is also sending out mailers in downtown SF to advise us that they’re soon going live also.
I’m guessing that GM is mostly looking at their competition and deciding that it isn’t worth the cash burn to go enter a relatively crowded market.
Waymo also regularly causes traffic jams because their cars can’t get out of each others way.
It will, but for Tesla only as Musk will grant himself federal approvals. The occasional pedestrian being dragged will be just small eggs being broken to make a nice omelette.
I remember when $10 billion could create a whole new division, or buy a whole car company from Cerberus. Wth were they spending $10 billion on for some Chevy Bolts with extra cameras? And then another billion a year on maintenance?
Nissan’s market cap is $9 billion right now, just saying
Yeah, but nobody wants Nissan.
Of course not. But, it still would have been a better use of $10 billion, I mean, Nissan and Infiniti at least have a combined 5.8% market share, add that to GM’s 16.9%, and they’d be only 2 points away from where they were before the Great Recession. /s
So you mean like instead, we could have “Uranus. A different kind of Saturn.”
“Uranus. Show us yours.”
“Venus. It starts with a ‘V'”
“Jupiter. It’s Big”
It would have been amazing if GM rebranded all of their divisions after planets as their bold 90s stroke.
Would the models be named after the different moons of Uranus? Like Puck? “Nothing feels as tight as a Uranus Puck!”
While I once aspired to build my own cars, I have never owned a company!
I was wondering about this, glad you chimed in, as I was about to ask if you owned Opel and Vauxhall at this point. I can’t keep track of these things!
Gee – It’s almost like level 5 self-driving is just a hype machine to pump up stock prices…
Self driving stock price
I mean that’s true basically only for Tesla, where it is absolutely a pump and dump scheme, especially given they somehow think they can achieve self driving with consumer grade webcams.
The sad thing are all of these companies who actually tried to do it in earnest and have discovered it’s a fools errand. They would have been better off half-assing it like Tesla and promising it forever but never delivering.
Tesla’s flaw was trying to mimic human drivers. I think he figured people drive with only their eyes, and computers are smarter than people, ergo all the car needs is cameras and a computer.
The thing is we don’t drive only with our eyes. Our skin is loaded with g sensors, our ears are a gyroscope, and sometimes smell and sound give us inputs critical to safely operating a vehicle. Not to mention the three and a half pound super computer between our ears that excels at the kind of parallel processing a piece of silicon using Von Neumann architecture could only dream of (assuming the silicon could dream).
Not to mention things like intuition, experience, and common sense.
Tesla seems to think everyone just drives with their eyes and that things like moving your head, shielding sun from your eyes with your hand, depth perception and proprioception don’t exist.
AVs OTOH have the option for V2V communication well beyond the car just ahead, not just basic V2V but where they plan to be in the near future. They can also keep better keep on roadway conditions, weather via external sensors and don’t ignore traffic laws because they are inconvenient.
AVs also don’t hit and run, DUI, try to impress others, hoon, sideshow or have a beer to hold.
Fair on all counts. And that would be great if we could ban all existing vehicles that don’t have that tech, or wave a magic wand and put that tech in all existing vehicles. That’s not going to happen.
I think it’s possible. At the bare minimum, direction and speed are all that are needed to communicate to the herd. Perhaps location? There is no reason that the V2V system would require anymore, for instance like control over the vehicle. As such, a new head unit that could communicate what is going on is all I would need in my old truck. I could respond if I knew something was going to happen. If other vehicles around me were automated, it’s possible to “make space” for my non-computer controlled response times. It wouldn’t be perfect, of course, but until we are a fully automated society, it won’t be perfect.
This would help a bit, but I don’t know that it would be enough. It could notify the “neighborhood” where it is and what speed and direction it is currently traveling, but not necessarily what it’s going to do next.
Adding blinker status could help, assuming they were properly activated by the driver (insert obligatory BMW joke here) but even then that might not be all that reliable. It’s not uncommon for me to accidentally toggle mine when flipping between high and low beams, for example.
I wonder if that would trigger some sort of a panic stop by an AV heading towards me on a two lane highway, due to it interpreting my accidental bump of the turn indicator as an “oh, shit, this idiot’s going to make a left turn right in front of me!”
I don’t see why an AV can’t determine the difference between an accidental and intentional turn signal exactly the same way you do: duration and position. Even better if the car in question sends a message that its pilot is looking in the mirrors and blind spots.
I believe that will effectively happen eventually, if/when the tech is fully proven and mainstreamed. Could just take the form of non-self-driving cars becoming prohibitively expensive to register and insure based on risk models, which would make them unaffordable for 99.9% of vehicle owners and remove them from the roads
Start with AV only traffic lanes and gradually expand those lanes as AV capable vehicles become more commonplace.
Once you limit routes, you have gotten something worse than a bus.
Who said anything about limiting routes? The AV only lanes would be akin to carpool lanes.