Having worked for GM I can say, one of the best things GM did was after they bought Cruise they kept it as a separate company and away from GM bureaucratic style of management where everything comes to die.
When an experimental project is made core to the company's traditional business, the experimental project is usually slowed down. The fewer dependencies between GM cars and Cruise, the better.
Vast majority of accidents are caused by drivers (or cyclers, pedestrians) who fail to safely engage in traffic. For the most part, cars don't kill people - people kill people (or, get themselves killed).
which is exactly why removing people out of the equation is the best way to solve the problem. even fully educated and hyper responsible person is not immune to judgement lapses, hallucinations, heart attacks, strokes, losses of consciousness and many more modes of failure.
what's worst - even if i'm 100% sure about myself, even if i take myself out of the equation - there are still thousands murderous humans driving around me.
Self-driving's problem isn't that it 'doesn't work in many use cases'. The problem is finding a single use case where today's L4 technology can actually be used given the technological, legal, PR and cost considerations.
Exactly - it's more about recognizing the shift that will happen if the technology can work and what that will mean.
GM seems to think it's important since they bought Cruise in the first place, but instead of seeing failure on this as an existential threat it seems more like a side project for them.
From the outside this seems like a bad strategy and will only succeed if it turns out self-driving is impossible (which seems unlikely).
Digital cameras were toys until suddenly they weren't and then Kodak died - even though they had invented the digital camera in the first place.
It's not enough to have the tech, you have to prioritize it in the business. Xerox PARC is another example of this type of failure.
Super Cruise, GM's Autopilot competitor, is planning to be fleet-wide by 2020[1]. But that is L2/L3 stuff, which is different than the L4 technology Cruise is focused on. You can't put L4 into GM's cars because it's not ready - it's probably dangerous and I'm doubtful it would be cost effective given the prices of sensors and the fact that you need to turn the car into a rolling datacenter.
While owned by General Motors, no. But with Softbank buying in, maybe GM is backing away from Cruise.
Is this another the Saudi Arabia sovereign wealth fund investment via Softbank? They seem to have become the world's largest source of dumb money.
Is there anybody, anywhere, yet who is deploying self-driving cars without "safety drivers" or insisting the driver remain active at the wheel? Cruise, no. Waymo, did for a while, then backed off. Volvo was supposed to be deploying by now but backed off. There are a few self-driving shuttle buses, but they're really slow, like 10mph, except when on dedicated pathways.
Companies that are claiming they will deploy in volume in the next two years are bullshitting. Nobody has Level 3 self driving working reliably in test yet. Let alone Level 4 or 5.
For all of the criticism that Tesla gets for a incremental approach - they still have the best semi-autonomous car in actual production and a difficult but obvious path to full autonomy.
These guys could completely flame out without ever getting anything of value out into the market.
I have a decent sense of where Tesla and Waymo are at in the SDC development competition, but where does Cruise stand? What kind of sensors, computation, vehicles, where are they testing? In my limited knowledge, I'd guess Waymo is ahead of these other players, but do we know who is "in second" or who is second-closest to rolling out actual fully autonomous services?
By disengagement reports cruise is second behind Waymo and these two are way ahead of everyone else. They are testing 160 or so Chevy bolts in California it seems like.
It is a mistake to use that report as some type of ranking due to it only measuring disengagements on public roads in California. There are some big names on that list that simply have no data like Tesla, Ford, VW, and Lyft.
Tesla isn't there because charlatan can't produce reports that would look good on Tesla. If it were he would be tweeting about it.
If a California based company can't employ 4 dudes driving around California to provide this report (which is kinda important to get regulatory approval), you have to wonder how bad Tesla really is.
The latest noise by Musk about Tesla's FSD is purely to raise money and take away the focus that he can't produce cars profitably. Nothing else
Interesting to call Elon Musk charlatan considering he put an actual electric car on the road that is on its way to sell 400k units in a year and kicking other non charlatan like BMW and GMs ass
They sold 400k cars with hardware they believe is sufficient for full self driving (perhaps with limited upgrades).
Tesla is the only one actually attempting to empower individuals with FSD tech. Everyone else has given up and/or wants to monopolize it for private fleets. Everyone claims to be shooting for the same FSD capability. Different strategies is, of course, wrongly conflated with "fraud."
I tend to think Tesla is behind Waymo. It's hard to say beyond that. But for actual mass deployment of serious hardware, no one is ahead of Tesla.
Charlatan in 2016: Teslas built right now has all the hardware required for FSD
Charlatan in 2019: The previous version by nVidia was such garbage (processing 1.5 fps, really?). Teslas built right now has all the hardware required for FSD.
Charlatan, a few minutes later: We are working on the next version of FSD hardware (that'd most definitely render 2019 obsolete)
There is Musk, the sometimes engineer, then there is Musk, the charlatan. An Engineer would never claim 'the latest hardware will be the greatest ever'
What I find interesting is that Nvidia claimed on their website that their hardware could do self driving. That’s the same hardware that Tesla has used and has to upgrade to the new Neural Net chip. Do you think Nvidia is going to pay for the hardware upgrade Teslas need? Tesla is doing so. They believe the current version of the NN is good enough for FSD, but that doesn’t mean it can’t be improved.
What I find so fascinating about this discussion is how Tesla is necessarily the most transparent player yet everyone else (with the same stated FSD goals) is given a pass due to secrecy. I suppose that means secrecy works.
What do you mean Tesla is the most transparent player? They absolutely never showed anything else than slides and a 2 minute video.
That's probably the least transparent from all the big players. You can go to Mountain View and see cars running right now. Or go to Phoenix and get into a Waymo car.
Again you got ahead of yourself with the Tesla hype.
Learning a problem is harder 3 years later isn’t unusual in software world. All the autonomous car companies said this is harder. Waymo has no driverless cars anywhere and there are quotes saying it would be ready.
Also, Elon always promised hardware upgrade if needed.
We're talking about the holy grail of AI in the automotive industry, not a SaaS side project.
This isn't assigning the Summer Vision Project in 1966.
Everyone knew that this was going to be a monumental problem, and the only person in the room who was claiming they already had all the tools to solve the problem (a problem that no one else is even willing to say they fully understood how to approach yet) is also the guy with a record for lying through his teeth so much there's a term for it.
I'm excited to see real attempts at FSD. Not from the company that gets most of it's hype from AP1, which used the same sensor suit GM was putting in cars years ago for off-label applications so they could go around waving "autopilot", which resulted in easily preventable deaths.
iPhone in 2008 had all the hardware to run an iPhone. iPhone in 2019 also has all the hardware to run an iPhone. You don't want to Tesla to continually improve their hardware ?
> Tesla is the only one actually attempting to empower individuals with FSD tech. Everyone else has given up and/or wants to monopolize it for private fleets.
There's a guy in a village next to me that is "attempting to empower individuals" with energy medicine (healing people with the touch of his hands). Everyone else, including medicine researchers, has given up on this.
Tesla claim they have all the hardware for FSD, but FSD is still science fiction. It literally doesn't exist. Waymo is closest but they use Lidar, which Tesla's aren't equipped with.
Maybe Tesla will work it out and make a great self driving car on cheap hardware. I'll believe it when I see it
Plenty of people pre-ordered No Man's Sky - then were upset when they received less than what they thought they'd been promised. That's the risk when you pay upfront for a product that doesn't exist.
Maybe people who've paid for the Tesla FSD package will receive what I imagine when I hear "full self driving". Or maybe, like No Man's Sky backers, they'll get something they're less happy with.
Musk has his head in the clouds. While this is good for his long-term plans to do the impossible, it means you can never believe him on progress or dates until you can touch the product.
Waymo and Cruise taxies have to operate in all conditions before they can roll out general purpose taxies. E.g. if you hail one and on the way home, there is construction and a cop directing traffic, you can’t take over and drive.
With a Tesla, they can deliver incremental value, without having to be able to handle every possible scenario.
Also aren’t Cruise and Waymo in same boat touting progress to raise money?
This is my feeling also.
Tesla is absolutely nowhere, but they keep making noise and announcements. And what is amazing is that it works. Most people think Tesla is up there with the other self driving car companies
Fine. They are at the point where they can eventually record a video on a fairly easy, almost empty, perfect condition, predefined suburban track. Noone knows how many attempts that video took.
This is the point that Waymo reached in 2012 if I remember correctly.
I don't know the details of the law, but I would bet that Teslas don't qualify on a technicality of the law rather than the fact that Tesla simply has not operated an autonomous vehicle on a public road in the state of California. You might be right that they are doing it to save embarrassment, but that is a rather cynical mindset considering it can also be explained by just not wanting to publicize their progress.
It is also worth pointing out that Tesla's ability to run software on its fleet of hundreds of thousands of cars gives them a huge advantage regarding data collection. If their cars are truly as safe as they claim (a huge if and one we have no real public proof of yet), the lack of having this report on disengagements in California is not going to be a stumbling block in getting regulatory approval.
EDIT: Since I am being downvoted, I looked up the regulation [1]. Teslas likely don't to qualify as "autonomous vehicles" according to definition b. Although that is a pretty bad definition in my opinion. Also I am not praising Tesla here, just pointing out that their ability to avoid this regulation is not an indication of the quality of their self-driving tech.
Considering that Elon Musk has never once failed to publicize even imagined progress (that would simply be par for any other business), it would boggle the mind to believe that he would be keeping any type of market-leading progress a secret, especially in the market where he has literally bet the future of the company and this type of news could encourage competitors to drop out of the market.
the lack of having this report on disengagements in California is not going to be a stumbling block in getting regulatory approval.
Actually, if Tesla hopes to legally sell self-driving cars in CA, that is going to be a huge stumbling block.
How many businesses volunteer themselves to be subject to regulation that doesn't apply to them, especially when that regulation gives their competitors insights into the company's R&D capabilities?
Tesla has sold over 50k Model 3s in California in 2018[1]. Regulators would be ridiculous if they put more value in the data from 111 Waymo vehicles compared to the data from tens of thousand of Tesla vehicles.
And once again, I am not saying Tesla is leading here or has superior technology. But people like you are letting their bias against Musk and Tesla cloud their judgement on this issue.
So you are saying instead if running the required test vehicles with trained drivers in California like Waymo does Tesla is running unsupervised Beta tests with 50k cars? For self-driving? Not sure if that's how new cars are certified...
Charlatan is an appropriate word for Musk but in a limited sense. He is surely making false claims about Tesla's self driving abilities which clearly border on fraud. But Tesla car itself is a very good car and has been selling like hot cakes.
I think Tesla is a very good EV. Quite possibly the best EV on the market. But it's pretty pathetic as a car, especially as a car at the price they ask for it.
I'm sure it's fast... But something that makes you look to the side and down to even see what it's doing (and uses touchscreen) not only isn't a good car, people who designed it should not be allowed to put crap on public roads in the first place.
Hell, the thing doesn't even have cooled seats you get it a Hyundai or something.
Nope. Great EV. Vastly overpriced golf cart as a car.
That's of course a separate can of worms, but I too never understood why people buy electric cars in their current form. Imagine you had a laptop that can last a month on full charge, vs your typical laptop of today that lasts a day at best - people would obviously choose the former. And yet, in cars, people choose the latter, and pay extra for the privilege. [1]
[1] Except that situation with cars is much worse than with laptops, as you can continue using a laptop while it's charging.
I can see people willing to put up with inconveniences for the sake of supposed environmental reasons, and some people just like the shiny new thing with touchscreens and computers and Elon's reality distortion field...
What is funny is when they try to convince people that a Tesla actually is a luxury car, which it quite obviously isn't. Model S lacks some meaningful amenities that a Hyunday at half the price has, build quality would make a Yugo gag, and if it does break down or gets in an accident you might be waiting months for it to be fixed. But I guess it's nice to convince yourself that there were some reasons you've paid $60K for for something less practical than a Yaris...
Nobody knows for sure since nobody is releasing anything meaningful (the disengagement reports are both voluntary and self-policed, with no standard for what constitutes a reported disengagement), but the general consensus is that Cruise is second behind Waymo.
Reporting the number of miles driven and the number of disengagements is required by law in CA. It is not voluntary.
The reason some larger companies don't have reports is because they either didn't do autonomous testing in CA (many test in AZ) or that they do not consider their tests "autonomous" ( level 4 or above). This is the case for Tesla autopilot.
Adding on to this - the self-driving car space is pretty small (in terms of companies with realistic self-driving aspirations) and there's a good amount of movement of employees between companies. While it's hard to make judgements from the reported numbers, the consensus in the space is definitely that the leaders are Waymo followed by Cruise.
This comment should be rated higher. It's exactly true. The data are self-selected and self-reported, and the specific criteria for what qualifies is considered a trade secret.
Anecdotally I see more Cruise cars in SF and more Waymo cars in Mountain View. I heard Cruise doesn't drive on freeways either, and that would affect disengagements per mile quite a bit.
Still Cruise's disengagement rate is astonishingly high: 1000 disengagements every 0.19 miles[1], according to the Wall Street Journal ;-).
That statistic is backwards - they have 0.19 disengagements per 1,000 miles, so roughly 5,000 miles per disengagement. But as previously mentioned, this is a wildly disingenuous number, literally orders of magnitude away from the reality of how their cars are performing on the road today.
$20B valuation for companies without products is a very, very frothy signal.
This kind of thing is waaay beyond what went on during the dot-com bubble, where there was a lot of value being created without any business model either.
Curious about the corporate structure behind this. Is Cruise a subdivision of GM or is it an independent company with GM just being majority shareholder?
1. Stability for the project. A self-driving R&D project internal to a larger company risks getting killed over the N years it takes to do the work without any payoff.
2. The investment probably looks better on the parent corporations budgets. They aren't sinking billions of dollars into R&D, they're investing in a third party which is sinking billions into R&D. The shares might even look good, if non-liquid, on paper.
3. Segments liability away from the larger company. When all self-driving cars in the world make a left turn at the same time due to an obscure calendar bug, it's just GM Cruise that gets sued out of existence.
Self driving right now is a moonshot for many of these companies, and competing with the resources that Google can throw at (both in terms of AI talent as well as sheer money) can be tough. Partnering with GM allows Honda a piece of this, allows for future adoption, and consolidates both of their revenues. If Cruise does become successful, the thought process might be that both Honda and GM obtain the tech therein, and differentiate themselves via other methods. Imagine this to be a new engine type that is too costly for just one manufacturer to research, and so they partner up with the implication that both get usage rights once the engine is done.
I work for Cruise. For safety, our drivers keep their hands on the wheel at all times, including when the vehicle is autonomous. The vehicle you saw may have been in manual mode, but seeing the driver crank the wheel usually isn’t a good way to tell.
I wonder how softbank’s vision fund has been doing, and whether that info is available somewhere. Seems like they’re in on every late stage splashy tech investment
Isn't Lyft also essentially a bet on autonomous driving? With (probably) an even shorter time horizon to burn all their cash? Otherwise it's hard to see them making enough profit to justify a 17B valuation.
Yeah right. But other than self driving cars I fail to see how they'll un-dig themselves out of the money losing setup before burning through all the fresh cash collected in the IPO. Hopefully I'll be wrong.
I see this criticism about Lyft and Uber all the time, but I don't understand how both companies could not at some point simply raise their prices to make the unit economics work. Sure, they'd lose some business… but given there is almost no fixed costs to running their business, they'd still have large profitable businesses. The only reason they don't is they're fighting for marketshare with each other while in growth mode. But once both companies have established their places in the world, saturated their markets, I don't see why they wouldn't simply raise their prices accordingly.
Revenue would drop, profit would rise. Eventually, the market reach an equilibrium.
Note they could also just raise the share they take for themselves and prices simultaneously. If prices go up 20% and they up their fee from 20% -> 33%, their income goes up 60% and their unit economics look a LOT better.
Well, up to now the mainly compete on price and booking comfort (apps) with traditional taxis. So taxi prices are somehow giving an upper price limit, maybe plus something for the better CX.
If they invented a high temperature superconductor that could ease their money problems as well, but neither of these ideas have anything to do with their current business model.
