What is this nonsense? Any truth to this article?
Basicly articles in Seeking Alpha are mostly clickbait with outrageous claims to get people to look. I recommend avoiding anything you see there (and trust anything you see there even less).
I don't know about the motivation. But holes in most positions I see there.
The numbers are certainly valid, but it's difficult to know what to make of them. For example, Tesla reports only having 550 miles of testing in the state of California in 2016 - seems mighty low for as well as the system is known to operate. They report only five days of testing, four of them (representing the vast majority of disengagements) in the rain.
What does that mean about the state of Teslas efforts compared with, for example, Waymo? Heck if I know.
Seeking Alpha is all individual opinions with no expertise required to publish. They only stay in business because there are always new suckers that haven't learned yet to skip over the site. The one thing they are good at is getting their links to show up in search results.
I totally agree that Tesla's dream is a dream. Current EAP 2.0 hardware may be enough for Auto Pilot, but it is far short of Self Driving. It does not have LIDAR and only has a single radar. It heavily depends on video camera, which does not work well on evening, rain/snow/fog days and requires lots of computer power.
I am expecting Tesla gets sued for false advertisement sooner or later.
Well, Lidar doesn't work very well on rain/snow/fog days either, and frankly "lots of computer power" is ridiculously cheap to purchase these days. It'll be interesting to see if Tesla can build equivalent autonomy using it's current sensor suite. Of course, if Lidar prices come down, there's also nothing keeping them from including it as an additional sensor in EAP 3.0.
What may have been reported were the miles on a few Tesla test vehicles. And even when those vehicles had the AP/EAP/FSD software disengaged, Tesla was still collecting data.
Tesla's strategy is different from the other companies listed in the report. Tesla has cars with AP 1.0 and 2.0 hardware driving every day. Even when AP is not enabled, Tesla is collecting data from those cars that can be used for training the autopilot software and verifying the software is ready to drive safely.
Installing AP 2.0 into every car is a very clever strategy, and as more AP 2.0 cars hit the road, Tesla will quickly gain more AP/FSD data than all of the other manufacturers combined.
For AP 1.0, Tesla is likely collecting more data in a single day than the other manufacturers combined over all of last year.
Because the other companies are relying only on test vehicles - and Tesla is using a rapidly expanding number of sold vehicles for testing, the article isn't making an apples-to-apples comparison, and if restricted only to test vehicles, Tesla will likely always come out near the bottom, because Tesla doesn't need to invest as much time with test vehicles with so many purchased cars collecting data.
I totally agree. People do not have LIDAR or Radar. That is why people cannot drive cars. Why would Tesla think they could drive cars using only the visible spectrum along with some help from Radar.
+1 in terms of having only one Radar, Tesla should really have a couple for redundancy.
However you and the article are fundamentally wrong in the assumption that it cannot work. Even if the cameras do not see any better than us in low light or fog the 8 cameras still cover more area than our 2 eyes. The other sensors without analogs in the human world, ultrasonic and radar give the car a distinct advantage over us poor humans.
- Should the Tesla vehicles have LIDAR? maybe, there are other options as well, what about an infrared forward facing camera?
- Will their be an AP 3.0 hardware suite, Yes..
- Can Tesla achieve Level 5 autopilot, Yes.
- Will the car be allowed to be driverless? Unlikely, fully autonomous yes but i think governments will keep drivers in the driver seat for a while yet. If for no other reason than liability and insurance reasons.
Time for the popcorn.
I did not read the article and I never click on SA but I can probably guessed what it says. There are other news article about the 550 test miles and number of disengagement. Like bp said Tesla sure was not testing AP1, AP2 or any production cars when there are cars on the field collecting millions of miles of data EVERYDAY. Those four test cars likely were just development tools for testing new engineering designs. It's like Steph Curry gets <10% shooting at practice when he shoots from cross the court or in the tunnel. He's still the best shooter the world has ever seen when he plays the real game.
Cameras collect a lot more data:
- Can read street signs, road markings, traffic lights, etc that can be analyzed.
- Structures, people, vehicles can be recognized.
- CMOS can detect beyond visible spectrum. Image amplified, filtered.
- There limitations with depth, rain, fog, snow. That is why radar is essential along with U/S. Plus precise GPS. Tesla should have included 360 deg radar. Probably expensive, power hungry.
- AP1.0 works quite well with 1 camera plus 1 radar and us.
Lidar gives 3D data, but still need camera to read. Does not work in bad weather. Needs radar. $$$
Lot of fake news around. 35M shares or 40% of available float is short. They want your shares bad!
I think that as radar tech evolves, like PE or phased array radar, Tesla will add 360 deg radar coverage and even Lidar. Meanwhile , my Model X has AP2.0, but the only feature that works is smart cruise control. No lane tracking anywhere.
Still my Model X P100 is a great car. Occasionally the FWDs software need to be reset, very soft leather, so I have to be careful with those perforated ventilate seats. What is impressive is the driving! Great view, handling, etc.
1) That's only the state of California. They may or may not be testing in other states, no one really knows as other states do not have reporting requirements.
2) Tesla does not need test vehicles driving around in perfect California weather. Instead they have their cars on the road being driven day in and day out by owners. Those cars are collecting and reporting data at an incredible rate as well as experiencing every possible scenario on road conditions.
What a shocker a short posting an article on seeking alpha bashing Tesla. This is a shameful manipulation of information with a clear motive. Oh well, that is precisely what shorts are going to do.
The mainstay of Tesla's business right now is building cars and those cars are of high quality that is unmatched. Even if most of the statements in the article are true, Tesla will catch up in the FSD capabilities. Right now they must focus on getting the M3 production going, and in the meantime are gathering data. Dreams--yes--but Elon Musk is clearly one who can turn dreams into reality. I grant that EAP (currently in quasi beta testing) has a long way to go, and is even scary to use at times. So when I tire of it--especially the nascent autosteer-- I turn it off, mash the accelerator and relish in the experience that my Model S is a joy to drive. When autonomous vehicles become commonplace, Tesla will be there, but autonomous is for function not exhilaration and I vote for the latter.
Ignore Seeking Alpha
I think Tesla may still need 2 cameras pointing side-ways on the noes of the car. If you want to turn into a road at a blind corner. Humans can lean forward, but the camera in the B-pillar is too far back (I think).
Redundancy is not necessarily required in full self drive, but the car would probably pull to the side of the road and stop if a sensor / radar fails, and you would have to go get it. On the other had, if the car is in an accident by itself, you will also have to go there.
What I am a little concerned about, is interference from other Tesla sensors - radar and ultrasonics from the Tesla next to you.
I would prefer it if all cars on the road didn't have Lidar - want to see the effects on eyes over long term first. We might have many blind people at age 60 from all the lasers everywhere...
Tesla's reported hours are what was required to produce their self-driving videos. No more, no less.
It is interesting to see some of the other manufacturers advertising their driving assist/self driving features.
But there's a huge difference between Tesla and the other manufacturers. Last fall, Tesla started delivering hardware they believe can support full self driving - in every car. And as Tesla makes progress in software and validation, they'll continue to enable more features, getting closer to full self driving.
The other manufacturers have full self driving in a small number of test cars - and cars they are delivering now, will most likely have only driver assist hardware, and only if the car is purchased with the features. They'll likely only begin delivering full self driving hardware after they've completed validation - and by then, Tesla will have been delivering FSD capable cars for several years.
Of course, there is a possibility Tesla is wrong - and the current hardware will never achieve FSD. It's possible the AP 2.0 sensor suite isn't capable of detecting everything. Or FSD approval requires additional hardware such as vehicle-to-vehicle or vehicle-to-infrastructure communications. And if that happens, how will Tesla handle all of the cars that were purchased with a promise of FSD.
FSD should never require vehicle to vehicle or vehicle to infrastructure, because if the other car / traffic light is not working, it would make FSD dangerous.
I agree the capability should be added to enhance safety, but it should never be dependent on it.
If it was forced, it would take decades before it would be implemented in every small town and old car or truck on the road
All companies produce nice videos. Tesla is the only company that put cars in public's hands and letting people to see themselves.
@JHB10 Initially there will be V2F capable roads that admits only cars with V2V/V2F capabilities. Eventually all roads will be like that. This is the future no doubt.
.. and letting people to see -it- themselves.
@JHB10 - "What I am a little concerned about, is interference from other Tesla sensors - radar and ultrasonics from the Tesla next to you."
Radar data sent out and received is coded to be unique to a specific car and manufacturer. No two radars on the same band should make a mistake a return signal.
Not sure about ultrasonics, but I haven't heard of any confusion when cars are next to each other. The longer the range, the higher the frequency used. Old parking sensors were typically in the 25 kHz range, where the Tesla long range sensors are likely in the 60 kHz (higher frequencies extend the sensing distance). So the Tesla should not be confused by other car makers parking sensors. The ultrasonics might be coded to reduce detection of other vehicles sensors, perhaps by timed pulses, but I couldn't find any confirmation this is done or not.
I have the feeling that full autonomy over random routes under random weather, traffic, and road conditions is much more difficult to achieve than Tesla is letting on. Virtually every day I encounter conditions that a self-driving car would be severely challenged to cope with. I am skeptical but I think is is worth trying to achieve.
Certainly, autonomy can be achieved on controlled, repeating tracks. We have a local mall that is trying out small, autonomous busses. It is a start.
Infrastructure to vehicle communications may be possible via internet like signaling traffic lights with GPS tags,timing, etc. V2I comms could be implemented just for the interstate, with lane markers, speed limits, road repairs, detours, exits, hazards, etc.
I have wondered about the health risk of Lidar. What about radar? With so many cars with CW microwave radar turned on result in health hazards? I know airports ask airplanes to turn off radar on approach to airports.
Tesla is doing it right .. they are recognizing that the barrier to overcome in self driving is the brains of the system not the sensors. HW2 has way more sensors than a human as is.
As for why the low number of miles on fully autonomous ... Tesla uses machine learning using their entire fleet. The fleet just started collecting data ... the number of fully autonomous trips will be able to expand as the the system learns. The advantage of this strategy is obvious already ... HW1 equipped cars far surpass anything else on the market.
Perfectly stated. For those who do not realize how powerful this machine learning is you can take a look at how human brains develop. Babies are born with 100 billion neurons (hardware) but few connections (firmware) in between. Eventually adult will have 100 trillions, yes with a T, of neural connections with 80% of those developed during the first three years of life. Machines can actually learn at a much faster pace of how to recognize images or drive a car than human could. That's why Elon has made those comments of how powerful or dangerous AI could be.
If you are not worried about that your newborn baby child could not walk you need not to worry that your newborn AP car could not drive itself. Wait till it to learn how to walk, and also to run or even plays NBA point guard.
The advantage of Tesla of course is that Elon is the only one among all auto company CEO's who understands and knows how to utilize AI. Executives in other companies are probably as clueless on this subject as George is (sorry no offense ;)).
Someone mentioned the Tesla needs a very powerful super computer, for AP 2.0
"The advantage of Tesla of course is that Elon is the only one among all auto company CEO's who understands and knows how to utilize AI."
If only he better understood how to communicate with customers. (Yeah, I know...)
Autopilot? What autopilot? only 45MPH in highways? My new X doesn't have autopilot, nor self park or even auto wipers and many features I paid for, I am driving a car with 70% of the features I PAID FOR.. My wife mercedes Distronic cruise control works better than Tesla and when innovations or changes are available you can ask a human being who would provide real ANSWERS. You don't have to wait for reading tweeters from the leader of the company to find out when the feature you paid for will be available. Tweeter is the only source of information from Tesla, similarly to another leader.
I am still waiting to have at least 90% of the car I paid over $120,000. And of course don't bother calling or emailing Tesla Service, if you get an answer play the lottery.
No wonder why the Tesla used market is growing and now for the first time you can buy a brand new one with immediate delivery.
If Tesla does not change its communication and customer service, I think within five years will join Delorean.
Maybe so but that's not him to blame. Eienstien once said there were only three people in the world who understood his theory of relativity.
Sure we can blame Elon: it's his company. He makes the rules. If better customer communication were a priority with him, it would happen. Actually, it would have happened years ago.
Windward - get real. You bought knowing full well it would take some time for the new software to catch up. AP1 far exceeds anything on the market. Your AP2 will be self driving you to the yacht club 2 years before anything comparable. Chill out and enjoy the updates as they roll in ... your car will carry on improving beyond what you can imagine.
"Eienstien once said there were only three people in the world who understood his theory of relativity."
It wasn't Einstein who said that, but a reporter who didn't understand the theory of relativity who said it. It's blatantly obvious that anybody who teaches physics understands it, and so should their students.
As for FSD, Tesla might not have it done by the end of the year as they are saying, but to say it's not close is absurd. AP1 took longer than expected, not that Tesla gave a time frame. Out of countless articles and interviews when it was announced, I couldn't find one reference to a reporter who asked when the software would be released. This time around, Tesla has years of experience and a foundation to build upon. This time around, Tesla gave dates. They may miss the dates but that's a far cry from saying it's just dreams.
Haggy, the Special Theory, with my physics skills back in the day, I could claim I understood.
Not so much the General Theory. :-/
I actually have a Ph.D. degree in Physics. I don't understand the general theory of relativity either. That's a toughie. Glad to know Haggy nailed it. ;)
About the story of only three people who understood it here is a quote from the Wiki.
"In the early years after Einstein's theory was published, Sir Arthur Eddington lent his considerable prestige in the British scientific establishment in an effort to champion the work of this German scientist. Because the theory was so complex and abstruse (even today it is popularly considered the pinnacle of scientific thinking; in the early years it was even more so), it was rumored that only three people in the world understood it. There was an illuminating, though probably apocryphal, anecdote about this. As related by Ludwik Silberstein, during one of Eddington's lectures he asked "Professor Eddington, you must be one of three persons in the world who understands general relativity." Eddington paused, unable to answer. Silberstein continued "Don't be modest, Eddington!" Finally, Eddington replied "On the contrary, I'm trying to think who the third person is."
I bought a loaded Model X P100, including the AP2.0 options, knowing that it would take a long time for Tesla to roll out AP software. I do not want software that is unreliable... way too risky to turn it and depend on alpha versions.
Meanwhile, these cars way too much fun to let AP drives. Especially the Model S P100. I got a P85 and has no hi tech sensors. Still, it is a lot of to get on the highway and just stomp on the accelerator. The wife used to drive the Toyota minivan way under the speed limit like some assisted living resident. Since she is now driving the P85, she steps on it. I need a speed limited software.
Carik: Bubba: Tstolz: +100 for continuing to push back against impatient, ill informed and glass half empty people who only serve to hurt Tesla with their carping and thus their own best interests going forward.
Carik: I do have a question that you might be able to answer. On all my thousands of miles in the Tesla I have only encountered one truly blind corner. There is a speed limit of 30 mph in both directions. My question is: my new 2017 Tesla model S has longer range sensors than the older model and has one right on the side of the front bumper facing the oncoming traffic- do you think it will be sufficient to allow the car to make the turn under AP2?
Before you answer, if the answer is no, I can always bypass the once a year corner, by forcing the Sat Nav to take a different route :)
@windward - sell your Tesla. I'm sure you will find a buyer. Go back to driving your wife's wonderful Mercedes.
SO: +1 you got that right. windward: Yes please sell it and give us a rest from fake news. I owned a Mercedes S class with Distronic, air cooled seats, massaging seats blah blah etc. Over $100k. I took ownership as the dealership was closing, The sales guy gave me the key, locked the door and drove off - but I didn't drive anywhere - left rear tire absolutely flat. The Distronic was rubbish compared to Tesla's TACC, the car was boring compared to my new Tesla model S so don't blow smoke up our ...
SO_S90D no you are wrong No is not easy to sell a used Tesla, probably you will never know as you will always own a Tesla but no it is not easy to sell one, and yes I tried to sell my my 18 month old 20,000 miles S before trading it in for less than half what I paid, and.no I was not able to sell it for one year not one single call except one for 30% for what I paid for the car. That is why there is a growing used tesla inventory for sale, and that is why there is a button on the tesla website for NEW Inventory. Sales are going down along with the company. There is no customer service READ the forums NO customer service. Sorry to the cult followers for breaking the bubble, the car is great but the company sucks. Check out lucid air.
Strange opinions. I have a Tesla which is my second one and love the AI AP2. Probably the short stock holders are sweating a bit now. Keep trying as buying lower for me is a gift . I intend to hold forever
Just got back from my drive in 28 F weather on ice and snow, love the D and still grinning ;-)
windward: now that is definitely and absolutely Fake News. I sold my model S 3 months ago in 2016 to buy my new AP2 model S. Got a very fair price and they held the offer until I picked up the new car in January 2017. Of course you're going to lose the tax right of the top and some depreciation for being 2 years old but all cars suffer that so don't single out Tesla. However, it's a good fake story to cover up your real intentions, which are that you've never had any plan to sell your car but desperately need to cover your short position! I can't think of any other logical reason to explain your attack on Tesla with such obvious incorrect information.
Very astute comment. The new Tesla is simply elegant. I am also having fun getting updates on the AP2. My car is learning! This is changing our life for the better. It is amazing.