Which Company Is Winning In Autonomous Vehicle Development? One Study Suggests It's Not Tesla...

[@photopawn37]

How do you even package LiDAR to a consumer car without making the car look weird?

[@hellcat1988]

People expect mainstream vehicle manufacturer cars to have recalls and defects to the point we hardly even notice them anymore. Tesla's not being given the same expectations. It's like the difference between a shotgun spray and a sniper's bullet. People see the spray of problems with mainstream manufacturers and treat it as one problem, despite being spread out, with each presenting a problem that could be just as damaging as the hypothetical single fault in Tesla. Yeah the falcon wing doors might be able to damage limbs or sever fingers if you're REALLY stupid with them, but I know people who've had just as bad from conventional doors. Hell, I almost lost the tip of a finger on an old chevy pickup door. And no I don't think that's the only problem Tesla has had. It's just an example. When these other companies bring their tech to market and people start finding the flaws the techs were able to hide from reviewers or didn't even discover themselves, they'll still be forgiven for those problems unless they're serious and dangerous. Tesla may never get that level of expected fault forgiveness that Ford and the other global manufacturers get from their apathetic buyers.

[@smwforever45]

Reasons why I think Tesla has better chances than Ford, GM, Nissan etc.: 1. The overwhelming lead that Tesla currently has. Tesla: Has Level 2 autonomous cars ON THE ROADS and has announced to push AP2 to Level 5 autonomy within the next few months and years. To help them, they've got millions of miles of test data from public roads and actual people's driving habits. And they are actually doing something! My dad got his Model X with AP2 just three and a half weeks ago and there have already been two updates improving the Autopilot functionality. Most other car manufacturers: SAY to have higher-level autonomous cars - but no one except themselves can verify their statements because it's all behind closed doors. Plus, they barely have any autonomous cars on public roads and thus lack necessary data to improve their systems. 2. LIDAR sitting atop the cars? Seriously? Who would want to buy a car with such an ugly thing on the roof? Tesla'a technology might not be as advanced, but it looks a LOT nicer. 3. Established car manufacturers make most of their money off the stuff that currently sells well: ICE cars with level 1 autonomy (cruise control) at most. Research on autonomous cars simply doesn't pay off for them at the moment, why should they care to invest much time and/or money? Tesla, on the other hand, is not a classic car manufacturer. Elon Musk is an inventor, not a businessman. He has visions and he WILL try them out. No matter the money and time. These three things are also the reason I don't understand the comparably low ranking of Tesla in this study: a. You can't argue with the age of the companies and rank Tesla lower because they're new to the game. That's like saying Apple could never succeed with their iPhone because there's Nokia and they're an established company. You know what happened. b. As I stated above, I don't think LIDAR is the ultimate solution and thus I would not rank companies higher based on whether they use LIDAR or not. There needs to be something more compact... and less energy-consuming, too.

[@TheTCOLL]

Months or years? Which is it?

[@smwforever45]

I expect roughly two years, but obviously we're hoping for a faster release.

[@TheTCOLL]

Ford has been showing off theirs, a few other companies as well. Does Tesla have any proof?

[@smwforever45]

Way before Ford, Nissan etc., Tesla has shown two videos of an autonomously driving Model S that even parked itself at the end of the video. Never saw those?

[@Andy-em8xt]

1. Tesla no longer has a lead even in on road cars. The new Audi A8 has level 3 autonomy up to 37 mph. 2. The LIDAR in the prototypes is not the LIDAR that will be built into commercial cars. Again the Audi A8 has a commercial LIDAR system that is concealed quite well. 3. Car brands aren't dumb, they know they have to adapt for the future or die off. The old manufacturers have far more capital and resources at their disposal than Tesla and have well established R&D labs. Tesla is actually outsourcing a lot of their Automation and not really pushing the technology forward as much as the other manufactures. Your rosy almost cultish view of Elon Musk is no different than how Apple fans idolise Steve Jobs. Elon Musk is a visionary and has done a lot to push the industry foward but don't think for a second that the other companies are not quickly catching on.

[@AndrewduToit]

Like your new format. Well done and great topic too. Keep up the good work.

[@mortimersnead5821]

I want self driving taxis to completely take over, so I can turn my parking lot into a garden.

[@richardvasquez4171]

For a company that they say ranks 12th, they're spending way too much time explaining why Teslas not the industry leader.

[@zerix01]

Correct me if I am wrong but doesn't​ Tesla's system creat a cloud point model using their sonar sensors? This is the same result as LIDAR but at a reduced cost.

[@wiscokidd3279]

zerix01 I have heard that before. I believe it was a quoted by Elon if I'm not mistaken.

[@PatrickKormann]

No, AFAIK they're doing that with the radar. And there's only one front facing.

[@robmagee100]

www.tesla.com/blog/all-tesla-cars-being-produced-now-have-full-self-driving-hardware

[@LoanwordEggcorn]

Tesla is using the *video cameras* to create the point cloud. Radar is for longer range forward detection and ultrasonics for short range side and rear detection. The key to Tesla's autopilot is the image processing from the multiple video cameras, and neural network processing of the resulting data.

[@robmagee100]

Tesla is actually using and developing the radar processing to eventually be the primary source for the point cloud.... www.tesla.com/blog/upgrading-autopilot-seeing-world-radar

[@nicksturkenboom2879]

Very interesting report! May I ask you to where I can find the study you are using in your clip?

[@robmagee100]

I see the rankings as heavily weighed towards the oldest manufacturers. Of the factors used to weigh the results, 5 are weighted towards traditional, established manufacturers. Partners, Staying Power, Product Portfolio, Production Quality and Reliability, and most of all, Sales Marketing and Distribution. Autopilot is about Technology and companies like Google and Tesla have the best tech engineers in the world, with Tesla having the best hardware engineers as well as software. Elon made the decision to go with radar instead of Lidar because radar sees better through fog, smoke rain, etc much better than Lidar, and thus is much safer. Tesla has had to play catch up with programming since Mobileeye was dropped, and the new processors and sensors were added, but I see the Tesla suite as far more robust, potentially, than a Lidar-based system.

[@LoanwordEggcorn]

Tesla uses image processing from the multiple video cameras for most of its autopilot functionality. Radar and ultrasonics are secondary. Radar is used for longer range front detection.

[@APsupportsTerrorism]

If they have the best tech engineers in the world, why did Tesla have to outsource autonomous features - and then commit IP theft when they finally decided to go independent? Hmmmmmmm So many fanboys.

[@FrankTaylorLieder]

Comparing Tesla's released capability with what's in the labs for other companies is not really a level playing field. We've seen that Tesla has full autonomy in the labs. Dinging them for not using lidar is wrong too, they look at what the system does, not how it does it.

[@dr-sy1fs]

A "study" stating that lead in petrol is harmless for the environment or that cigarettes are good for your health, doesn't mean it's true. I'm sorry but I do not trust this "study" further than I can throw it.

[@robertrowland1061]

This supposed diminishment of Tesla is an attempt by the established car companies to make themselves appear ahead of the game, when in fact they are trying to catch up.

[@TheArmchairrocker]

Market penetration is one of the most important factors. If it doesn't make it to market, what good is it? Betamax was considered better than VHS. VHS won because of its market penetration

[@heidimitchell7387]

Until the Leaf and other manufacturers get their proto type "secret" cars on the road, Tesla is leading.

[@johankroes19]

I think that real world experience is very important with autonomous driving. Tesla has that, and the company is very ambitious, so i think that tesla will stay leader in affordable autonomous drive

[@EngineeringEssentials]

as said the autonomous system eats up to 4.5kw. Wich means 20 miles of range every hour. compared to Tesla's current autopilot 300w 🤔

[@samreciter]

I see this completely different. Todays prime technology for level 5 autonomy is deep learning neural networks - and that means training data - a lot of training data. The only two competitors in the field having the ability to collect this vast amount of data - are Waymo and Tesla, maybe Nissan - bei Baydo und den Asiaten allgemein weiß ich zu wenig. Waymo drives with lots of cars since years - and Tesla has it's users. Both have the needed capability inside their cars and the network capabilities to collect and adapt the data. I think this is key - no company that has some secretive testcars running on virtual worlds or test runs can keep up with this. Sensors getting cheaper - then you can swap or add them as you like. But learning to drive - having a neural net and world mapping that is fast and reliable - that can cope with humans, animals, weather - all this unpredictable stuff - that is nothing you can create without real world data. I really think - although by now they can lie to themself by adding fancy, expensive sensoring - all the others car makers don't even realize how far back in the race they are.

[@PatrickKormann]

I think big data is overrated in that respect. You have thousands of cars being driven around sampling some data. So what? That car has no idea whether it has detected the important things, it has no idea what the driver is thinking and it has no idea if the driver is even doing the right thing. Furthermore Tesla will only get a small fraction of the data. Test vehicles driving themselves and being supervised by trained driver/supervisors and with lots and lots of sensors give much more valuable data.

[@samreciter]

You would think - but that's not the case. Google and other big data companies have shown that more data trumps better data by far in case of training general neural networks. And what Tesla is doing with autopilot 2 - collecting all sensor data and how the driver reacts in a co-driver manner - is the best anyways. Waymo and others have shown that - to not scare humans to death - autonomous cars must even adapt to human like driving - e.g. not the best possible way to turn - with margins the technology is comfortable with - but the way humans are used to.

[@kriss2005]

It's the opposite. Autonomous driving in every situation is done by using neural networks. Neural networks need data. See Mobileye presentations. These legacy car makers have no idea what are they doing. See Ford demo needing an on-board generator.

[@samreciter]

As people don't do everything correct - artificial hand curated data is no good fit to train a NN for real life traffic. In a world all vehicles would be autonomous it would be better - but you would still have animals, pedestrians and weather, etc.

[@PatrickKormann]

Yes but I was talking about the driver not reacting correctly. But the main problem is, that the car has no means of knowing if it's own model of the world currently is correct or not. Regarding google using big data - they don't with their self driving car. They're using a very controlled setup. And it makes sense. Just think about it, a car can learn driving on the street with big data. But it can't learn from the reactions of people. Because people are not going to tell the AI why they did what they did. Maybe they have been asleep and overcorrecting, maybe they have seen something the sensors have missed, maybe they were just bored, maybe they braked because they needed to stop there to pick someone up. AI cannot learn to drive that way.

[@darkice241]

WHAT EXACTUALLY ARE THEY COMPARING? I'm very confused on your interpretation of the report as if they are benchmarking the current Tesla autopilot on the market VS. what's being developed behind closed doors from other companies. The report should be comparing the highest capability that a company has in development, not what's being offered on the market. And if that's the case, google's waymo should easily be among the top 3.

[@MichaelEdmond]

DANIEL CHENG exactly, i find it weird how it's shown that the autonomous leaf is better than model s but is not at market yet... and still GM and ford are top of the charts yet we have heard next to nothing from them, Daimler rates higher than Tesla yet i have only seen prototype or concept things on the road. it's as if the scoring is strangely in favour of old, American companies... the same companies that are pressuring the gov to roll back their emission rules ect... hmmmmmmmmm... i am not sold on this graph, while i agree Tesla is not top, i have seen nothing from GM or ford that suggest they are ahead of Tesla. (in autonomous cars)

[@joolzvega7298]

Sounds like this study was done with the intent of slighting Tesla

[@Scout339th]

"Published on Apr 5, 2017" Nah, it's Tesla. You forget to take into account that the current Tesla cars all get updates to better autopilot. It doesn't even matter if anyone else is ahead of them at the moment, because the sheer amount of people that own Teslas are larger than that of any other autonomous vehicle out there. 1. "Which company will be the first to market" Well, Tesla was the first to talk about FULL autonomy, not just some "Parallel parking" garbage. 2. "Which company has the edge" Tesla has the edge, but the only people in competition that comes close to them at the moment is Google's fully autonomous car, but it isn't affordable, OR released to the public, thus helping my statement with question 1, and also ties into question 3. 3. "Which company will make autonomous vehicle technology everyone can afford, and want." Tesla Model 3. It didn't release with autonomy, but has the hardware to support it in the future. Think the model 3 is too expensive? They will continue to make more affordable cars down the road, *it was even mentioned in Tesla's Roadmap...* 2:40 Reason why Tesla didn't make it in top 10? Because it's being constantly developed. That, and many people that rate these things might be against Tesla for an unknown bias, just like large oil companies. That rating system is clearly flawed. 3:02 I won't leave the comment section alone, this is ridiculous. The poor messenger giving all this garbage information. Please slap the Author for me. 3:48 This is because the best form of development is using it in the *real world*, not behind closed doors. Was tesla the first to the Autonomy race? No. Are they the fastest DEVELOPING autonomous cars? YES. 6:20 Thank you for your insight, and that is the flaw in the rating system. It favors that of the HISTORY of the makers, rather than the work that they put out NOW. 7:48 Telsa is governed by Elon Musk, he will easily find something that will keep him ahead of the competition when more competition arises (which is exactly what he wants. More electric vehicles other than Tesla in the market.) Sources: www.tesla.com/blog/master-plan-part-deux www.ted.com/talks/elon_musk_the_future_we_re_building_and_boring electrek.co/2017/04/29/elon-musk-tesla-plan-level-5-full-autonomous-driving/ This video is so flawed its ridiculous.

[@madhusreedharanpathirapilli]

That is a good piece of information. Thanks.

[@mrdsn189]

HI Nikki, I think you have a typo at 0:52. Not sure if it would make you want to re-upload though....

[@pangit9999]

mrdsn189 yeah I think she accidentally used the title from a previous video!

[@arturasp9738]

so... the analyst says that the best autonomous cars are in the future... wow...

[@AllanSustainabilityFan]

I'm not sure I agree with the report. I would say Google and Tesla are (imho) probably the top contenders if going by miles driven, which I understand is how the machine learning happens, the more miles you put on the road, the more the system improves: electrek.co/2016/11/13/tesla-autopilot-billion-miles-data-self-driving-program/ The only complication here is the fact that Tesla switched to nVidia hardware around the time those miles were gathered, so worst case scenario is they had to do it again from scratch - doubtful - but then they have all these cars their customers drive to help them gather that data for them once again. Just my 2 cents. Time will tell I guess.

[@ScorchedEarthRevenge]

Tesla has fully autonomous driving functionality that they keep behind closed doors too. If this report is comparing like with like they wouldn't have compared Teslas production systems to the prototypes of other companies. They would have compared Teslas prototype to other companies' prototypes.

[@prabhat8413]

I don't see how you think this was biased. It was independently produced out of Detroit by a Ford guy who graduated from General Motors Institute of Technology (n.k.a. Kettering University). Ford and GM being ranked first and second is not evidence of bias given the independent analysis.

[@nettlesoup]

baba bindash And with Kettering alumni such as GM CEO Mary Barra and Ford VP Raj Nair, there is no chance of a conflict of interest. Tesla would surely want to reveal all of their secrets to this very independent analysis.

[@Andy-em8xt]

Can you actually cite a source for this? Looking at the original report and researching the authors I couldn't find any link to General Motors or Ford. 62.nl.dealer-preview.co/SiteAssets/LB-AV-17-Navigant-Research_FINAL.pdf

[@alileevil]

So what if he graduated from a non Ivy league University? Universities in the USA must meet certain standards. I wonder if you actually read the report or just dismissing his report as a Tesla fanboy.

[@chgwwh]

Any "study" could be out of date by now.

[@sizzlingbacon8952]

the problem with lidar in the future if so many people use lidar it might affect another car with lidar so camras are the way to go

[@felixsu375]

The report is misleading. The one thing that determines how good your deep learning systems is is data. Tesla has 700 million miles of driving data and is generating 1 million miles per day. Waymo, after 5 years, has 1.5 million miles of driving data. The other car companies aren't even close. That will determine who wins because as Musk has stated, it isn't the driving conditions that you see 90% of the time that will determine how safe your autonomous car is. It's the boundary conditions of the last 10%. They're the only ones that have enough data to put an autonomous car on the road today. The other companies aren't even close. They have less than 1% of the driving data Tesla has. They're actually years behind.

[@artemaung5274]

Felix Su Waymo hit 3 million miles autonomous driving. Tesla has a lot of human drivers data. That's a big difference. I do think Tesla has a chance to overcome Waymo in 2-5 years because of mass produced cars and data, but RIGHT NOW no one even close to Waymo! over 5000 miles between disengages in 2016 after 1000 miles in 2015. it's not a stretch to think they will have 10000 - 50000 miles between disengages this year. Other manufactures can't even break past few hundred highway miles, and likes of Nissan and Uber can't even do it longer than 1 mile on average! You just vastly underestimate how far ahead Waymo is right now.

[@jojodroid31]

The most advanced system I've seen which isn't just a research project like Google's one is Nissan's Intelligent Mobility. Is planned for 2020 but already works right now. Basically fully autonomous.

[@jonaboy3]

Of course lidar is going to work a bit better, it is also going to cost wayy to much for the foreseeable future and also looks like shit. Tesla is the first company that will offer lvl 5 at a reasonable price, no prototype bullshit.

[@1man1bike1road]

tesla is between level 2 and 3 right now probably 2. level 5 might not be achieved in 10 years

[@Andy-em8xt]

The new Audi A8 has a commercial lidar system that costs ~$500 and is the first production car to have it. Already lidar is becoming commercially viable in cars and Tesla has no excuse.

[@nickjames5366]

Driverless trains run in a controlled environment, planes fly on autopilot in a controlled environment. Cars on highways drive in a semi controlled environment (depending on the country). Urban driving is a wild environment. Who is legally responsible in a situation where an autonomous vehicle has to make the decision of killing a) the texting teenager pedestrian, b) the exuberant dog running across the road, c) the driver of the vehicle coming in the opposite direction when the autonomous vehicle has chosen to swerve to avoid A or B. The authorised, licensed, insured driver should be alert and in control of the vehicle at all times.

[@nickjames5366]

Before autonomous vehcles are allowed on the road, legislation needs to be introduced.

[@nickjames5366]

Sorry, I should have included my ex-wife in that scenario - she thinks it okay to drive down the Center lane of a road at half the speed limit. This is why she in now my ex-wife.

[@nickjames5366]

And I should have included my late mother, who was a danger to shipping and low flying aircraft when she was behind the wheel. Lol.

[@1man1bike1road]

i remember a lorry crash in ironbridge telford where the driver decided to crash into a shop full of people probably because of his fear of drowning in the river severn. Not as he said he thought he would kill more on the river bank. First thought for most drivers is always for themself maybe a mother would die rather than kill kids though. Autonomous cars will never think of its own survival before others. Humans are basically selfish

[@joseluiz7017]

It’s 2020 now and yeah, Tesla is winning!

[@asleep-cult6877]

Try again when you have a model 3

[@LoanwordEggcorn]

Tesla is clearly the leader and has the broadest plans for mass deployment of autonomous driving, both in the existing Model S and Model X fleets and also the upcoming, mass market Model 3. They also have a very clear path to market with standardized autonomous driving hardware in all current and future cars, including neural network hardware. More importantly they have over the air software updates and fleet learning. Who else has that already deployed in production and for sale? No one. Tesla is also collecting and automatically learning from more than 1 million miles of real world autonomous driving data every day. That's also a huge lead. Telsa has already proven that Lidar is not needed. The fact that the other companies are still stuck on it shows how slow-moving and inflexible the other companies are.

[@N8844H]

A bit like rating auto manufacturers by the "quality" of their concept vehicles, isn't it?

[@APsupportsTerrorism]

Nissan has been fielding ProPilot in Japan since 2016 model year (read: 2015), and will debut in US in a week. #NotJustAPrototype #NoIPtheftNecessary #NoOutsourcing

[@arlindbanushi]

The winner is Oshkosh. In 2005 made a fully autonomus truck. Capeble of going anywhere. Having even terrain stability control, and land mine control. So Oshkosh already sold that vehicle over 10 years ago. I understand that it's not on the list because of it's client.

[@mr88cet]

Interesting! Another advantage Tesla has is massive amounts of practical data from thousands of real-world users.

[@morosis82]

More than 200k users now.

[@arjensmit6074]

Where can we find that study ? Maybe add a link in description ?

[@DoodleDoo]

this is great for everyone, more competition leads to lower cost.

[@BrankTech]

LIDAR needs to rotate to work, fixed position cameras don't. Actually LIDAR needs to rotate at specific speeds to work. To me that's a deal breaker if I'm relying on that to drive my car. It's just one more moving piece and Point of Failure that I'd rather not have.

[@LarlemMagic]

Why does it need to rotate to work? What stops binocular cameras from being good enough?

[@morosis82]

slimjimmy, not enough people recognise that.

[@ra6865]

The other cars use much more energy and their sensers are clunky and not implemented in the design of the car. The tech they use is also way more expensive. So while they wait 2-3 years for the price to go down Tesla Autopilot 10 will probably be out and have the same capability. This was a strange report

[@carpenterfamily6198]

I think in the end, 'We' the consumers win. This technology will be great for baby-boomers as these cars can drive them around even after they can no longer drive themselves. I really don't care which company wins, but I do think more than one company will offer this product.

[@zapfanzapfan]

The LIDAR-system probably has to become a bit less of an eyesore before going into sleek production cars.

[@Poxenium]

I hate these analysts, who look at everything from a strict, traditional business standpoint. They are the ones, who praise companies like Comcast, in monopoly positions, which can screw over customers as much as they want, making lots of profit.

[@FurEngel]

"One study...", there is a ton of misdirection right now being pushed by the traders who have huge shorts on TSLA. Who is the leader of EV? Telsa. Who is the leader of vehicle tech? Telsa. Who is winning the autonomous vehicle? Tesla. Any other answer is a lie.

[@john0constantine]

It seems to me you fail to understand that the difference between assisted and fully automated driving is not gradual but a principal difference. The cited study features a lot of issues that seem irrelevant in view of the technical challanges yet unsolved.

[@patmcdaniel7995]

You seem to be assuming Tesla will not advance anymore. And that LIDIR issue is pure bias. Please explain how it is not.

[@francesconicoletti2547]

The variables you mention in the report seem to come from car guys praising other car guys for doing car guy things. Nowhere is there any mention of 'successfully rolling out new technology ' . Ford & GM have only done that in the last half century under government duress & after much resistance and fumbling.The GM Volt for instance is a half measure,without the fast charge as standard that would make in a general purpose car. Ford actually claims that complying with proposed fuel economy standards would destroy the auto industry, if we take it at its word, that level of technological incompetence means that it could never deploy anything as new and complex as autonomous vehicles.

[@toddschroeder8971]

Well geez. This is a tough one to talk about. You did an excellent job of trying to be fair. IMO though I have to say. When it comes to what company is "winning". At the moment it should be recognized that Tesla is the best currently. Anything else that is not being produced and a prototype doesn't count. Just the same as FF or any other prototype. Its simply not out. Now, once everyone else actually produces something that is available for the public to buy and drive. Who knows. There are so many brilliant people working on this. It's not likely Tesla will hold that crown. And that's a good thing. It will cause everyone to step up t whatever level that is. I am excited for what's to come no matter who brings it. Safer is what we all want. :)

[@Patchuchan]

The only way to see who's really winning at this point would be to look at the code. Who's demoing raw prototype hackerish code and who has working code written to real time control standards?

[@TheRayweather]

you keep mentioning GM a company that's defunct in a few years.

[@walternyberg]

As one who enjoy driving, I have to say, this is an option I would not even consider before I start to drool to mush! :-)

[@zcustard]

did Ford pay for the study? i've not heard anything from them

[@nicksutton2964]

At 8:04 you say "...who will ultimately win the autonomous vehicle race?" What do you mean by "win"? Do you mean first to market? While that would be a major achievement it says nothing of improvements, of competition. The process is on-going just like any other competitive market. If the others don't "win" it won't stop them from trying and we will see more and better systems as the technology improves. I expect that we will also see off-the-shelf complete system kits that smaller manufacturers will be able to fit to their cars. So perhaps the answer to your question is right there - those who buy and use the vehicles ...."us!"

[@robsmith1a]

I'm not sure it really matters who wins the technological race because the legislative process and more to the point the public acceptance of self driving cars will be way behind the technology. EVs are a practical purchase for a lot of drivers but there is still a lot of resistance to them because people generally stick to what they know and believe in until the case to change is totally compelling. People are terrible unsafe drivers in reality but to get them to trust a computer to drive a car is a heck of a leap of faith in most cases even if it is demonstrably safer The other thing is that a lot of people (me included) actually enjoy driving.

[@shawnueda8909]

There is no substitute for real life experience. None of the "old" established companies, be it GM, Ford, BMW... have "better" technology, it still has to prove itself in the real world. Tesla is proving itself day in, day out and everyday. Is LIDAR a necessary requirement? Animals, humans, birds don't have it after billions of years of evolution, so why is it necessary? So, as a retired engineer, I would have to say that the study is flawed when it decided that LIDAR is a requirement for being more advanced. total BS.. My bigger beef is what happens to all these system when its twelve years old and degraded because Joe-6pak can't afford to maintain the car? Will the system degrade gracefully or catastrophically? or will it just shutdown and let the owner know that autopilot is offline? Hope it goes off line.

[@p1703]

Life experience is not going to be substituted as a whole thing - it's only about controlling a vehicle safely - this requires less intelligence than an average human has got. But foresightful and riskless handling of a car will be programmed (this is what a human driver needs to learn first). Correct, LiDAR is not necessary in principle, but it can gain superior safety, plus compensate a lack of artificial intelligence in certain situations. So, both of your arguments seen in a combination build a serious argument against your opinion. And if an autonomous system realizes it cannot handle a situation or it is not up-to-date, it will stop the car safely (if needed) and go off-line afterwards - this is a basic prerequisite which is necessary for consumer market.

[@Patchuchan]

One to to consider these SDVs are at best about as smart as an insect so the 3D data needs to be as clean as possible. The code to process 3d data from a lidar is simpler and thus easier to debug. But software support is a big issue and I doubt they'll keep writing new software for a vehicle that is 10 years old but a vehicle needs to last 15 years to break even on the carbon debt of it's manufacture.

[@isaiahfreeman]

I am putting my money on ford based on the potential production level, price, and current business moves. When they do release I see them leading or being just behind GM. Which could translate into some nice returns.

[@jamesirwin7677]

The reason they aren't using LIDAR is price. It's still way too expensive. Until the price drops its up to the cars AI to handle things.

[@professor0magic]

you didn't mention the price of lidar. If it gets a hundred times cheaper over the next five years, it will be relevant, but at 10k for each sensor, it isn't relevant today, or for the next few years.

[@patmcdaniel7995]

Some of those parameter are total BS. You know it.

[@Szarko32c]

Let's be serious. Other companies had those technology, but they sale it - they just use lane, stop or park assisatance, but don't intergrate the technology fully to keep car affordable. Also customer dosn't want it and in many markets it's illegal.

[@jasonbaxter3658]

You don't need Lidar to calculate distance with a moving camera. Calculating distances and shapes etc with cameras is very complicated but its possible, we do it fine with our own eyes. Personally I don't think LIDAR is practical due to price and how silly it looks

[@beastie_387]

how do you know Tesla doesn't have a much more advanced prototype then company's like Nissan so you don't so your comparing something to something that you might know nothing about I think that's a bit messed up

[@LarsPallesen]

It'll be interesting to see which auto manufacturer emerges as the market leader in autonomous driving. Ford and GM may have some awesome technology brewing in their labs. And who knows, maybe we'll see it in their cars in a couple years or so. But until then, can we all just agree that you can't claim to be the world leader in autonomous cars when you haven't even released one car with autonomy hardware?

[@mikelewis6726]

I have to common on this one if these cars are so great oh come we never heard about them we always hear about Tesla's autonomous driving and parking, Nissan Leaf I never see Ford in the spotlight how can I judge another car to Tesla if I don't see the car or hear about it.

[@AntonyThorburn]

'a fully qualified autonomous driver'. THE NEW DRIVING TEST??? Hahaha, classic.

[@p1703]

I'd say this topic should not be about winning at all. Several companies are developing their own concept and in the end, safety and reliability is all what matters. Instead, learning from each other is the most important thing today. Tesla is very innovative, yes. But I don't care if their Autopilot hit the market as first one. The first fatal accident using autonomous driving was caused by a Tesla, too. So my opinion is, it's much safer to test fully-autonomous software (which should be able to handle ANY traffic situation) in a small fleet of testing vehicles, monitored by instructed operators. So Waymo's, Uber's/Volvo's strategy is much more reliable and safer than offering incomplete technology to people who don't understand even driving by themselves ;) . And I guess, Waymo uses the most advanced technology, which is even capable of understanding human gestures/behaving and handling surprising situations. THIS is what driving a car is about, not who is selling as the first one.

[@marinovalencia5441]

Tesla, Uber and Google actually have more real world experience and patents in their portfolio. Those Big Auto companies spend a lot of money on acquisitions and R&D but never get it right on the money. Out of all those in the top 10, BMW shows signs of being a serious player in advancing the automobile.

[@onwike]

Long term watcher and subscriber... The issue with this report is that it takes Navigant's "tech Research" at face value and rationalizes it. Some pointers. The issue ALL automakers are facing with autonomy is how do you approximate and beat the visual recognition algorithms in our brains, with a computer. Most of these companies are going with MobilEye which relies on algorithms built by researchers which have to be tweaked by hand to account for corner cases. In fact, as is the case with anything that has to do with visual algorithms, the top 99% of cases are easy to recognize and deal with. The issues arise with corner cases and situations that the researcher could not foresee. The serious players are either using AI to create visual algorithms or purchasing companies that are using AI... Tesla is using AI To use an AI to create algorithms for any purpose, you need reams and reams of real data (read real world miles). This is why California and other states have allowed public testing as it is an essential step. THERE IS NO SUCH THING AS HIDING YOUR AUTONOMOUS TESTING!!!!!!!!!!! IT MAKES NO SENSE!!!!!!!!!!!!!..... Tesla is collecting data on public roads on it's own AND using it's entire fleet of production AP2 vehicles as datasources. So, if this sounds familiar, it should. This nonsense is the same silliness currently being spouted about other companies and EVs. For any company to compete with Tesla on EVs, they will need at least as many batteries as Tesla. To date, no other company in the world has access to the battery supply chain that Tesla does. As such, *ahem* NO OTHER COMPANY IS REMOTELY CLOSE TO TESLA ON EVS *ahem* Recap -Tesla's lead on autonomous production car research is damn near insurmountable (yea, Waymo has more miles, but they aren't planning on producing an actual car only selling their data) -Tesla's lead on EVs is blatantly enormous -Stop reading Navigant, they don't actually do research. I mean, look at their actual staff. /rant

[@jdelacruz6854]

By the time the other automakers come up with autonomous cars, there will be 100s of thousands of autonomous Teslas on the road. Updated over the air and being sold already with all the necessary hardware, I don't see how this is even a contest.

[@TheTCOLL]

Ford has plans to roll out by 2021. I haven't seen any fully autonomous Tesla and they currently do not have it. So who is to say who is ahead?

[@jdelacruz6854]

TheTCOLL There are videos of automated Teslas driving from point to point and parking itself at the end. They are working on it. And no one else currently has anything even close in the market. If we follow the current trend who is likely to be ahead?

[@truefalse7393]

(Where are GM, FORD and others going to get all the new batteries they need?)? Why would you buy an autonomous ice car? It's a checkmate on time and infrastructure, I don't think people are seeing the big picture.

[@ilanmallard9609]

It makes sense, but i believe that Tesla will come trough, because its system will already have "experience" and some of their cars already have the capability, and comparing to other cars, like leaf, fusion, pacifica shown in this video Tesla's system is hidden within the cars, and even though the other cars probably will too, Tesla is already doing it, but it will be impossible to know until other cars come out as fully autonomous and not just behind closed doors, and i believe Tesla also has fully autonomous (fully functional like the leaf) behind closed doors, and are releasing it little by little into their cars and when the time comes they will release it and it will be better because it will have learned a little with all the years of it driving, different from others.

[@livesimplyandhumbly]

As both an engineer and biochemist my point of view is more the merger of mechanical and organic. ALLLLL these sensors are just the result of paranoia, humans drive with two eyes, in fact, only one eye is required. One eye in the electronics world equates to just one camera, the dual camera is a luxury. The key to autonomous cars is A.I. The brain of the computer. The hardware that processes the data stream. Think of it this way. A simple wide-angle camera is enough for a human behind a monitor or glasses to drive the car. Despite the lack of sonars, lasers, and cameras everywhere, a human can drive a vehicle with just a single camera. Future autonomous cars will eventually reduce down to two stereoscopic cameras like the human eyes. But for reassurance purposes, the first autonomous vehicles will use a mess of sensors. GPUs can simulate the massive processing required to simulate what the brain does with the stream of data coming from our eyes. Combined with GPS and motion sensors it would be enough.

[@morosis82]

I get where you're coming from, but the point of the extra sensors is that it can do things we simply can't. Like sense a hard stop from a car in front of the car in front of you, due to bouncing radar off the road under the car in front of you. Seeing to the left, right, back and front all at the same time and being able to make split second decisions based on that information. Not only that, but as someone in the software world, the ideal would be that it makes predictive decisions all the time that it may very rarely take, but are there in the case of a prediction coming true - for instance, it might predict separate optimal paths for happy path, and various potential crash scenarios, in advance so that it can respond to late input in a fraction of a second, say when a car comes out of a side street too fast with little visibility. As a driver I actually drive like this to an extent (mostly comes from also being a cyclist that likes to stay unsquished), but it's exhausting when you drive a lot. Luckily I do not.

[@Guesswhokk]

hmm... bench-marking safety based on these concept cars. Some choose to develop their cars on publicly, others choose to develop it with their designer. Currently, I can't trust neither with my life inside it.

[@Tom_Mroz]

We all love "independent" research. Management of Ford, GM, Renault etc really loathe all what SpaceX creator is doing - as he is crushing their business model.

[@evilplaguedoctor5158]

unfortunately ford & GM aren't (or don't seem to anyways) be taking the change in the automotive industry seriously. so my gut feeling says that they are going to take a big financial hit in the next 10 years, and may have to scrap luxury features like their auto pilot program.

[@TheRayweather]

Google is the most advanced and you didn't mention it.

[@FritzSchober]

I think that fast charging networks are very important and only companies that offer tons of cheap fast chargers will win.

[@Tydud3]

3:09 fine I won't shoot the messenger.

[@alileevil]

Its going to be the Japanese and the Germans who are going to be leaders in the market. They have a well established consumer base and a production base all across the world. Their experience in the production line will make sure that production meets demand. Once the tech gets cheaper, Tesla will be in for a shock. That said, I do respect Tesla as they showed that its possible to come up with a practical long range electric vehicle. They must step up their production standards if they are to remain relevant.

[@kriss2005]

... and Audi is the leader of electric cars from the press releases.

[@karlp8484]

People see a Tesla which doesn't need that ridiculous Noddy whirling thing (or things) stuck on the roof to give a reasonable amount of autonomy, versus the LIDAR reliant vehicles. This immediately gives the impression that Teslas sensor tech is more advanced because it's almost invisible. And doesn't look bloody stupid like current LIDAR. But solid state blended LIDAR is just around the corner, and it's going to be more reliable than cameras and ultrasonics/radar. This where they are coming from, but for the next 3 years at least, Teslas are going to rule here. LIDAR's currently $35,000 a pop and that's going to have to change.

[@tigerdust542]

what does SALES MARKETING , PRODUCTION STRATEGY OR PRODUCT PORTFOLIO have to do with the practical application of being number 1 in applying autopilot ...nothing ..you can't put Tesla in a box with an out dated business model

[@lgear]

The review is a load of tripe. It uses measures that Tesla is naturally behind on, being a young and not as established company (yet). The big Tech companies including Tesla will be number one, as this is about software and machine learning - they have the best employees and software engineers in the world for that. And it looks like their route to market will be as direct competitors, not selling their advanced software platforms.

[@qrthrse1]

To me it sounds like neither technology will win for years to come. Cameras are pointless in heavy rain, snow, and fog. Just because cameras are recording driving conditions, it doesn't mean that the next trip will have the same conditions. I actually think that Tesla will be great for the price and already has thousands of cars on the road, although it can't achieve level 4 or 5 with cameras, so it can never really be autonomous. Lidar is superior by far, and can achieve level 5, and already has, but the cars will probably cost $120,000 or more. So Waymo cars won't be affordable. I can see Google becoming the huge emergent ride share company and using the tech to make money through owning the cars, or licensing to Tesla.

[@terjes7548]

I see many people belive that Tesla is keeping their fully self driving capable software away from its customers. I know they did a demo video on the systems capabilities, but i'm pretty sure it was done with software developed by Nvidia, and not their own.

[@jimmyporter8941]

Tesla uses an NVidia computer, and has access to all the software NVidia has developed. So if NVidia can do it, Tesla can do it.

[@terjes7548]

But it seems like autopilot 2.0 is preforming much worse on local roads today then in their demo. If they also own the software from Nvidia, why would they not build upon it?

[@mil3761]

I think the news that Tesla's market cap beat GM's was wrong. It was based on inaccurate information given by Google. Tesla is still around $1-2 billion short of taking the #1 spot.

[@Steppenkater]

So humans use only vision to drive a car and nobody says that this is not enough. And humans are many times distracted or tired or are watching in the wrong direction. Tesla's autopilot system with 8 cameras (and in addition: radar and sonar) which is vigilant all the time while driving does avoid all this problems. The only point that has to be improved is recognition of objects and situations. And there is only one way to accomplish that: Lots of computer power, a learning neural network and lots of data to be able to learn in the first place. And that is what Tesla is focusing on. It will take some time but in the end it will drive significant better than the avarage human. And that is all what is needed.

[@archigoel]

This Navigat article seems to be product of dumb AI. They ignore simple, powerful things to make Tesla bad. like:1. Need of Lidar- we humans have 2 visual sensors, Tesla has 8+ radar+ ultrasound. So it has superhuman sensor capability needed to drive 10 times safer then humans. Whats missing is Brain (hardware + software), which Tesla is slowly updating.2. Cost of lidar: its projected to come down in price by 2020, so till then, Auto Makers cant offer autopilot ..... giving Tesla a Huge USP. By 2019, Tesla would have prod capacity of 500,000 cars/year with L4 autopilot, which other automakes cant match till 2021, giving them huge first mover advantage.2. Tesla battery gigafactory, and alien dreadnought: There huge battery supply chain advantage cant be matched till 2021, givng 5hem time to scale and consolidate.3. Direst selling Model, SuperCharger, wiFi Updates: these elements give Tesla USP, which again cant be matched in near term by automakers.

[@PatrickKormann]

True, but it will be hard enough to do it with all those sensors. It's way more difficult with less sensors.

[@pradeepsoundalgekar4277]

Nvidia CEO : “No more than 4 years to have fully autonomous cars on the road” . For More Details Visit : tfortrends.com/nvidia-ceo-no-more-than-4-years-to-have-fully-autonomous-cars-on-the-road/

[@heesingsia4634]

Don't think autonomous drive will be feasible unless the whole city is autonomous. Human error will always be a problem

[@Lunker22]

If you really understood deep learning you would understand that what you are really asking is "which company has the most real data!" Therefore: you can not rule out Tesla or Google.The xy plot is meaningless.

[@earthwizz]

Tesla will have another couple of years development of AP and a couple of years of M3 production before others plan to bring either comparable cars or AP to market. They probably won't be waiting around for others to catch up.

[@TuanLe-yq7vx]

I appreciate the report but I just don't see how ford or gm can deliver. So unless it's available they got nada.

[@transportevolved]

+Tuan Le so isn't the same true for Model 3?

[@jimmyporter8941]

No. We have reasonable estimates for delivery of Model 3. We know they are tooled up to deliver it. It's happening. So far Ford and GM autonomous driving is vapourware.

[@TuanLe-yq7vx]

Transport Evolved you can place your order on the Tesla3. Delivery will start this year. Where as Ford and GM are probably going back to making ice cars with the Trump's regulations roll back.

[@Airborn14]

Tesla and Kia

[@pierreoffice3823]

all brands are surfers, Tesla is the wave

[@APsupportsTerrorism]

Ironic since Tesla is the only one that outsourced their autonomous driving, then hired away their tech leaders initiating an IP lawsuit which they settled (unfavorably). In other words, Tesla hasn't actually developed anything. Just outsourced and then stole tech. Meanwhile, Nissan is all in-house and actually ahead of Tesla selling ProPilot in the real mass market in Japan (not $60k "mass market" luxury segment)

[@manoman0]

That goes for the ICE as well. Nobody cares.

[@PatrickKormann]

I think that's true. Tesla is very aggressive with their timeframe, yet I haven't seen any indication they are even close to other solutions. Google has been doing this 'forever' and yet they're not claiming to be first to the market.. guess why? There are still lots and lots of problems to solve.

[@jimmyporter8941]

Google has a couple of problems. First of all they started developing before the breakthrough in deep learning in 2012, so they have likely gone down a lot of dead ends. Secondly they've lost their most important staff. They've been coasting for years.

[@PatrickKormann]

Well, if anybody is into deep learning, then it's google. They've lost staff, but I doubt they have problems finding replacements (and btw so has Tesla).

[@gerrylangford6485]

It doesn't matter whose system is the most advanced, Tesla are producing cars with its system for the general public and it doesn't matter how advanced your system is if it's only ever a trial version. getting on with it is the basis of Elon Musk's success with Space X and Tesla.

[@toptobottom247]

Lidar technology is not the leader. Non of those companies hasn't figured out how to intergrate the system into their cars where it doesn't hurt vehicle design. Also Lidar consumes too much energy as noted by Unplugged on it's test drive with the Propilot.

[@Somun-a]

Lidar itself is not the cause of the power consumption. It's the fact that they are using separate "computers" for each of the sensors. It's a prototype.

[@toptobottom247]

I wonder what Nissan is using for their Propilot that's currently working in the 2017 Serena

[@Patchuchan]

Also prototypes tend to run sloppier and slower code that needs more powerful computers. An algorithm written python or java is going to run a lot slower than one written in C.