This is bizarre, I’m in non-car autonomy but I talk daily with car autonomy folks. Everyone has been expecting a consolidation of companies through acquisitions but drive.ai showed us that it’s probably not going to happen. At The Information mobility event last month one of the SoftBank fellas said that autonomous cars will be ‘won’ by those who can outspend everyone else. Looks like backing up billions with more billions might be the playbook for now.
It's R&D and depends on a lot of new innovation to be solved (new ASIC hardware for real-time inferencing, new ML model types -- potentially a new breakthrough even).
> SoftBank fellas said that autonomous cars will be ‘won’ by those who can outspend everyone else.
Listen to them more. What else do you expect fund managers to say?
We will not see any real self driving in our generation, and may be not even after. That is much more for practical consideration than that of technical possibility.
Second to that, just how many self driving and AI startups are plain fraud? The amount of companies getting 9 digit valuations with nothing more than OpenCV hello worlds should make people to at least scratch their heads
> We will not see any real self driving in our generation, and may be not even after.
I think we can once we build purposeful infrastructure (think cars "on rails" with mesh-coordinated intersections).
> Second to that, just how many self driving and AI startups are plain fraud? The amount of companies getting 9 digit valuations with nothing more than OpenCV hello worlds should make people to at least scratch their heads
Argo, although lesser known (but still out of early Waymo and CMU's Robotics Institute), is every bit as competitive as the other major players. If anything, they are less bombastic than the Elon Musks of the world falsely tweeting that we would all be taking naps in our cars by last year.
I was seeking this definite info for quite a while. This kind of in may help my friends too. Keep on posting the great work. Some are helpful content in there. It is decent to see your site. I get a kick out of the chance to share these sorts of items to my friends. They like to read the websites for increasing great contemplations and information. Here you can check out my <a href="https://www.clazwork.com">best essay writing service</a> to having a wonderful support for your writing.
>> We will not see any real self driving in our generation,
I agree.
But we'll be driven a large percent of trips, in the near future.
There's on-demand shared(few people in a vehicle) riding - which can be quite cheap AND fast. ridewithvia.com seem to scale with that model pretty fast. A combination of that with strong public transport systems and regulations may beat cars.
But also, a combination of partial self-driving(highway, etc), with remote control driving, could work.
I think the corporate types are assuming that throwing money and engineers at the problem is going to "win" it, but I'm not sure it will. Google has plenty of both of those things and they seem to be stalling at the moment, or the very least they're not confident enough to go out and claim the market for themselves.
I think the engineers have been probably been wildly optimistic as usual, and the last 5% of car autonomy is going to take as much effort as the first 95%, or maybe double that.
Not sure how many Data Scientists you know but I've worked with hundreds over the years and have never met a single one who was irrationally optimistic about the work they do. In fact generally it is the opposite.
But you know the sort of people who are wildly optimistic and prone to over-exaggeration. Executives like Musk.
as another datapoint, I worked at GRASP lab and half of my classmates went Waymo, or some other self driving car company. They go there because of the calibur of co-workers, many of them do no believe self driving cars on residential roads is possible within the next 10 years. In fact, one of the robotics professors here (who is no stranger to PR hype) said: self driving cars with pedestrians on the roads is impossible
Most are optimistic about highway driving in California though.
Hm, I would say that some consolidtion happened with that deal. Quote: "VW is also handing over Autonomous Intelligent Driving, the self-driving subsidiary that was launched just two years ago to develop autonomous vehicle technology for the Volkswagen Group. AID is valued at $1.6 billion.
The Munich-based AID team will become Argo’s European headquarters, a move that will expand its staff 40% to more than 700 employees."
AID has been Audi's L4/L5 development arm, which was separate from VW's main research arm. It also helps to judge the size of this deal, because VW only seems to invest one billion in capital and the remaining 1.6 billion investment is AID. It would be interesting to know, how they came up with that valuation ...
> SoftBank fellas said that autonomous cars will be ‘won’ by those who can outspend everyone else.
Lol, of course they said that as their entire investment strategy is to throw more dumb money at startups than anyone else can possibly match. Their vision fund VC has been investing $100 billion dollars and they're planning on doing another $100 billion dollar fund. "To put the fund’s magnitude in perspective, its size is almost double the investments made by U.S. venture firms last year. PitchBook data shows that VC fundraising in the U.S. totaled $53.9 billion last year across more than 200 funds, and that was the largest annual raise in at least a decade." [0]
Unrelated, but does anyone know how many electric cars have an autonomy of more than 700km, while also charging fully in less than one hour? Is that even possible with the current technology?
Currently - none that I know of. Tesla is currently approaching that number with 592km(370 miles) of range.
As for charging _fully_ - that's rarely done in practice, because over 80% li-ion batteries charge much slower. Bjørn Nyland recently beat the record[0] for longest distance travelled in an EV in 24h (2781km) by stopping to charge from 10% to 50%, because only then he could use the full power of superchargers.
Most of the more expensive EVs come pretty close to that and seem to be competing on getting the numbers better there. E.g. I think Tesla is planning to break the 400 mile barrier pretty soon. 300 mile is pretty normal already. Most EVs charge pretty quickly to about 80%. 20 minutes to half an hour gets you plenty of range. Mostly EV owners don't charge to 100% unless they are at home or doing some overnight stop.
I think most normal drivers on a cross country trip need bathroom breaks, lunch/snacks/drinks, leg stretching etc. every 2-3 hours. Great opportunity to plug in and recharge enough for another 2-3 hours. Range anxiety is mostly irrational and mostly a non technical problem as your car probably has more endurance than you.
Have EV companies completely abandoned the swappable battery idea?
I always thought thatd be the endgame. Some standardized battery design where you roll up on top of a station and it swaps the battery underneath the car.
Or maybe this is something your car does for you before arriving. Which eliminates the amount of dead time tens of thousands cars will need to spend while using up expensive property in cities while charging.
>I think most normal drivers on a cross country trip need bathroom breaks, lunch/snacks/drinks, leg stretching etc. every 2-3 hours. Great opportunity to plug in and recharge enough for another 2-3 hours.
Everytime I hear this I wonder if the people saying it have ever taken a road trip. Stopping every 2-3 hours for a half hour on a long road trip seems downright crazy to me.
This is beside the point that during the busy season of summer travel there are sometimes extensive lines at gas stations on popular routes. Imagine if every car took 15-45 minutes to fuel up?
Range anxiety is real. EVs are the ultimate commuting cars: known distance, charge at home over night. But for the immediate future, over long distance, gasoline cars are hard to beat.
> Stopping every 2-3 hours for a half hour on a long road trip seems downright crazy to me.
It seems crazy to you because you haven't done it. It seems perfectly normal and acceptable to people who drive Teslas over long distances. I see people going for lunch, shopping, sightseeing at Superchargers. If you know you'll have to wait 20+ minutes, you just find something to do instead of biting your fingernails.
SO basically, if you adjust your behavior to match the car, it will, eventually, get you where you're going. And if you really want to enjoy lunch and sightseeing at such beautiful places as the Supercharger station in Kettleman City. Not everyone is willing to make these tradeoffs, especially for a car that costs twice as much as a comparable ICE one.
This is going nowhere really. When I had an ICE car, I never drove 12+ hours/day because it was too exhausting. Now I've done 5 such trips in the past 8 months alone and it was no issue because the breaks and AP keep me refreshed. And there's no comparable ICE car because they are all noisy, smelly and uncomfortable, while most are much slower too (acceleration). Also, there's Sentry Mode: https://www.youtube.com/watch?v=gRGarDcUELE
LOL. Seriously. Can we please stop with this asinine assumption that only people who did not try a Tesla can possibly dislike them? It got old 5 years ago.
Sure, Tesla has pretty good acceleration (other than the roadster, nothing you couldn't get with an ICE). When I was 20 years younger, I might have even cared. Now I am somehow more interested in meaningful comforts and conveniences, and outside of the SV gaggle of kids who thing that being a computer on wheels, with a big screen means that Tesla is a "luxury" car, presenting even Model S as somehow competing with any of the big name luxury brands is just laughable. Hell, even my wife's Hyundai plug-in has things that you can't get in any Tesla. For half the price. Well, but Tesla has fart jokes. I'll take colled seats, thank you very much.
I wouldn't even start with the AP, it's not any better than anybody else's driver assist systems. Except when it drives you into a fire truck or median barrier, of course. Then it could be considered a natural selection tool, I guess /s
FWIW, in the same period I did 2 2000+ mile drives, and a few in 300-500 range, and in all of them I needed to actually get where I was going in some reasonable time. Maybe with stops at some vista points, but those generally do not have superchargers anywhere near. Could I make it in a Tesla? Sure. I could make it in a Leaf probably, too. It would just take significantly more time, and I would need to spend much more time at places that have chargers, and those (just like places with gas stations on the Interstate) usually aren't all that exiting.
For Tesla's target market, people with more money than actual need for a car as a tool, it actually all makes sense and works quite well. But I somehow doubt that this market is large enough to "drive German automakers out of business in 2 years" as some claim here, or even to make a significant dent in ICE car sales, absent some drastic legal or tax pressure.
> Can we please stop with this asinine assumption that only people who did not try a Tesla can possibly dislike them? It got old 5 years ago.
Still works with everyone I tried. Do you know what really got old something like 7 years ago? The claim that you can't drive long distances in a Tesla. Hansjörg Gemmingen just reported hitting 900000 Km in his Model S (from 2015 I believe) after driving ~50000 Km in 2 months.
> Tesla has pretty good acceleration (other than the roadster, nothing you couldn't get with an ICE)
OK, how much is an ICE that gets the 2.6s 0-100 Km/h of the Model S Performance? And what comfort does it offer?
> FWIW, in the same period I did 2 2000+ mile drives
In 1 day? Impressive. Or we are talking about completely different things.
> But I somehow doubt that this market is large enough to "drive German automakers out of business in 2 years" as some claim here, or even to make a significant dent in ICE car sales, absent some drastic legal or tax pressure.
German newspapers already reported on recent losses by Daimer, BMW and explicitly mentioned EV (and their failing transition) as one reason.
Didn't work with me, and I like shiny new things as much as anyone. But being old and cynical I want that shiny to be actually useful for something.
I've never heard anyone say that you can't drive long distance in an electric car. The difference is how long will it take you.
Isn't 2.6 sec 0-60 something you only get with the "ludicrous" upgrade, where it counts how often you actually use it? There are faster ICE cars.
Good God, no, I don't think you can even do 2000 miles in a day (if you ever sleep) while staying even remotely legal. But I did quite a few San Jose to Las Vegas drives in 8 hours. 2000+ were when I first drove my wife's car from San Jose to Austin, which took me 2 days. Bloody thing goes 600+ miles on a tank, which is more than I can take. Could that be done in a Tesla? Sure, but it would take longer and involve planning the trip around chargers even if they were everywhere. Second time I was moving my car, which is technically faster, but has more reasonable mileage, and made two long stops, but only because instead of morning, as planned, left in the evening (apparently if an SV apartment complex says that the rental office is open on a Saturday, it does not mean that anyone will actually be there to give the keys to). It was still much faster than would be possible in anything you need to charge. And somehow all the interesting little spots where I felt like stopping and taking pictures didn't have any Superchargers anywhere near. Nor, often, any gas stations.
Automakers were regularly having losses long before there were any Tesla to blame.
It's pretty clear that ICE will be replaced by something, quite possibly electric, at some point in the future, and they need to prepare for it, but pretending that in 2 years electric cars will be anything but a statistical error in the number of cars sold, let alone that anyone is going out of business over it, is just a load of wishful thinking SV BS.
Then, of course, I just don't understand how anyone can, in a good consciousness, give Elon even a red cent.
My understanding is that for road safety, getting out of the car every 2-3 500-600 km in a single day. On the extremely rare occasions that I do; I'll plan to have some food in between and will likely end up stopping multiple times. That leaves plenty of room to charge. I don't think I've ever filled up a car and then drove it non stop until I nearly ran out of fuel. And I've driven some shitty cars with quite poor range.
Anyway, range anxiety would be justified if it was hard to find a charging point and or if you'd have barely enough range to get to the next charging point. Neither is the case. Your car will tell you how much juice you have left and where you can get some more. Unlike gas stations, charging points are all over the place and not bottle necked on fuel truck deliveries to handful of locations. So, there's no need to queue up to charge.
I think most normal drivers on a cross country trip need bathroom breaks, lunch/snacks/drinks, leg stretching etc. every 2-3 hours. Great opportunity to plug in and recharge enough for another 2-3 hours. Range anxiety is mostly irrational and mostly a non technical problem as your car probably has more endurance than you.
I agree with the rest of your post, but this is completely untrue. No one trying to do serious distance stops every 2-3 hours, and especially not for the 20 minutes for a Tesla 50% supercharge top up. Maybe 10 minutes every 3-4 hours is realistic.
Well the inconvenience of having to stop for an extra 30-40 minutes over a journey that basically is going to eat your whole day in any case might be considered a show stopper for some hard core road warriors. But, for normal people, that sounds like an entirely reasonable tradeoff. Especially when you consider that the fuel cost you'll be saving easily pays for a nice motel + dinner. But yes, if you are traveling beyond the range of your EV, you will at some point have to plan a longer break.
> I think most normal drivers on a cross country trip need bathroom breaks, lunch/snacks/drinks, leg stretching etc. every 2-3 hours.
I don't think this is true. I recently drove 12 hours to Austin, TX and I stopped only once to get gas after 6 hours. I don't really like to waste a bunch of time on stops when I have a destination in mind.
That said, I don't have a family or anything, so I could see how that changes when you have a couple children with you.
I know a lot of people who own EVs. Every single one of them also has a gas vehicle in the family. No one uses EVs for road trips. Afaict they don't belong to single-car households.
Geo-fenced with some limitations (ie unable to drive in severe weather). So it'll only work in some areas, and will likely have further limitations like avoiding alleys, roundabouts, and whatever else it finds difficult but it not necessary to navigate from point A to point B. All of this is fully driverless and you can sleep in the back of the car.
SDS stands for Self-Driving Service. SAE stands for Society of Automotive Engineers.
SDS usually stands for self-driving system. Argo provides only the sensors+compute HW+software that can be integrated into different cars from different manufacturers
It's a fairly high level of automation. One step below total automation. That is, the system controls all aspects of driving, but there are still situations where it will ask the human driver to take over. However, even if the human doesn't actually take over, the system can still handle it. It's basically capable of dealing with an odd situation, but it's not confident about it.
If the car is able to function even when noone responds to the request for intervention, how is that different from level five? Does it mean that it safely pulls over, where level 3 would just drop control of the vehicle in a dangerous scenario?
Also, does level 5 entail general intelligence, in order to do things like follow the instructions of a traffic cop or parking attendant?
Yes (or just stop - the idea is that it will not crash).
> general intelligence
Your examples would most likely be situations that would be trained for. I doubt that any system in the next few (5-50) years will be able to adopt to a completely new situation as well as a human.
Yes, but there is still vagueness. Depending upon where you are and the circumstances, there may be no "safely pull over." I've certainly gotten into very low visibility situations on highways (snow squall, very heavy rain) where put on hazards and crawl along as best you can and pull over to the side of the road are pretty much equally bad alternatives.
I don't think this is a problem with the classification, I think this is a problem with how hard it is to build a self-driving car even if you give it these limitations. You're just making a point for how hard a Level 4 system is to build, not how hard it is to define.
That article misrepresents what Musk actually said. The actual conversation was more like: Q) When will you be able to fall asleep and let the car drive? A) Uh dunno maybe late 2020.
Even if you believe in Musk, he tends to run late so late 2020 seems optimistic. Assuming you are being safety conscious - some idiot will probably do it anyway and post to youtube.
It seems to me that self driving systems could do with an algorithmic breakthrough or two before they work satisfactorily. If you look at the Autonomy event video https://youtu.be/Ucp0TTmvqOE?t=6973 where Karpathy explains how the systems work, the Japanese Spaniel problem a couple of minutes in to that is the kind of thing that needs fixing.
The problem of Perception has a long road ahead at any of these companies and Argo is taking a very long position with this funding round. The headcount growth is rather worrisome but other than Aptiv there aren't many fleets that span multiple continents.
It's best not to think in GB terms when talking about AD datasets. E.g., when you record raw data of a multisensor setup (lidars, radars, cameras), the data rate can reach 10+ TB/h. Camera-only datasets are in comparison much smaller.
Taken out of the argoverse dataset description:
- One dataset with 3D tracking annotations for 113 scenes
- One dataset with 327,793 interesting vehicle trajectories extracted from over 1000 driving hours
- Two high-definition (HD) maps with lane centerlines, traffic direction, ground height, and more
Right, the dataset is not remotely useful for commercial use. For one thing, the license prohibits such use.
Yes, it's certainly not yet clear if the dataset is large enough to capture useful variance. But unlike kitti or cityscapes, it's large enough to present a computational challenge to most of the machines & budgets in research use today, so there's a pretty good chance it'll help push the state of the art... Perhaps for more than one art. The API code itself has a lot of low-hanging fruit: https://github.com/argoai/argoverse-api
One attractive aspect of these datasets is that they help open up the question of safety for public discussion. For example, now anybody can throw off-the-shelf object detection at these datasets and see what a realistic F1-score looks like for objects at 30m, 50m, 100m, etc...
In argoverse and nuscenes, you have track labels, so you can furthermore factor the velocity difference into how you weigh the error. Have you ever been hit by an Uber? Even 5mph can cause a lot of damage.
Wouldn't it be better to use a simulated environment first. Maybe something using pybullet[1] and a script which maps real world to STL or OBJ files[2]
Today's virtual worlds are not accurate enought to allow to develop perception algorithms in simulation. In order to develop sensor fusion, you also need to simulate the output of all sensors including their specific characteristics. Apart from the model quality, there is another challenge: Simulation runtime, which is substantial (!) and - to my best knowledge - not even close to realtime.
If you want to develop driving algorithms that sit on top of the perception stack, then this becomes simpler. You can work on the object level (object being simulated cars, pedestrians, ...) and statistically model perception errors. This is a lot faster, which is e.g. important, if you want to run large-scale reinforcement learning to develop your driving strategy.
In any case, in the future, I would say that we will see a lot more simulation (don't forget, all major players build heavily on simulation - just look into the numbers on how many miles Waymo simulates every day) and potentially going down all the way to the sensing level, because it allows you to develop and especially debug along the whole sense - plan - act stack.
Also, today there exist also hybrid approaches. You take real recordings and abstract them into a simulatable format that you can then, e.g., use to variate and derive artificial scenarios for simulation. This can be used to analyse the influence of different situative paramters on the behavior of a function to pinpoint which parameter(s) caused certain troublesome behavior that have been observed in real drives.
How would something like Waymo's daily simulation training work? Is it just feeding it a larger set of random obstacles every day?
Wouldn't it quickly have negligible returns once it optimizes for the current simulation capabilities? Or are they constantly tweaking both the car model and the simulation data set?
That videos mesmerizing. It’s like perfection in driving. I can imagine the future when we’re traversing highways at 200km/hr that speed once cars are all networked.
At least you will have autnomous traffic jam assistants much earlier, which allows you to spend more time on HN while stuck in traffic on your way to work on true self-driving vehicles ;)
Incumbent car companies seem to have recently realized they are about ~2 years away from being put out of business by Tesla. In the past two weeks we've seen the CEO of BMW step down with speculation it was over BMW's botched EV strategy. Now the VW / Ford alliance on electric and self driving.
They clearly are at no risk of being put out of business by Tesla. All electric car sales are still only a fraction of ICE and given Tesla’s stagnant sales nuns it looks like it will remain that way for a while.
Tech early adopters have bought and most others aren’t interested until the price falls significantly and the reliability becomes something like a civic.
Tesla's sales and sales of EV's in general look like they are in the early days of an exponential growth curve. They have been at around 100% or better YOY sales growth for the past 3 quarters. At that rate it will only take a few years to go from niche to dominating the market. Car sales in general are flat or in slight decline and Tesla is seeing 100% YOY growth. It doesn't take a rocket scientist to see that if that trend continues for a couple more years some of the traditional auto makers are going to go out of business. The real nail in the coffin though will be if Tesla crack full self driving by 2021 which is what they stated they think they will do.
Even if the entire world now wanted Teslas and nothing but Teslas, the manufacturing ability is just not there. Inevitably the vast majority of the market would go to other companies. That was always the relation between Tesla and established car companies: Tesla has better EV technology, but others had the better manufacturing ability by far.
In theory, this should favour the established companies - Tesla doesn't patent their EV advances, and they aren't the only EV player anyway. It should be easier to add EV tech to existing chain than to establish an entire new supply chain. But every established company is very invested in ICE, and has large internal lobbies.
It's interesting to see who gets the better of this - the established car companies' objective advantage or their internal political disadvantage. The recent moves might suggest these established companies are getting their internal politics in order, which should allow them to do EV better.
I realize that is the prevailing opinion. I think it's wrong. No incumbent out there has has yet made an electric vehicle that can even match Tesla's 2012 model S on range and performance. The electric cars they have come out with are money losers for them. Fiat's CEO famously told people he hoped no one bought the Fiat 500 Electric because the company loses $14,000 on every sale.
Other companies are behind on EV proper (aside). Commercially, it doesn't matter all that much.
Tesla can't sell nonexistent cars, and it can't increase capacity to the point of even getting near to taking over the market. The existing companies have more than enough time to fill in demand with their existing EVs - which will 'win' even if only because these models have cars which can be sold, though I wouldn't bet against the other companies having a compelling offer too. With scale profitability will come.
Of course, the existing companies could sleep and lose their chance, but this doesn't seem likely.
(aside) Note that comparison should also include price, and on that scale other EVs are already better. There should be room at the bottom of the market as well.
Given that VW was caught cheating emissions on a global scale, and far worse than other carmakers, i find it surprising that anyone on hackernews would consider buying their cars.
I can't think of a reason folks here would hate on Toyota, Honda, etc: companies that build practical, efficient, cheap cars, have well regarded engineering, and have met with incredible economic success.
False equivalency to say all car brands are like VW, which lied and cheated about emissions whilst marketing their product as Clean Diesel, and left their own customers and everyone else on planet Earth holding the bag.
I buy goods/services based on the future expected utility and effects of my purchase. If VW has a vehicle that's otherwise the best option for me to purchase, I'm not clear that there's any benefit to me, based on their past actions, opting to go with the next best alternative instead.
Boycotts come at the cost of the participants but to the benefit of the greater good. You’re being asked not to cross the picket line.
If you don’t believe VWs transgressions were enough to warrant a boycott, that’s a fair point to make. I’m not in the car market so I don’t even have an opinion on it, myself. But to the parent’s credit: they never meant to talk about you benefitting personally. It’s about teaching car makers that actions have consequences. For the greater good.
Put another way: how bad does VW have to mess up before you consider more than your personal benefit?
Toyota had an Unintended Acceleration problem that actually put peoples' lives at stake, and they denied everything / covered it up [1]. I don't recall people boycotting Toyota.
[1] Toyota to Pay $1.2B for Hiding Deadly ‘Unintended Acceleration’
They certainly didn't put the fault there by design, and over a number of years. I think VW has done irreversible damage to the environment systematically over a long number of years.
My understanding is just about every car maker does this, and VW is just the company to fall on the sword. Hopefully in a few years there will be more data and journalism on this to substantiate or refute these rumors.
Indeed, I cite a source corroborating this in my cousin post.
Every car manufacturer that is building ICE vehicles is damaging the environment. Even though VW fudged some numbers in their report like everybody else did, the fact that there are even cars on the road to begin with is by far the main contributor to global warming, much more than any incremental damage (if any) caused by people who could have been mislead into buying a "clean diesel" vehicle based on sheer (mis-)reporting.
If there was a TDI next to an F-250 on the road, and I asked someone to point to the biggest offender, everyone would point at the F-250. It's really unfortunate to see VW become the scapegoat here in the US (but unfortunately also quite predictable since they are a "foreign" car manufacturer).
I'm not saying I won't participate in a boycott, I'm saying it's not clear that a boycott of VW benefits the greater good, because a) It's not clear that current or future VW is behaving or will behave any worse than other car manufacturers. and b) There's pretty free movement in decision-makers between various car manufacturers.
I don't anthropomorphize companies; they aren't people. They can be economically incentivized or legislatively regulated, but they can't learn.
> Put another way: how bad does VW have to mess up before you consider more than your personal benefit?
There'd have to be some indication that me buying a VW would be contributing to future mess ups.
"Since the Volkswagen 'Dieselgate' emissions cheating scandal broke in September 2015, nearly every major automaker including--Daimler, General Motors, Suzuki and Mitsubishi have all been caught up in falsifying fuel economy figures or cheating on emissions testing."
I am not justifying VW's actions, but it's pretty obvious to any critical reader of news articles that VW was just the unlucky one whose head met the chopping block -- so in terms of a single targeted boycott of VW as a consumer, I don't think it makes sense. They still make great cars, despite some bad apples that work there (every company has them -- also see Lenovo / SuperFish).
If this trend continues, I really expect there to be a massive bidding war for Tesla before too long. When bears/shorts on tesla Complain about Tesla’s market cap, they seem to always forget the fact that Tesla is well ahead of any of these players in terms of both deployed hardware and data collection. This merger and investment brings Argo’s valuation up to you around $8 billion.
Tesla is ahead in many regards but they are too debt laden for a buyout at this point. A larger player would be smart to wait until the economy turns and Tesla can no longer meet debt obligations, then buy them out of bankruptcy.
> they are too debt laden for a buyout at this point
Enterprise value is indifferent to capital structure. A leverage-sensitive buyer would buy out the high-yielding debt in addition to the equity. Purchasing out of bankruptcy is riskier, in many levels, compared with a negotiated sale.
What aspect of two competitors cooperating to compete against a third in a market where there are no clear winners yet, merits an antitrust investigation?
There are already several other big SDC alliances like Honda / GM Cruise, Toyota / Uber, and even the three-way Nissan-Renault / Jaguar Landrover / Waymo.
116 comments
[ 5.2 ms ] story [ 174 ms ] threadListen to them more. What else do you expect fund managers to say?
We will not see any real self driving in our generation, and may be not even after. That is much more for practical consideration than that of technical possibility.
Second to that, just how many self driving and AI startups are plain fraud? The amount of companies getting 9 digit valuations with nothing more than OpenCV hello worlds should make people to at least scratch their heads
I think we can once we build purposeful infrastructure (think cars "on rails" with mesh-coordinated intersections).
> Second to that, just how many self driving and AI startups are plain fraud? The amount of companies getting 9 digit valuations with nothing more than OpenCV hello worlds should make people to at least scratch their heads
Argo, although lesser known (but still out of early Waymo and CMU's Robotics Institute), is every bit as competitive as the other major players. If anything, they are less bombastic than the Elon Musks of the world falsely tweeting that we would all be taking naps in our cars by last year.
In that way, surely, but you will need to build a purpose built road network for that. Some self driving buses in China work like that.
New, growing cities with centralised planning can afford that, others not so much.
I agree.
But we'll be driven a large percent of trips, in the near future.
There's on-demand shared(few people in a vehicle) riding - which can be quite cheap AND fast. ridewithvia.com seem to scale with that model pretty fast. A combination of that with strong public transport systems and regulations may beat cars.
But also, a combination of partial self-driving(highway, etc), with remote control driving, could work.
I think the engineers have been probably been wildly optimistic as usual, and the last 5% of car autonomy is going to take as much effort as the first 95%, or maybe double that.
But you know the sort of people who are wildly optimistic and prone to over-exaggeration. Executives like Musk.
Most are optimistic about highway driving in California though.
This may not be true if there is still a fundamental leap to be made
Rather hundreds or thousands of small leaps. For example solving problems like this:
https://sites.google.com/view/lidar-adv
The Munich-based AID team will become Argo’s European headquarters, a move that will expand its staff 40% to more than 700 employees."
AID has been Audi's L4/L5 development arm, which was separate from VW's main research arm. It also helps to judge the size of this deal, because VW only seems to invest one billion in capital and the remaining 1.6 billion investment is AID. It would be interesting to know, how they came up with that valuation ...
Lol, of course they said that as their entire investment strategy is to throw more dumb money at startups than anyone else can possibly match. Their vision fund VC has been investing $100 billion dollars and they're planning on doing another $100 billion dollar fund. "To put the fund’s magnitude in perspective, its size is almost double the investments made by U.S. venture firms last year. PitchBook data shows that VC fundraising in the U.S. totaled $53.9 billion last year across more than 200 funds, and that was the largest annual raise in at least a decade." [0]
[0]: https://www.cnbc.com/2019/05/17/softbanks-100-billion-vision...
I don't think this will match your expectations, but this is the best take on what you want I can think of
At some point, you can not improving the range any more by adding more batteries because any gain will be eaten by more rolling resistance.
At some point you will arrive to the point when your vehicle starts to resemble a battery stack with two wheels.
As for charging _fully_ - that's rarely done in practice, because over 80% li-ion batteries charge much slower. Bjørn Nyland recently beat the record[0] for longest distance travelled in an EV in 24h (2781km) by stopping to charge from 10% to 50%, because only then he could use the full power of superchargers.
[0] https://youtu.be/R-2Yj-uVeB0
I think most normal drivers on a cross country trip need bathroom breaks, lunch/snacks/drinks, leg stretching etc. every 2-3 hours. Great opportunity to plug in and recharge enough for another 2-3 hours. Range anxiety is mostly irrational and mostly a non technical problem as your car probably has more endurance than you.
I always thought thatd be the endgame. Some standardized battery design where you roll up on top of a station and it swaps the battery underneath the car.
Or maybe this is something your car does for you before arriving. Which eliminates the amount of dead time tens of thousands cars will need to spend while using up expensive property in cities while charging.
https://en.m.wikipedia.org/wiki/Better_Place_(company)
https://cleantechnica.com/2018/11/24/nio-builds-battery-swap...
Everytime I hear this I wonder if the people saying it have ever taken a road trip. Stopping every 2-3 hours for a half hour on a long road trip seems downright crazy to me.
This is beside the point that during the busy season of summer travel there are sometimes extensive lines at gas stations on popular routes. Imagine if every car took 15-45 minutes to fuel up?
Range anxiety is real. EVs are the ultimate commuting cars: known distance, charge at home over night. But for the immediate future, over long distance, gasoline cars are hard to beat.
It seems crazy to you because you haven't done it. It seems perfectly normal and acceptable to people who drive Teslas over long distances. I see people going for lunch, shopping, sightseeing at Superchargers. If you know you'll have to wait 20+ minutes, you just find something to do instead of biting your fingernails.
You'll find out eventually.
Sure, Tesla has pretty good acceleration (other than the roadster, nothing you couldn't get with an ICE). When I was 20 years younger, I might have even cared. Now I am somehow more interested in meaningful comforts and conveniences, and outside of the SV gaggle of kids who thing that being a computer on wheels, with a big screen means that Tesla is a "luxury" car, presenting even Model S as somehow competing with any of the big name luxury brands is just laughable. Hell, even my wife's Hyundai plug-in has things that you can't get in any Tesla. For half the price. Well, but Tesla has fart jokes. I'll take colled seats, thank you very much.
I wouldn't even start with the AP, it's not any better than anybody else's driver assist systems. Except when it drives you into a fire truck or median barrier, of course. Then it could be considered a natural selection tool, I guess /s
FWIW, in the same period I did 2 2000+ mile drives, and a few in 300-500 range, and in all of them I needed to actually get where I was going in some reasonable time. Maybe with stops at some vista points, but those generally do not have superchargers anywhere near. Could I make it in a Tesla? Sure. I could make it in a Leaf probably, too. It would just take significantly more time, and I would need to spend much more time at places that have chargers, and those (just like places with gas stations on the Interstate) usually aren't all that exiting.
For Tesla's target market, people with more money than actual need for a car as a tool, it actually all makes sense and works quite well. But I somehow doubt that this market is large enough to "drive German automakers out of business in 2 years" as some claim here, or even to make a significant dent in ICE car sales, absent some drastic legal or tax pressure.
Still works with everyone I tried. Do you know what really got old something like 7 years ago? The claim that you can't drive long distances in a Tesla. Hansjörg Gemmingen just reported hitting 900000 Km in his Model S (from 2015 I believe) after driving ~50000 Km in 2 months.
> Tesla has pretty good acceleration (other than the roadster, nothing you couldn't get with an ICE)
OK, how much is an ICE that gets the 2.6s 0-100 Km/h of the Model S Performance? And what comfort does it offer?
> FWIW, in the same period I did 2 2000+ mile drives
In 1 day? Impressive. Or we are talking about completely different things.
> But I somehow doubt that this market is large enough to "drive German automakers out of business in 2 years" as some claim here, or even to make a significant dent in ICE car sales, absent some drastic legal or tax pressure.
German newspapers already reported on recent losses by Daimer, BMW and explicitly mentioned EV (and their failing transition) as one reason.
I've never heard anyone say that you can't drive long distance in an electric car. The difference is how long will it take you.
Isn't 2.6 sec 0-60 something you only get with the "ludicrous" upgrade, where it counts how often you actually use it? There are faster ICE cars.
Good God, no, I don't think you can even do 2000 miles in a day (if you ever sleep) while staying even remotely legal. But I did quite a few San Jose to Las Vegas drives in 8 hours. 2000+ were when I first drove my wife's car from San Jose to Austin, which took me 2 days. Bloody thing goes 600+ miles on a tank, which is more than I can take. Could that be done in a Tesla? Sure, but it would take longer and involve planning the trip around chargers even if they were everywhere. Second time I was moving my car, which is technically faster, but has more reasonable mileage, and made two long stops, but only because instead of morning, as planned, left in the evening (apparently if an SV apartment complex says that the rental office is open on a Saturday, it does not mean that anyone will actually be there to give the keys to). It was still much faster than would be possible in anything you need to charge. And somehow all the interesting little spots where I felt like stopping and taking pictures didn't have any Superchargers anywhere near. Nor, often, any gas stations.
Automakers were regularly having losses long before there were any Tesla to blame.
It's pretty clear that ICE will be replaced by something, quite possibly electric, at some point in the future, and they need to prepare for it, but pretending that in 2 years electric cars will be anything but a statistical error in the number of cars sold, let alone that anyone is going out of business over it, is just a load of wishful thinking SV BS.
Then, of course, I just don't understand how anyone can, in a good consciousness, give Elon even a red cent.
Anyway, range anxiety would be justified if it was hard to find a charging point and or if you'd have barely enough range to get to the next charging point. Neither is the case. Your car will tell you how much juice you have left and where you can get some more. Unlike gas stations, charging points are all over the place and not bottle necked on fuel truck deliveries to handful of locations. So, there's no need to queue up to charge.
I agree with the rest of your post, but this is completely untrue. No one trying to do serious distance stops every 2-3 hours, and especially not for the 20 minutes for a Tesla 50% supercharge top up. Maybe 10 minutes every 3-4 hours is realistic.
I don't think this is true. I recently drove 12 hours to Austin, TX and I stopped only once to get gas after 6 hours. I don't really like to waste a bunch of time on stops when I have a destination in mind.
That said, I don't have a family or anything, so I could see how that changes when you have a couple children with you.
Perhaps naive question from an outsider to the AV world: what does "SAE Level 4-capable SDS" mean?
SDS stands for Self-Driving Service. SAE stands for Society of Automotive Engineers.
https://www.sae.org/news/press-room/2018/12/sae-internationa...
It's a fairly high level of automation. One step below total automation. That is, the system controls all aspects of driving, but there are still situations where it will ask the human driver to take over. However, even if the human doesn't actually take over, the system can still handle it. It's basically capable of dealing with an odd situation, but it's not confident about it.
If the car is able to function even when noone responds to the request for intervention, how is that different from level five? Does it mean that it safely pulls over, where level 3 would just drop control of the vehicle in a dangerous scenario?
Also, does level 5 entail general intelligence, in order to do things like follow the instructions of a traffic cop or parking attendant?
Yes (or just stop - the idea is that it will not crash).
> general intelligence
Your examples would most likely be situations that would be trained for. I doubt that any system in the next few (5-50) years will be able to adopt to a completely new situation as well as a human.
[1] https://www.wired.com/story/elon-musk-tesla-full-self-drivin...
https://bgr.com/2019/07/09/tesla-autopilot-team-elon-musk-sh...
Even if you believe in Musk, he tends to run late so late 2020 seems optimistic. Assuming you are being safety conscious - some idiot will probably do it anyway and post to youtube.
It seems to me that self driving systems could do with an algorithmic breakthrough or two before they work satisfactorily. If you look at the Autonomy event video https://youtu.be/Ucp0TTmvqOE?t=6973 where Karpathy explains how the systems work, the Japanese Spaniel problem a couple of minutes in to that is the kind of thing that needs fixing.
https://electrek.co/2015/12/21/tesla-ceo-elon-musk-drops-pre...
The problem of Perception has a long road ahead at any of these companies and Argo is taking a very long position with this funding round. The headcount growth is rather worrisome but other than Aptiv there aren't many fleets that span multiple continents.
Taken out of the argoverse dataset description: - One dataset with 3D tracking annotations for 113 scenes - One dataset with 327,793 interesting vehicle trajectories extracted from over 1000 driving hours - Two high-definition (HD) maps with lane centerlines, traffic direction, ground height, and more
1000 driving hours is ok'ish for research (imho).
Yes, it's certainly not yet clear if the dataset is large enough to capture useful variance. But unlike kitti or cityscapes, it's large enough to present a computational challenge to most of the machines & budgets in research use today, so there's a pretty good chance it'll help push the state of the art... Perhaps for more than one art. The API code itself has a lot of low-hanging fruit: https://github.com/argoai/argoverse-api
nuscenes is good too https://www.nuscenes.org/
Waymo will have one as well some day https://waymo.com/open/
One attractive aspect of these datasets is that they help open up the question of safety for public discussion. For example, now anybody can throw off-the-shelf object detection at these datasets and see what a realistic F1-score looks like for objects at 30m, 50m, 100m, etc...
In argoverse and nuscenes, you have track labels, so you can furthermore factor the velocity difference into how you weigh the error. Have you ever been hit by an Uber? Even 5mph can cause a lot of damage.
Wouldn't it be better to use a simulated environment first. Maybe something using pybullet[1] and a script which maps real world to STL or OBJ files[2]
1. https://pybullet.org/wordpress/ 2. https://github.com/mkagenius/osm2maya
Today's virtual worlds are not accurate enought to allow to develop perception algorithms in simulation. In order to develop sensor fusion, you also need to simulate the output of all sensors including their specific characteristics. Apart from the model quality, there is another challenge: Simulation runtime, which is substantial (!) and - to my best knowledge - not even close to realtime.
If you want to develop driving algorithms that sit on top of the perception stack, then this becomes simpler. You can work on the object level (object being simulated cars, pedestrians, ...) and statistically model perception errors. This is a lot faster, which is e.g. important, if you want to run large-scale reinforcement learning to develop your driving strategy.
In any case, in the future, I would say that we will see a lot more simulation (don't forget, all major players build heavily on simulation - just look into the numbers on how many miles Waymo simulates every day) and potentially going down all the way to the sensing level, because it allows you to develop and especially debug along the whole sense - plan - act stack.
Also, today there exist also hybrid approaches. You take real recordings and abstract them into a simulatable format that you can then, e.g., use to variate and derive artificial scenarios for simulation. This can be used to analyse the influence of different situative paramters on the behavior of a function to pinpoint which parameter(s) caused certain troublesome behavior that have been observed in real drives.
Wouldn't it quickly have negligible returns once it optimizes for the current simulation capabilities? Or are they constantly tweaking both the car model and the simulation data set?
Cue up the [0] Porsche 919 vid to show what a human can do in a car...
[0] https://youtu.be/KsLi7HgSuhI
Tech early adopters have bought and most others aren’t interested until the price falls significantly and the reliability becomes something like a civic.
In theory, this should favour the established companies - Tesla doesn't patent their EV advances, and they aren't the only EV player anyway. It should be easier to add EV tech to existing chain than to establish an entire new supply chain. But every established company is very invested in ICE, and has large internal lobbies.
It's interesting to see who gets the better of this - the established car companies' objective advantage or their internal political disadvantage. The recent moves might suggest these established companies are getting their internal politics in order, which should allow them to do EV better.
Tesla can't sell nonexistent cars, and it can't increase capacity to the point of even getting near to taking over the market. The existing companies have more than enough time to fill in demand with their existing EVs - which will 'win' even if only because these models have cars which can be sold, though I wouldn't bet against the other companies having a compelling offer too. With scale profitability will come.
Of course, the existing companies could sleep and lose their chance, but this doesn't seem likely.
(aside) Note that comparison should also include price, and on that scale other EVs are already better. There should be room at the bottom of the market as well.
False equivalency to say all car brands are like VW, which lied and cheated about emissions whilst marketing their product as Clean Diesel, and left their own customers and everyone else on planet Earth holding the bag.
If you don’t believe VWs transgressions were enough to warrant a boycott, that’s a fair point to make. I’m not in the car market so I don’t even have an opinion on it, myself. But to the parent’s credit: they never meant to talk about you benefitting personally. It’s about teaching car makers that actions have consequences. For the greater good.
Put another way: how bad does VW have to mess up before you consider more than your personal benefit?
[1] Toyota to Pay $1.2B for Hiding Deadly ‘Unintended Acceleration’
https://abcnews.go.com/Blotter/toyota-pay-12b-hiding-deadly-...
Every car manufacturer that is building ICE vehicles is damaging the environment. Even though VW fudged some numbers in their report like everybody else did, the fact that there are even cars on the road to begin with is by far the main contributor to global warming, much more than any incremental damage (if any) caused by people who could have been mislead into buying a "clean diesel" vehicle based on sheer (mis-)reporting.
If there was a TDI next to an F-250 on the road, and I asked someone to point to the biggest offender, everyone would point at the F-250. It's really unfortunate to see VW become the scapegoat here in the US (but unfortunately also quite predictable since they are a "foreign" car manufacturer).
I don't anthropomorphize companies; they aren't people. They can be economically incentivized or legislatively regulated, but they can't learn.
> Put another way: how bad does VW have to mess up before you consider more than your personal benefit?
There'd have to be some indication that me buying a VW would be contributing to future mess ups.
I am not justifying VW's actions, but it's pretty obvious to any critical reader of news articles that VW was just the unlucky one whose head met the chopping block -- so in terms of a single targeted boycott of VW as a consumer, I don't think it makes sense. They still make great cars, despite some bad apples that work there (every company has them -- also see Lenovo / SuperFish).
This is far from being obvious given that the competition is collecting data with much more advanced sensors.
Enterprise value is indifferent to capital structure. A leverage-sensitive buyer would buy out the high-yielding debt in addition to the equity. Purchasing out of bankruptcy is riskier, in many levels, compared with a negotiated sale.