Slashdot Mirror
Tesla Autopilot Crisis Deepens With Loss of Third Autopilot Boss In 18 Months (arstechnica.com)
An anonymous reader quotes a report from Ars Technica: It is no secret that Tesla's Autopilot project is struggling. Last summer, we covered a report that Tesla was bleeding talent from its Autopilot division. Tesla Autopilot head Sterling Anderson quit Tesla at the end of 2016. His replacement was Chris Lattner, who had previously created the Swift programming language at Apple. But Lattner only lasted six months before departing last June. Now Lattner's replacement, Jim Keller, is leaving Tesla as well.
Keller was a well-known chip designer at AMD before he was recruited to lead Tesla's hardware engineering efforts for Autopilot in 2016. Keller has been working to develop custom silicon for Autopilot, potentially replacing the Nvidia chips being used in today's Tesla vehicles. When Lattner left Tesla last June, Keller was given broader authority over the Autopilot program as a whole. Keller's departure comes just weeks after the death of Walter Huang, a driver whose Model X vehicle slammed into a concrete lane divider in Mountain View, California. Tesla has said Autopilot was engaged at the time of the crash. Tesla has since gotten into public feuds with both Huang's family and the National Transportation Safety Board, the federal agency investigating the crash. "Today is Jim Keller's last day at Tesla, where he has overseen low-voltage hardware, Autopilot software and infotainment," Tesla said in a statement to Electrek. "Prior to joining Tesla, Jim's core passion was microprocessor engineering, and he's now joining a company where he'll be able to once again focus on this exclusively."
Keller was a well-known chip designer at AMD before he was recruited to lead Tesla's hardware engineering efforts for Autopilot in 2016. Keller has been working to develop custom silicon for Autopilot, potentially replacing the Nvidia chips being used in today's Tesla vehicles. When Lattner left Tesla last June, Keller was given broader authority over the Autopilot program as a whole. Keller's departure comes just weeks after the death of Walter Huang, a driver whose Model X vehicle slammed into a concrete lane divider in Mountain View, California. Tesla has said Autopilot was engaged at the time of the crash. Tesla has since gotten into public feuds with both Huang's family and the National Transportation Safety Board, the federal agency investigating the crash. "Today is Jim Keller's last day at Tesla, where he has overseen low-voltage hardware, Autopilot software and infotainment," Tesla said in a statement to Electrek. "Prior to joining Tesla, Jim's core passion was microprocessor engineering, and he's now joining a company where he'll be able to once again focus on this exclusively."
LIDAR is the way to go and both Google and Apple know this.
The problem is, puck-sized LIDAR systems, as seen in 8-packs on the Apple dev car, cost 8000 a piece and that is why Testa uses cheapo-cams and parking radar.
In case anyone is wondering why you're seeing so many stories like this about Tesla all of the sudden, here is your answer. In particular, this chart.
Tesla is the most shorted stock in the US right now. There is literally no company in the US that more people have a financial interest in seeing fail than Tesla. A third of the stock is in short positions. The problem is that this is incredibly dangerous from the perspective of a short squeeze. Shorts hold the stock price down - the massive surge in short selling countered the benefits (from a stock price perspective) of the major increase in Model 3 production rates. But this can only be taken so far; it's not like they're going to be able to short 100% of the stock. If Model 3 production continues to rise like it's been doing - and along with it, the stock price - not only will some short sellers want to liquidate, but others will be contractually forced to liquidate. This is done by purchasing an equivalent number of shares of TSLA to cover their short. This purchase in turn raises the stock price. With such a massive percentage of Tesla shorted, this can easily snowball, where the obligations of some shorts to purchase cause the next to be forced to purchase, and the next, and so forth - all purchased at whatever price Tesla happens to be at the moment. The shorts would need to acquire literally 1/3rd of Tesla's stock in a short period of time.
Needless to say, this would be a financial disaster for them. If Tesla underperforms what the market expects of them, longs lose some money. But if Tesla overperforms what the market expects, shorts lose a huge amount of money. It's highly asymmetric.
So in case you were wondering if it was a coincidence that all of the sudden everyone and their cousin suddenly started bashing Tesla in the news - even for something as mundane as another company poaching talent from Tesla - no, it's not a coincidence.
No matter how kind you are, German children are kinder.
I suspect Tesla's method of using less hardware will be the main path in 15 years for autonomy, once we have car to car communications and car to traffic control communications as standard equipment in every vehicle and bugs worked out. Some cars with humans or lasers can communicate safe passages and routes through construction, and other road conditions (wet/ice/snow...) And lesser equipped cars can then navigate recently validated routes more safely.
But now the NTSB is very likely going to step into national standards for autonomy, and it doesn't appear Tesla is ready to meet the likely minimum standard, such as redundant navigation (operate without GPS, or without optical recognition.) and redundant systems.
Hardware only gets cheaper, the future of self-driving cars is more and redundant sensors. And no car is going to rely on another car to tell it what is a safe route.
I stole this Sig
In case anyone is wondering: all of this "Autopilot is doomed" stuff comes out at a time when it just had one of its most massive updates in its history (unfortunately, it hadn't significantly rolled out before the fatal crash in California). It no longer filters out stationary objects, it handles roads with unusual lane widths, two-direction roads with no central lane markings, and will deliberately "break the rules" when needed for safety (for example, driving into the shoulder when a truck is about to hit you, or when the normal "rules of the road" have suddenly changed.
But of course, you're not going to see a million articles about that because that's not the obligate doom-and-gloom.
I consider myself a self-driving pessimist. I think there's far too many rules that we process, with too complex reasoning, for self-driving to be immediately around the corner. I have slowly been becoming more optimistic with the realization of how much more one can enable a car to "see" than humans (for example, using radar brightness at different wavelengths to determine road smoothness / traction conditions ahead, or past altimetry data to determine the depth of water on a road), but still think it's going to be a long time before full self driving becomes mainstream. But I also believe that, if properly implemented, combined human-computer systems can be much better than either alone - with the computer bringing new senses and "constant attention" to the picture, and the human bringing their brain. The key is ensuring that the human pays full attention. Making them regularly torque the wheel is one thing, but even better looks to be where the tech is headed - eye tracking. With eye tracking, they can't stop paying attention to the road. And I can't see how in such a situation that "human + computer" is not better than "human alone".
No matter how kind you are, German children are kinder.
Also, the NTSB does not set standard for autonomy. The NTSB is an investigation board. The NHTSA sets standards in the US.
No matter how kind you are, German children are kinder.
If you believe this, jump aboard the short train. There's a nice echo chamber over at Seeking Alpha to assist you in parting with your money.
No matter how kind you are, German children are kinder.
Humans seem to get by ok with minimal hardware. Two cameras and one massively parallel, but extremely buggy and unreliable CPU that completely fabricates a lot of what you see based on its best guess.
But, all the same, I wish they never got into this auto pilot self driving thing. It is a distraction from getting the affordable electric car done. That is the most important thing to get done.
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
... with several billion years worth of AI training data
In 15 years?
The average car age on the road USA is approaching 12 years. in 2003 is was 9.7 years, in 2016 it was 11.6 years.
In 15 years the average car will be similar to what's available now. Most cars won't have self-driving capability or car to car communication.
There are plenty of cars from the 70's, 80's and 90's on the roads today. In 15 years there still will be. Maybe fewer from the 80's as they start turning to rust.
Maybe in 150 years.
PE ratios are great for a company with relatively static business operations in a relatively static field, but they're terribly misleading for growing companies. Ex: in late June of 2014, Amazon had a PE of over 500. Was it a bad buy? Hardly; people who bought it then would be 4 1/2 times richer today.
PE ratios don't cut it for growth companies that are dumping tons of money into capex, expanded options for future sales and building a new market. Quite the opposite , PE ratios paint precisely the *wrong* picture. In such a case, they're more a measure of how aggressive the company is being toward growth.
For such growth companies, the real question is how large of a market they're aiming for and how likely they are to get there. E.g. the reason Tesla shot up so much after Model 3 preorders opened is that nearly half a million people put down real money to wait over a year for a BMW-priced electric car that they'd never even seen in person, much less driven, without any advertising . That's pent-up, untapped demand. That's a market. Tesla is priced based on the potential shown, relative to the risk in getting there. Its current profits are almost irrelevant except for helping fund its route to "there".
One can disagree about the exact market size and the risks. It's expected! But what should not be in dispute is that PE is totally the wrong tool to employ here.
No matter how kind you are, German children are kinder.
As Buffett once put it, the market can remain irrational longer than you can remain solvent. The sure thing is that eventually investors will get tired of Elon's sweet crooning and flashy distractions and will stop giving him more money to throw after the billions he's frittered away. The unsure thing is exactly how long that will take. But I'm heartened at the (dare I say) S-curve of reality that finally seems to be setting in.
I don't believe Tesla deserves a $48B market cap-- worth more than Ford's $45B and close to Honda's $60B.
Tesla hopes to sell 500,000 vehicles this year, and 1,000,000 next year, both with a net outflow of cash. Tesla is also significantly encumbered by debt and will likely have to dilute existing stockholders more. Honda sells 5 million vehicles a year with an operating profit plus all kinds of other "stuff".
Yes, electric cars will be important, and ultimately TSLA may move as many vehicles and deliver as much profit as Honda. Maybe. But-- how much of that upside will current stockholders enjoy (vs. new stockholders and bondholders), and how long will it take to reach that point (time value of money)?
if AI drivers perform the same, or even half as bad as humans, they would be forbidden.
what I am saying is : humans are bad drivers
And what everyone else is saying is: Tesla's "auto pilot", Uber's full-fledged self driving cars, etc. are worse. Much worse. They're parlor tricks that work in a rigged scenario at best. They're decades away from being as safe and flexible as even a drunk human.
And no car is going to rely on another car to tell it what is a safe route.
That's actually exactly what Autopilot does. In traffic it follows the car in front. The lead car turns blue on the display when the Tesla is locked on to it.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
Tesla is not really "significantly encumbered with debt". There's 200-something million due in November, and next spring there will be another round due that brings the total up to $1,1B. 500k vehicles with an average sale price of $45k is $22,5B. At 25% margins, it's $5,6B. Knock that down by SG&A, R&D, etc - the debt is not at all disproportional compared to the business activities it was acquired to fund.
Tesla has no plans to stop at the vehicle numbers you state (although your timing figures are outdated - the slow Model 3 launch delayed some of the timing). Model 3 is to peak at 500k without advertising, 700k when Tesla starts advertising in the future. Model Y is expected to be 700k without advertising. No figures that I'm aware of have been announced for their Semi expectations, but let it suffice to say that freight vehicles are a very large market in their own right. After the Y Tesla will be launching a pickup, which is a huge market. Etc. Tesla is not stopping at the Model 3.
Additionally: Honda's 5M vehicles per year are nowhere near $45k average sale price.
As for "how long": Model 3 should be hitting "around 5k/wk" at the end of June (they're trying for 6k then, but a miss is expected, even by them). The line should be tuned to 6k or higher in Q3 and potentially Q4. Tesla plans to duplicate the 3 line, but the delays and the good news on the China front make it increasingly likely that they'll duplicate it directly into China to have local production for the asian market, rather than duplicating at Fremont. Gigafactory, after having paused in size for nearly a year, is as we speak undergoing a significant expansion; they're expanding the base to the full target width and they're building a new peripheral parking lot so that they free up the current parking area for building expansion.
Y is to be announced this summer. I anticipate that they want to have Model 3 production up to ~5k/wk before announcing it so that they don't get accused of trying to draw attention away from Model 3 production rate issues. Expect at least half a million reservations for it within the first couple months. Model Y is built on the Model 3 architecture, so it's not going to take near as long to go into production (Model 3 is a brand new architecture). I expect about a year, maybe 1 1/2, before first deliveries, with a much faster scaleup. So we're talking mid-late 2019. I expect a pickup in late 2020 or early 2021. By that point, Semi will have gone from pilot orders to full fleet orders. So we're talking maybe 3-4 years before Tesla becomes massive.
Also, the problem with comparing to other automakers production today, is that the more the market Tesla expands into, the more they take away from other automakers. So not only does Tesla become larger, but its competitors become smaller. In short, a bet on Tesla's success is simultaneously a "short" against other automakers. Right now, according to surveys, Model 3 is mainly hacking away at a market that's about an even mix of customers from entry-level luxury brands (BMW, Audi, Cadillac, etc) and customers from eco-friendly vehicles (hybrids, other EVs, etc). But as prices continue to drop, the market segments that it draws customers from will also move downmarket.
No matter how kind you are, German children are kinder.
You're talking about V2V? Yes, they have spectrum and some implementations. They don't even seem to have a single agreed to or dictated protocol that would allow, for example, BMWs to talk to Cadillacs. And no, there is no current Implement_by-date although several have been proposed. NHSTA proposed a standard at the end of 2016, but it seems to have quietly died in late 2017
Not that I'm opposed to V2V? Seems like a perfectly OK idea actually if everyone can be persuaded to speak the same digital language.
But what they have does not seem remotely what will actually be needed to support autonomous vehicles.
You can't see ANYTHING from a car, You've got to get out of the goddamned contraption and walk...Edward Abbey
Some hardware gets cheaper some of the time. I've noticed for instance that the old-fashioned mechanical watches that were the staple of the last century have not gotten any cheaper at all. Recently the price of high-end GPUs has skyrocketed.
Mass production of some goods can cause them to become cheaper but this is by no means a foregone conclusion.
How many billion miles travelled per day? The relevant statistic is the number of fatalities per billion miles (or km) traveled. See here. We don't know for sure the relevant statistics for driver-assisted vehicles.
There is a very good chance that driver assistance is going to improve the fatalities statistics, but this needs to be done correctly.