Waymo is patronized by Alphabet for as long as it needs to be. This has been good, as it takes off a lot of pressure to virtue signal fake progress to impatient investors; a fate many of Waymo's potential competitors haven fallen victim to.
When Uber and Tesla were acting like cowboys and creating plenty of drama for the media Waymo (under the sensible and boring leadership of non-techy car industry vet John Krafcik) was quietly structuring itself to be in the game for the long haul, following the whole "Fuck You Money" episode that saw most of the Google Self Driving Car Project's original talent leave for easy Venture Capital elsewhere.
Waymo is very tight lipped about their operational costs, I would really love to know how close they are to being revenue positive. Recent moves by Waymo suggest they believe they are getting close.
They're doing 100k paid trips/wk now, per the CEO in a recent LinkedIn post. If average trip value is (very generously) $20, that's still only $100MM in annual revenue. I would bet that doesn't even pay for the direct physical + digital costs of operating the cars, let alone salaries, office leases, etc.
Waymo has started onto a very good trendline now, for sure, and a path to revenue-positive is becoming clearer. But they have a loooooooong way to go before they are offsetting even their operating costs, let alone beginning to pay down their (absolutely massive) fixed costs.
I think the bear case is actually pretty tough for Waymo. Their scaling is not cheap.
- The cars are, by all account, expensive as hell; I've seen $200k+
- The cars require ongoing cleaning, maintenance, mechanical work
- There's a massive customer support operation; not just people to call for support when your car gets stuck, but people who can step in and remote-instruct the vehicle.
- Expanding to a new city requires a massive data-gathering operation before any paid trips can begin.
None of this is taking into account the massive fixed costs that come with being a tech company - hundreds of $400k+TC employees, server bills in the tens of millions, thousands of top-of-the-line GPUs for training clusters.
Given how fast Waymo is scaling, we should look at forward revenue projections, if available. Also, when it comes to TAM, Uber is a good standard to compare against. In Q2 2024[1], Uber had
- 10.7B revenue (42.8B ARR), out of which 6.134B (24.5B ARR) was from "mobility" and others from delivery/freight.
- Real fun part: Gross booking revenue for mobility was 20.554B, but revenue for Uber itself was 6.134B, or 30% of gross revenue. If 70% is going to drivers, that is Waymo's opportunity. It can price at 40% discount to Uber's prices while still reaping twice the revenue per ride. Start with the most lucrative markets where Uber is already profitable and expand from there.
Waymo has to recoup their research costs which are far from trivial, so I doubt prices will drop. Futhermore, there's an agreement for Uber to run the Waymo's in their app, futher complicating the calculus
Consumer cars really aren't meant to last over 100k miles. I know a lot of uber drivers will put 30k miles or more on per year. Also if a car becomes a standard widget made to be easily repairable with parts that are standard widgets, repairs and maintenance would be cheaper.
E.g. If you needed to clean the carpet in a vehicle, make the carpet easily removable and washable in an industrial washing machine.
> Consumer cars really aren't meant to last over 100k miles
What on earth are you talking about? A Honda Civic or Toyota Corolla will easily last 200k miles. In places with salt and rust, the body will probably fall apart before the engine and transmission give out.
"Many non-commercial vehicles (both auto and truck) have exceeded 1,000,000 miles (1,600,000 km). For instance, in 2013, East Patchogue, New York resident Irv Gordon (1940-2018) had accumulated 3,000,000 miles (4,800,000 km) in his 1966 Volvo P1800. The car had amassed 3,200,000 miles (5,100,000 km) by Gordon's death on 15 November 2018."
https://en.wikipedia.org/wiki/Car_longevityhttps://autos.yahoo.com/news/the-first-car-to-3-million-mile...
Honestly, I'd not use the word many, perhaps a significant number....
There is absolutely no way they are remotely close to being revenue positive (unless you're asking strictly if they are charging more than the gas they're burning without counting anything else). They have like 700 very expensive taxis and a burn rate measured in billions of dollars per year.
> (unless you're asking strictly if they are charging more than the gas they're burning without counting anything else).
Yes, I assume they're asking about the operating margin. That's the number everybody cares about. If it's positive, then Waymo can theoretically become revenue positive by adding more cars and regions to their fleet.
That number needs to include car depreciation + any remote 'safety drivers' + the maintenance depot overhead + fuel and insurance + any other costs that scale with the number of cars then.
Exactly. They have revenue in the low 6 figures per vehicle annually, so they probably need to get the number of operational employees per vehicle down to around 1/10. Seems feasible, but they're probably not anywhere close to that yet.
On the other hand, HN has consistently underestimated them for the last two years or so as they've begun to commercialize. Their trajectory has been 10x growth in rides served per year for a while now. At the current scale it's hard to imagine they could grow another 10x in the next year without being close to break even on operating cost.
10x growth would be ~$1B revenue, 7000 cars. Let's assume it requires doubling head count from 2500 to 5000. That's $130K/car, $250K/employee. sounds break-even-ish.
Yep, I look forward to reading good books about it in a decade or so. I think there are lots of lessons to be learned here for how to do a certain kind of difficult new thing.
I'd like to read that book too... but "how to do a certain kind of difficult new thing" in their case involved more than solid execution. It also needed an owner/investor with extremely deep pockets and technical capabilities that vastly accelerated their work, who was willing to throw huge amounts of money at the problem for years, along with favorable economic conditions (zero-percent interest rates) for long periods of time, at the right period of time. I give Waymo tons of credit for their work here, but the conditions that enabled it aren't easily repeatable IMO.
The reason it's interesting (IMO) is that other companies that also had owner/investors with extremely deep pockets and technical capabilities have been throwing huge amounts of money at it in the same favorable interest rate environment, but with seemingly a lot less success.
I think their (apparent) success probably does have a lot to do with some combination of having the deepest pockets and (I think?) being a few years earlier to get started than everyone else, but still, it also looks like they had a better strategy, and I'm pretty curious how that worked.
It is fascinating how Google simultaneously provides a great lesson in how to leverage their primary competency of AI, as well as how not to (losing to OpenAI in LLMs).
> Waymo is very tight lipped about their operational costs, I would really love to know how close they are to being revenue positive. Recent moves by Waymo suggest they believe they are getting close.
What specific "moves" by them indicate they are close to being profitable? All the industry research I get indicates they are still negative to the tune of $1B+ a year.
Scaling out now helps along their regulatory moat, being the first trusted operator in a given market means they get a seat at the table for setting the rules
Considering just how many of their cars I see driving around Atlanta covered in sensor arrays every day, I would certainly hope they’re making real progress.
Anyone knows whether there are fundamental differences in the approach waymo took vs tesla, or are they following the same algorithms and techniques, with more polishing or experience in the "small details that makes all the difference" ?
This an explanation from a layman on the auto side.
Tesla has focused on attempting to solve all cases using vision and generalize it as much as possible. The pro is you would be able to drop FSD capabilities anywhere in the US and it will work. The con is the tail of edge cases take significantly more effort then the first 80%.
Waymo instead of being generalized, works off of a detailed map for each region they drive in. They have a complete expectation of every detail on those roads and so then only need to account for the dynamic unknowns on the road. The pro is that they are getting to hands off driving sooner. The con is that anywhere its deployed needs detailed mapping and for that mapping to be kept up to date.
> Waymo instead of being generalized, works off of a detailed map for each region they drive in. They have a complete expectation of every detail on those roads and so then only need to account for the dynamic unknowns on the road.
This isn't entirely true. Everything at Waymo is built to be generalizable. Engineers and execs at Waymo have said to multiple times, as recently as yesterday [1].
They are able to drive just fine without up-to-date maps as well. It's an assumption built in to the system. The cars are also self-mapping. Ultimately, maps are just another input to the general driving software. The same driving software is deployed across all their cities.
A cache is exactly how I explain maps to people. They are saving precious compute cycles not computing the same static objects over and over again, by N number of vehicles.
A con does not have to be a flaw. But I would see it as a tradeoff. With all of these companies its hard to sometimes see beneath the curtain of what they promote.
They test in many cities (Buffalo, Tahoe, Bellevue, NYC) . But they won't deploy to any cities which they haven't mapped because they consider it critical to safety.
So the answer is no and that's why I outline it as a potential con. They are unable to just drop deployment into a new city, they as a matter of history, map out the entire city and keep an active map of the driving territory. I have no idea who wins in this dog and pony show but I think its a valid potential con and a easy way to see some of the decision differences in companies.
Does Google Earth/Maps/StreetView not count as mapping? I was feeling certain that all that infrastructure and data was invested into Waymo. It would seem that Google had a huge advantage from running mapping/camera cars for years and years, before SDC were even a glimmer in their eyes.
The mapping that waymo uses is much more detailed than streetview, the biggest difference being that it includes lidar data. I wouldn't be surprised if they combined the two for some of their latest data updates, but the pre-existing streetview data is not obviously enough to give an advantage.
Obtaining a detailed map doesn't seem difficult. This is the same company that created Street View. E.g. SF has ~1000 miles of streets [1], that would only take 50 hours of paying for manual driving to map all of SF at 20mph.
Never said it was difficult but that their driving fleet relies on a lidar map of the region it serves. Its a definite trade-off so I would list it as a con but it does not mean one is better than the other.
And probably because of this strategy, by the time their system is successfully deployed in several cities, the technology to make fully self driving has become much more accessible and they can relatively easily add that incrementally.
The other pro is cost: cameras are much, much cheaper than LiDAR. Tesla is making the bet that they can close the performance gap between cameras and LiDAR faster than the cost of LiDAR will come down.
I wonder if this will ever be a significant factor. How much can a LiDAR setup cost? 4k? Maybe 2k if built in-house at car manufacturer scale in the long term? Is that significant when buying a whole car?
I’d advise you to ignore lay explanations of the space - outside of the industry most of the discourse about self driving cars is poisoned by Elon’s deceptive presentations and his followers who parrot what he says.
If you want a grounded explanation of how Tesla’s stack works, follow @greentheonly on twitter. He’s a Tesla reverse engineer who regularly posts about the software that’s actually running on the car.
If you want an explanation about how real AV companies stacks work, I’d read Sebastian Thrun’s robotics textbook - then imagine what’s outlined in that book but with ML plugged in to a ton of spaces throughout the stack. This is also similar to how Tesla’s stack works, btw - greens just good to follow because a lot of people refuse to believe Tesla isn’t running some kind of “LLM but for driving” fully end to end black box model.
Tesla won’t launch a robotaxi anytime soon because they can’t use remote support or HD maps - although I think they’ve been stepping up their mapping efforts. Even the demo at Universal studios a few weeks ago was HD mapped - per @greentheonlys twitter.
I worked in the space for years and have seen the internal of both a traditional robotaxi company’s stack and Tesla’s.
For reference, Sebastian Thrun led the Stanford team that won the Darpa (self driving car) Grand Challenge in 2005, and then joined Google to lead Waymo (then called the Google Self-Driving Car Project), among other accomplishments.
Tesla chose to recklessly endanger lives, killing at least 50 people by releasing an unsafe beta test. Their goal was to capture market share from Waymo.
This is the thing that drives me nuts: wantonly ignoring regulations designed to protect consumers, and claiming that it was going to save more people than it hurt, in the long run. Tesla's participation in the market has not and will not make self driving arrive earlier than if they had refrained from participating. All it has done is given them an opportunity to compete for market share, and engage in pump and dump scams by misrepresenting their progress to naive investors.
People will tell you they're fundamentally different, but they are in fact the same. First, there are two different, independent aspects: Perception and Action. Seeing the world, vs taking action in that world.
For perception Waymo uses more sensors than Tesla. It uses lidar to construct a 3D scene of what the car is seeing, while Tesla uses SLAM-like techniques with their cameras. What people are missing is that these are only a small part of the perception problem - lidar returns a monochromatic 3D scene so it cannot see labels, markings, read signs, lights etc. LIDAR simply doesn't carry the necessary information needed to navigate the world, thus it is a secondary source of info. Cameras in motion do carry all the information needed, so the big difference between them is only one part of the overall perception stack.
Once you have constructed a labeled, accurate 3D scene (whether by lidar or SLAM), "action" is the same and there is no difference between Tesla and Waymo here. They both have to learn how to drive safely using the same information, so it's going to be a lot like LLMs where there's difference between LLAMA/Claude/GPT, but they're also all kinda the same thing.
The fact that you don’t know the difference in capabilities between cameras and LiDAR means you shouldn’t really be commenting. LiDAR allows you to also ‘see’ things in conditions where weather is bad visibility is limited. Cameras cannot allow you to do this.
Going through this thread, it is mind blowing how people let their fanboyism for others talk nonsense. So many Elon lovers here. Can all those Elon lovers just jump in the backseat of a Tesla and turn on FSD (which is an oxymoron) and jump on a high way or something? The world would be a better place. I can’t wait to see their faces and a the drivel they will come up with next year when Elon announces delays and other bull**t fit why he can’t launch what he promised. It’s going to be hilarious.
Tesla is taking a minimalist approach to every important factor: Specialized vs unmodified production cars; back seat in-vehicle UI + app vs app-only for sitting in back; multiple sensor types vs camera-only; intensive mapping of a service area vs crowdsourced; only Waymo-owned vehicles vs private cars seconded to the service.
The biggest small detail is that Waymo's expansion is gated on achieving performance goals like very little need for remote supervision because Google won't hire a building full of remote monitors. Tesla claims this will be possible "by next year."
Fundamentally different - people will flock to LiDAR vs Vision when debating this, but the more fundamental difference is that Waymo uses HD mapping + remote support.
Thanks, i was expecting something beyond just the lidar thing.
Can you elaborate more about mapping + remote ? I've never tested neither a tesla autopilot nor a waymo (european here).
edit : also, are there differences in the core algorithms ? Tesla seems to be full AI / ML. Is waymo the same ? ( as the company is older i wonder if they haven't built more things manually)
Neither of them are “full AI/ML”, they’re both traditional robotics systems with ML used for detection/prediction/planning at certain steps. Elon will sometimes say something about moving to a “new ML stack”, but Tesla reverse engineers regularly look inside of what’s running in the cars and that’s not the case at all.
Contrary to what other people in this thread are saying, the remote support isn’t remote direct driving of the car - essentially what will happen is that if the car finds itself in a situation where it’s unsure of how to proceed and it’s safe to stop, it will pause for a few seconds and wait for a remote operator to clarify a situation for it.
A good example might be road construction - if the car detects new road construction work that doesn’t match its map of the area, and its onboard systems determine that it’s not sure how to proceed through the construction with confidence, it will send what it thinks the top five likeliest ways to proceed to a remote operator. The operator then selects the proper path (or says that none of them are proper). The car will then follow the path presented by the operator, but actual driving behavior /collision detection / pathfinding is still determined locally. Think of it like ordering a unit around in StarCraft.
You can actually see this behavior in the car when it runs into a difficult situation. It tells you it's asking for assistance or something similar, and pauses for a few seconds.
I also made the mistake of assuming the remote operator drives the car but if you watch Waymo's technical videos, it's clear that the AI is in control of the car all all times and the remote operator is just doing near real time labelling of what the car is seeing.
Same way reverse engineer get access to internals of other devices - a bunch of tricks. :^)
In one case, greentheonly realized some fraction of Tesla’s cars are shipped out of the factory still in dev mode, with debug mode enabled and increased privileges. He found someone with a car like this who was down to helped and swapped part of his cars hardware with their car, and from then on was able to get a much better view of what was running on his car.
Unfortunately twitter is awful to search and a lot of his info is buried deep in old threads, but a few (old) examples to illustrate that he regularly does this.
It's a fallacy to really compare them directly. Waymo has active cars in service that are available to the general public. Tesla right now just has promises that they have been unable to deliver on for almost a decade now, as their approach is pretty radical.
It's a mile driven by the car, not by a human. An easy concept to understand, unless you don't want to understand it.
If you want to play the "ultimate goal" game, the ultimate goal is to do it all profitably at scale -- and Tesla is way ahead on that front, which is why their fleet self-drives almost as much per day as Waymo's fleet has ever driven.
Time will tell, but anyone who can't make a case for both "Tesla wins" and "Waymo wins" scenarios is a fanboy with deeply compromised thought processes.
If miles driven by cars under human supervision counted, Toyota cruise control would clock the highest. Autonomy is a binary: it’s either driverless or it’s not. There’s no need to invent terms such as “self driving mile”.
Tesla may be profitable, but nowhere close to a working solution. So how far ahead are they really?
Tesla's system is so fragile it needs a human ready to take over at any second to prevent a crash. Compared to Waymo Tesla FSD is like a kid using training wheels.
Touched on it to your original question but they require cars to drive through their service regions on a regular basis to keep a very detailed map of the environment. Not just the 2d map of a representation of the entire environment. On the remote part, Waymo will encounter trouble and they have support to take control of the vehicle remotely.
People get hung up on the vision stack and mapping. Obviously, any self-driving car needs a way to perceive the world around it, and Tesla’s approach is very different from Waymo.
The software that actually does the driving, ie turns the wheel and works the pedals, is very similar between the two companies. Tesla took a big step in Waymo’s direction earlier this year when they replaced their optimization based path planner with a neural network.
I would recommend watching this video, where the same YouTuber is using FSD which takes an exit and needs intervention to avoid running directly into a divider.
You seem to be stating this as some kind of gotcha.
1. Since that video, FSD 12.5.6.1 has been released, and v13 (which is what is rumored to have been used on Cybercab) is just around the corner. It is completely disingenuius to point to a video 7 months old (on FSD 12.3!) and insinuate that it is representative of the current experience.
2. In FSD, interventions are in your hands. With Waymo in the same situation, your fate lies solely in the remote operator watching your vehicle and how quickly they are able to react. FSD is obviously not perfect but the rate of interventions plummets with every new major release.
Your understanding of remote support is entirely wrong. At no point can a remote operator "drive" a Waymo. They confirm or change plans when the car gets stuck -- that's it.
If you look at the rate of progress over time you see a monotonically improving system that has no apparent halting of improvement. Likewise you have examples of competitors making progress as well in different areas each converging to a pretty, what appears to be, inevitable conclusion of full autonomy.
No one knows when that happens. But it feels pretty certain it’s happening.
I’ve been using FSD for 5 years now. It’s gone from glorified cruise control to something I generally don’t need to intervene with on city streets in that time. Will it improve that fast over the next five years? I doubt it. But it doesn’t have to because the residual problems are much fewer if harder. At this point, especially given the rate of AI improvement overall, I am confident in that five years those problems will largely if not entirely disappear.
Do I take a nap in the back seat? Of course not. Should it be marketed in its condition? I don’t know. But I do know the joint probability of me making a mistake as the attentive operator and it making a mistake while in control is significantly lower than either alone. The fact it makes mistakes at times is obviously concerning to me as a driver, and the fact I also make mistakes actually doesn’t concern me nearly as much as it should. However - I catch its mistakes, and it doesn’t make mine. Why is it rational to be more upset about the machine making a mistake than a human? It’s not - but humans are taught logic and are never rational.
But it's not a robotaxi. Even its level of sensor and compute redundancy is not ready to be a robotaxi. Nothing shown at the Cybercab event changed that. They go for form over reliability every time.
With HW3, they ate their redundant compute node because they underestimated the compute required for the task.
Now even with the redundant node utilized for non-redundant purposes, that doesn't seem to be enough as they are finally admitting HW3 will never not be "supervised".
And then the many years of lying about its upcoming readiness. There are websites out there where you can find all of Musk's quotes about it being just around the corner, or their current generation of vehicles all becoming money-making robotaxis with a little software update worldwide.
There's no indication at all they'll break out of the 100-120 miles per safety disengagement they currently sit at (community tracker, Tesla themselves doesn't publish reviewable safety data).
You being happy that your car can finally make zero intervention trips is NOT the standard necessary for taking the driver out of the seat.
These comments on that video perfectly capture my reaction:
> This is FUD. Who or what speeds up to 55 mph to enter a rotary? How is it that the posted limit and presumably map data indicates 55mp seconds before the rotary? What do you expect FSD which is training on humans using vision and the maps to do? I saw a dinky little rotary sign AT the rotary. I'd slam on the brakes or have an accident too.
> Why would the car come to a stop? I don't see a stop sign, and most roundabouts are yield and I don't see another car blocking your way. Why enter a roundabout at 55? You are wrong, not the car or FSD. You don't know the correct way to drive a roundabout.
FSD mimics human behavior. If you are speeding into a roundabout at 55 mph, you are the one in the wrong, not FSD. It's honestly kind of incredible the ridiculous lengths people go to try to discredit FSD.
That YouTube commenter you quote pretty clearly did not pay attention to the video.
> FSD mimics human behavior. If you are speeding into a roundabout at 55 mph, you are the one in the wrong, not FSD. It's honestly kind of incredible the ridiculous lengths people go to try to discredit FSD.
That's just rephrasing the YouTube comment. Try watching the video yourself. Particularly watch test #7.
Here's a summary:
• The car is on a highway, traveling at normal highway speed of 55 mph. There is no visual indication that there is a roundabout somewhere up ahead.
• After traveling ~3800 feet there is a sign that indicates a roundabout and says the roundabout speed is 15 mph. The roundabout is not yet visible.
• The car continues at highway speed past another sign ~600 feet past the first that also shows that there will be an upcoming roundabout. The road starts curving after that sign, and the roundabout starts coming into view ~600 feet further down the road.
• The car continues approaching the roundabout at highway speed until the human intervenes. He tried to give the car as much time as possible for FSD to decide to slow down. In some of the tests he waited long enough that when he did hit the brakes he had to brake very very aggressively to slow down to 15 mph before entering the roundabout.
Even if it does not have that roundabout in its map and did not read the signs so it is not expecting a roundabout there shouldn't is see it as a sharp bend in the road that should not be taken at highway speed and slow down?
> > This is FUD. Who or what speeds up to 55 mph to enter a rotary? How is it that the posted limit and presumably map data indicates 55mp seconds before the rotary? What do you expect FSD which is training on humans using vision and the maps to do? I saw a dinky little rotary sign AT the rotary. I'd slam on the brakes or have an accident too.
This is a laughable hot take. "seconds before".
Watching the video, it starts with him at 43mph, and he drives at 55mph for FORTY SECONDS before encountering the roundabout.
All these clowns saying "Oh, in the real world he'd have slowed down for that roundabout".
No. He wouldn't have started slowing down two-thirds of a mile away (40 seconds at 55mph). This is a garbage argument.
You're not going to be taking that roundabout at 50 mph. That why, significantly before the roundabout, there's a 15 mph roundabout warning that FSD completely ignores.
Almost crashing the car IS a gotcha for any vehicle purporting to be autonomous. Especially when Musk seems to be betting Tesla on it working vastly better than it currently is.
One is doing it completely driverless and has to get it right every single time. The other has a driver ready to intervene and just needs a single intervention-free drive.
The biggest difference is the presence of lidar in Waymo vehicles which means they have an accurate physical representation of the world vs the vision only based approach that Tesla is taking.
So you’re going to ignore all the data Tesla is collecting, and it’s purely AI approach?
Very biased comments here, I get it, everyone hates Elon, but let’s not lie to ourselves.
For all their faults, one thing you can't say about Google is they don't know how to scale. Prior to Google Maps, Street View would have seemed impossibly expensive, but now we take it for granted. If they need to do a LIDAR run with a car over every public road in the US for them to deploy it, it seems like they could just do that. All they'd have to do is add LIDAR sensors to the existing Street View imaging fleet and continue running them.
If on the chance Lidar is needed, which I know it's not because my car drives me in every type of situation I've thrown at it, then camera only data is still valuable because it gathers so many crazy edge cases.
Waymo's latest hardware uses 13 cameras, 4 LIDARs, 6 radars and at least one mic. Def not early-optimized. I have not found a BOM for the whole package. That's a significantly different idea of how much, and how varied, sensor input it takes.
to my knowledge, Tesla has gone the computer vision route where they are solely relying on cameras and algorithms, while Waymo went the way of more traditional LIDAR and other scanners to close achieve the safe full self drive.
The disadvantage of using the LIDAR and full sensor stack is largely price.
So does Tesla’s. I use it daily. From home going through a busy city, onto a major highway with rush hour traffic, into a downtown area to work. It can do this without me touching the wheel or pedal for the entire length of the drive. I have a hw4 S plaid and it’s made dramatic improvements over this last year. I’m blown away at how good it is (also blown away by waymo).
Well, we know exactly how Waymo's remote operators help out: https://waymo.com/blog/2024/05/fleet-response/. They can't prevent accidents in real time like the Tesla drivers do and can't "control" or "drive" the vehicles.
Tesla FSD is impressive for a driver assist system. But that's all it is — a driver assist. They need orders of magnitude improvement to match Waymo's performance and go driverless.
It still needs to be supervised for the edge cases, but the standard city roads and highways are a solved problem. I think some of the complex roads where you have to quickly cross two way traffic that doesn’t stop can be difficult, I don’t use fsd in that situation, it’s even hard for a human. Sometimes I’ll give it a nudge when it’s being too safe. There’s a construction area that I hit which would have caused the car to take a non optimal path, so I take over there on a regular basis, those issues do get fixed though. That’s about the only issues I have. It can now do things like drive down my long private unmapped driveway without issues.
My work is about 10 miles way in the Seattle area. I can go to and from with zero interventions until I get to my works parking garage
You kind of have it correct, but Tesla is using vision, AI, and huge amounts of data. It’s like the chat-gpt of autonomous driving.
The data is the most important part, to solve real world driving everywhere, you need huge amounts of data for all the edge cases. Tesla has millions of cars on the road gathering this data, vs a couple of thousand for Waymo
Data quantity is useless if the data is of low quality. You need to be able to judge the car's performance in simulations to guide training. Elon admitted in the latest quarterly this is a huge problem for Tesla -- they have to do many millions of miles of simulations to compare two models. Higher fidelity data would cut this number by many orders of magnitude.
You either didn't understand what Elon said, or are deliberately misinterpreting what he said - I listened to the earnings call myself. He said it's taking longer to train the models because the miles between interventions is getting so large that it takes a while to see which model is better when they're comparing different models. It's not a "huge problem", it's a good problem.
> Tesla has millions of cars on the road gathering this data, vs a couple of thousand for Waymo
Except that Alphabet has been mapping and scanning for years, since before Waymo. And, Waymo vehicles are on the road while waiting for a fare, so they can use that time for mapping, while Teslas are reliant on where their owners go.
Tesla thinks they can do everything with nothing but cameras. Everyone else is using multiple sensors to give the computer additional info to verify what the vision is sensing. At this point it seems extremely unlikely Tesla will ever produce fsd with vision only. Probably why they added radar back into the model s refresh. It is currently not enabled afaik.
I find it fascinating the crew of Tesla fanboys on this site that run around downvoting anyone who isn’t drinking the koolade, but never have anything factual or constructive to add to the conversation.
There is literally no indication at all that Tesla is right and every other expert in the industry is wrong about vision only never working. We’re a decade in and they still can’t prevent phantom braking that a 2010 ford Taurus didn’t suffer from.
There isn't a big story to this. Does Waymo share details about how and when its vehicles are driven remotely? Since the answer is no, you can only assume that it is pretty often.
Really kudos to Zoox for turning a profile about them into this unveiling of Waymo. I fell for Waymo's sleight of hand. Aspirationally, we really do want self driving cars.
Occasionally there are problems of being stuck (at 0 mph) due to a lack of aggressiveness or the refusal to violate a rule, or being in a loop the car can't figure a way out of. Often times this is just a safety precaution, at this early stage, of wanting a human in the loop to confirm a plan, but sometimes it's because the system isn't robust enough to properly make the plan and I'm sure that's where a lot of the development effort is currently pointed -- reducing these events to as close to zero as possible. That's the long tail. It never really quite goes away.
They aren't. The car asks the remote operator about its options, the operator suggests an option, the car proceeds. Nobody is driving in realtime over LTE, that would be insane.
> Nobody is driving in realtime over LTE, that would be insane.
To be pedantic that's not really a good argument against the possibility of remote driving. All of Waymo's service areas are inside the coverage areas of AT&T's 5G+, Verizon's 5G Ultra Wide, and T-Mobile's 5G Ultra Capacity networks.
Just because they figured out an efficient way for the cars to be driven remotely doesn't mean that the cars are not being driven remotely. This ventures a bit into a subjective sense of what is self driving, and clearly, Waymos overstate their self driving ability through a variety of very clever, very powerful sleights of hand.
Waymo shared that their disengagement rate is 17k miles per disengagement meaning they are driven by human test drivers once per 17k miles, across all causes. It doesn't include assists by humans because that's not driving: it's merely remote humans answering a question about where to drive in case of a novel obstacle on the road. You'd be nuts to think remote humans actually drive: the internet infrastructure is not designed for guaranteed low latency.
I also wouldn't trust NYT reporting on anything related to Big Tech: they have a well-documented pattern of anti-big-tech biased reporting.
In this comment, you are saying the verbatim words "where to drive" to mean the navigation that Google Maps does, but then you used the verbatim words "where to drive" to mean resolving questions about obstacles. So I guess, yes, resolving questions about obstacles is indeed remote driving, and navigation questions from Google Maps is not remote driving. But everyone already knows that.
The sort of obvious definition of self driving means no human intervention whatsoever, which Waymos also fail. So I don't know. Why doesn't Waymo use Google Maps instead of people to tell the car, "where to drive"?
The disengagement rate is calculated when a human test driver is in the vehicle and takes over driving. It's a specific term from autonomous vehicle testing. If a self-driving vehicle has no human drivers in it and it does not know how to drive, it simply pulls over if possible and stops. That's not among the definition of disengagement.
This has happened multiple times in San Francisco. You can find videos of driverless Waymo cars getting stuck. You can also find a hilarious instance where a stuck Waymo was driven by the emergency personnel after it became stuck.
What a terrible business, past, present, future, any way you want to stack it. Once whatever executive sponsors of this science experiment age out, you can expect this company to no longer exist. At this point, it's basically a redirection of wealth from Google shareholders to small group of people who are running a very expensive, medium size taxi cab company that needs a constant stream of outside capital to keep running.
The fact that they decided to use expensive Jaguars for their fleet just shows they are openly laughing in the faces of Google shareholders. They know how preposterous the entire venture is. At least they have had the foresight not to kill anyone (yet).
Big Tech is largely set of interlocking duopolies.
* Ads: Google/Meta
* Cloud: Amazon/MS
* Mobile: Google/Apple
* Desktop: Apple/MSFT
* VR: Apple/Meta
There's also smaller players and sometimes strong third players (eg GCP), but by and large there are no single winners. Waymo and Tesla are looking like a good bet on the self-driving duopoly. I think a lot of people are skeptical on Tesla because "LIDAR", but lidar is largely irrelevant. It is only used in 3D scene reconstruction, it has nothing to do with actual driving which is the hard part.
> lidar is largely irrelevant. It is only used in 3D scene reconstruction
Source? AIUI lidar is actively used as a sensor/input to detect absolute distance between things, which seems like 80% of what you need to drive safely.
Lidar produces a monochromatic 3D scene, which you can also extract with cameras using SLAM techniques. But these scenes simply do not carry the information needed to navigate the world - you only get one colour, so you cannot see signs, markings, lights, anything. LIDAR is thus a small, secondary input to the problem of "perception".
Once you've accurately perceived the world, the hard part is the action, and the judgement needed to drive safely. It doesn't matter how you got your labeled 3D scene (LIDAR or SLAM), you have to figure out how to navigate it safely and this is more like 90% of the self-driving challenge. If you look at videos of the latest FSD in action (12.5) perception is not an issue. Teslas see everything just fine, they don't disengage because they didn't see a car, but because their judgement is not human level yet.
That is one way, yes, there are others. But if you find my writing unconvincing, here's something you can try: Watch videos of FSD driving and note how often it fails to see objects like cars, trucks, people etc. FSD has 99 problems, but depth perception ain't one.
Can you stop talking nonsense. There is so much difference between cameras and lidar. Weather and fog being one clear example. It’s unbelievable what people will do to suck up to elon.
Completely agree. It’s frustrating how every thread on self-driving devolves into the same argument about cameras vs. LIDAR. Everyone points to Tesla FSD disengagements as evidence, when the vast majority of disengagements are caused by the route planning, not the vision stack.
I’ve watched countless hours of FSD footage and have FSD on my Model 3, and I’ve never seen it incorrectly map the position of a vehicle. Both Tesla and Waymo seem to have reliable perception stacks, the actual decision-making is the hard part, like you said.
I'm really glad that Google/Waymo isn't the only player in this space. Everything Google invests in is ultimately to benefit their ad business. If Tesla wasn't also competing here, the future would be robot taxis that relentlessly bombard you with ads from start to finish and we would have no choice but to tolerate it.
This will cover another 4 years of R&D. They have a very long way to go before the economics of their business make sense. They are pretty far from making money on a per ride basis. They'll have to show significant per ride profitability before they can scale.
They could give the rides away for free, and it would not change anything. Somebody willing to risk $1b on Waymo doesn't credibly care about the kopecks they are earning on rides today. If they did, they would just sit on the sidelines, and wait until it is a public company, and buy post lockup expiry crash priced shares then.
I wonder what Waymo sees as the limiting factor to expanding to the top 24 US metropolitan areas in the next 12 months: affordable vehicle availability (e.g. tariffs on Zeekr), all-weather performance in northern climates, regulatory limitations?
I've heard that just acquiring and setting up the real estate and setting up depots takes a while. After all there aren't a bunch of robotaxi hubs sitting around in cities waiting to be bought or leased. That's where Tesla could have an advantage (if they could get robotaxis working), since in theory they could bootstrap using their supercharger network.
I haven't read much about Waymo depots, but from what I can tell they are : empty parking lot + chargers. If so, that seems like there would be ample options for this in almost any city. Amazon built a charging network in short order (https://news.ycombinator.com/item?id=40083622). Perhaps, their peak (kWh) charging requirements are higher than I would guess.
Nope, Waymo depots are secured garages in industrial zones.
I've been to two in the Phoenix area (Lost & Found FTW).
They are secure facilities near freight train tracks, and their neighbors are logistics companies. I saw Waymo vehicles coming and going, but I did not spot their point of entry into the building. There was no "big parking lot" full of Waymos or chargers visible from the outside.
The visitor/guest entrance is just a security desk and a waiting room with about 3 chairs. Even during business hours, the front door may be locked. You announce your purpose and the security guard admits you. I was once told that I'd need to wait outside, no problem: the weather was fine, and picnic tables were nearby. Once an employee retrieved my Lost & Found items, I was sent on my way.
What's the exit here? They've raised over $10 billion at VC multiples, valuation around $50 billion. Do they expect this to be a $1 trillion company, even then the return would be a modest 20x.
I feel when that much money is at stake and such a high exit is expected, this could distort incentives. I feel like investors would prefer to set their money on fire for some tiny probability of that 20x+ return rather than look to actually make a modest profitable business
Do they expect this to be a $1 trillion company, even then the return would be a modest 20x.
The ideal end game for self-driving taxis is replacing personally-owned cars. Instead of anyone having a car of their own they'd just order one for any journey they make. If ordering a taxi is easier and cheaper than owning a car, and potentially faster and safer because it's self-driving, it's a no brainer for anyone who sees a car as a cost rather than a recreational vehicle.
If they manage to do that then Waymo will be significantly more valuable than a trillion dollars.
As long as US keeps the Chinese manufacturers out..
Baidu Apollo now has a $28K self driving vehicle [0], I doubt Waymo is ever getting close to those numbers. They might replace some US personal cars, but they won't be able to compete globally.
edit: Waymo wanted to use Zeekr vehicles before, I guess that ship has sailed now - they will have to get by with their new Hyundai contract [1]
> I doubt Waymo is ever getting close to those numbers
Why? In theory, Waymo should be able to copy everything Baidu is doing, e.g. manufacture in a low-labor-cost country like India.
This is mainly about the technology, that's the hard problem. Optimizing the costs is something that has been done many times before and we know how to do it.
At some point the technology is good enough, then the cheapest car wins. Waymo doesn't have vertical integration necessary to rapidly iterate and it can't afford to take risks - it's still not operating their commercial service on the highways, not because the tech isn't there, but because of the liability and PR damages if there's a fatal crash (much more likely at highway speeds).
Baidu can have their test car kill a pedestrian every day for the next 10 years if necessary, that's not exactly an even playing field.
Waymo is providing the self-driving AI, it's currently tied to a specific hardware, but it will become adaptable to a wide range of hardware. It will be used in both cheap and expensive cars. If the technology is good enough, it will work with cheap cars too.
They will start using highways with regular passangers probably by the end of this year. Maybe Baidu can launch on highways with a 10x higher accident rate, but I don't think it's a significant advantage to the business.
When operations are scaled up and routine, robotaxis will have a short life, probably 3 years or less. Even at $100k/car every 3 years, that's probably 10% of LTV for the cars.
Uber knows this. They just did not have the money or tech Alphabet could bring to bear.
Yes, wear: Robotaxis will probably go farther in a day than full-time rideshare drivers, maybe by 2X or more. Plus sensors and on-board compute will advance quickly. Waymo is on their 6th generation. 10 sets of tires in that time would not be surprising. That many miles also bumps up against battery lifespan.
But aren’t most cars on the road at the same high traffic times?
I need a ride a 9am and 5pm just like half the city. People keep talking about fleets being in demand and your Tesla becoming a taxi while you sit on your office but wouldn’t everyone else’s also?
Didn't Uber raise over $30 billion from VCs before going public? It seems likely that they could continue for many more years at the current burn rate or even higher.
Google controls/dominates search, much of the mobile phone market, OS on many devices, some of our programming languages, build tools, infrastructure, maps, ads, video, email, and now transportation? Anything I’m missing?
You are being disingenuous, "summon" could be considered as fully driverless, so there are a couple of miles. They are getting there for sure, just not at the similar level, which is expected because customers do not care about it that much.
This is silly. The conversation is about L4 autonomy. Summon isn't fully driverless by any stretch of imagination. It requires the driver to be within a few feet and hold down a button in the app (essentially a kill switch). Moreover, the liability is still with the driver.
Disingenuous would be trying to counter an argument that Tesla has no driverless operations on public roads with a suggestion that a parking lot party trick counts.
average miles between intervention in similar driving scenarios would be a better metric. I’m struggling to understand how anyone thinks the robotaxi thing is competition though
> In reality Tesla FSD is still in very primitive shape.
From the article. Have the people saying this actually tried the most recent FSD build?
I used it yesterday, and it reacted flawlessly to road construction that blocked a lane, then had cones demarcating a lane that didn't follow existing road markings. Then it saw that a driver wanted to turn left from parking lot while approaching a stop light and made space for the driver to turn out. It also flawlessly executed a merge involving turning into a suicide lane and waiting for traffic to pass.
It's very good at adapting to changes and can work on roads its never seen before. There is no other production car that can do what FSD can do currently.
Does this mean it's ready for full autonomy tomorrow? No, but I could use supervised FSD today if I wanted to drive 1600 miles from Austin to Toronto with hands mostly off the wheel.
The pace at which FSD is now improving is quite remarkable. In the long run it's clear that this approach is going to be WAY more flexible than Waymo's, because Tesla has so many more vehicles on the road collecting so much more data, and throwing vast amounts of compute at vast amounts of data is basically the best way to see rapid improvement for the foreseeable future.
It's crazy that when I walk to the taco stand down the street in my sleepy little neighborhood I'll see waymos just hanging out waiting for a ride, and if I wanted I could just hop in one and be robot chauffeured around the city.
It's funny how the future arrives and intermingles with the present. Stuff I could only dream of as a kid, things that seemed impossibly far off are here right now.
I'll probably have a Rosie in my house before I die and that's pretty incredible.
If you mean "driven or steered remotely": They were never teleoperated. I'm assuming that would be completely reckless given the nature of radio networks and possibly because of regulations. The car will call a service once it can't make a decision (this is only when the car is fully stopped), and a human will decide to send out a driver or make a decision in the moment.
These guys are all talk and no action. They wouldn’t dare risk their life in the backseat while fsd is turned on but happy to be a keyboard warrior for Elon. It’s hilarious.
Yup, this. I use FSD almost every day now that it's available for Cybertruck, and it's already quite excellent. It pulls off human-like maneuvers that are kind of gobsmacking.
Then jump in the backseat if it’s gobsmacking and just as good as a waymo. Tell it to drive you to the other side of the city while you sleep in the backseat. Put your money where your mouth is.
I can understand the average person not realizing what Teslas can do today, but I'm highly disappointed that so-called "tech" people on this message forum have no idea.
See the thing is they do and you don’t. If you don’t know and understand why waymo is far ahead of elons lies and will never catch up with its current tech direction and strategy for supervised fsd (which is an oxymoron) then you should only be disappointed in yourself.
If you’re so confident of the tech go jump in the back of a Tesla and let it drive you 100kms. Let’s see if you live to tell the tale. I’m happy to also do the same but in a waymo.
Robotaxi will use the same stack as other Tesla vehicles, which is deployed on millions of vehicles. The <$30k price point and lack of controls/supervision are the only things that truly differentiate it from a Model 3.
Waymo is patronized by Alphabet for as long as it needs to be. This has been good, as it takes off a lot of pressure to virtue signal fake progress to impatient investors; a fate many of Waymo's potential competitors haven fallen victim to.
When Uber and Tesla were acting like cowboys and creating plenty of drama for the media Waymo (under the sensible and boring leadership of non-techy car industry vet John Krafcik) was quietly structuring itself to be in the game for the long haul, following the whole "Fuck You Money" episode that saw most of the Google Self Driving Car Project's original talent leave for easy Venture Capital elsewhere.
Waymo is very tight lipped about their operational costs, I would really love to know how close they are to being revenue positive. Recent moves by Waymo suggest they believe they are getting close.
They're doing 100k paid trips/wk now, per the CEO in a recent LinkedIn post. If average trip value is (very generously) $20, that's still only $100MM in annual revenue. I would bet that doesn't even pay for the direct physical + digital costs of operating the cars, let alone salaries, office leases, etc.
Waymo has started onto a very good trendline now, for sure, and a path to revenue-positive is becoming clearer. But they have a loooooooong way to go before they are offsetting even their operating costs, let alone beginning to pay down their (absolutely massive) fixed costs.
I think the bear case is actually pretty tough for Waymo. Their scaling is not cheap.
- The cars are, by all account, expensive as hell; I've seen $200k+
- The cars require ongoing cleaning, maintenance, mechanical work
- There's a massive customer support operation; not just people to call for support when your car gets stuck, but people who can step in and remote-instruct the vehicle.
- Expanding to a new city requires a massive data-gathering operation before any paid trips can begin.
None of this is taking into account the massive fixed costs that come with being a tech company - hundreds of $400k+TC employees, server bills in the tens of millions, thousands of top-of-the-line GPUs for training clusters.
> that's still only $100MM in annual revenue.
Given how fast Waymo is scaling, we should look at forward revenue projections, if available. Also, when it comes to TAM, Uber is a good standard to compare against. In Q2 2024[1], Uber had
- 10.7B revenue (42.8B ARR), out of which 6.134B (24.5B ARR) was from "mobility" and others from delivery/freight.
- Real fun part: Gross booking revenue for mobility was 20.554B, but revenue for Uber itself was 6.134B, or 30% of gross revenue. If 70% is going to drivers, that is Waymo's opportunity. It can price at 40% discount to Uber's prices while still reaping twice the revenue per ride. Start with the most lucrative markets where Uber is already profitable and expand from there.
[1] https://s23.q4cdn.com/407969754/files/doc_earnings/2024/q2/e...
Waymo has to recoup their research costs which are far from trivial, so I doubt prices will drop. Futhermore, there's an agreement for Uber to run the Waymo's in their app, futher complicating the calculus
https://investor.uber.com/news-events/news/press-release-det...
The drivers cut covers the capital cost of the car, maintenance and fuel. Not just labor.
Consumer cars really aren't meant to last over 100k miles. I know a lot of uber drivers will put 30k miles or more on per year. Also if a car becomes a standard widget made to be easily repairable with parts that are standard widgets, repairs and maintenance would be cheaper.
E.g. If you needed to clean the carpet in a vehicle, make the carpet easily removable and washable in an industrial washing machine.
> Consumer cars really aren't meant to last over 100k miles
What on earth are you talking about? A Honda Civic or Toyota Corolla will easily last 200k miles. In places with salt and rust, the body will probably fall apart before the engine and transmission give out.
"Many non-commercial vehicles (both auto and truck) have exceeded 1,000,000 miles (1,600,000 km). For instance, in 2013, East Patchogue, New York resident Irv Gordon (1940-2018) had accumulated 3,000,000 miles (4,800,000 km) in his 1966 Volvo P1800. The car had amassed 3,200,000 miles (5,100,000 km) by Gordon's death on 15 November 2018." https://en.wikipedia.org/wiki/Car_longevity https://autos.yahoo.com/news/the-first-car-to-3-million-mile...
Honestly, I'd not use the word many, perhaps a significant number....
Parent said 100k miles. I agree that 1m miles is much rarer.
Then why do those cars not come with a standard 100k warranty?
To quote Charles Babbage "I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question".
Does everything you buy keel over and die the minute its warranty is over? Do you throw away your car the first time it needs a repair?
With 700 cars, 100MM is $140K per car per year so that should be covering direct car non-employee expenses.
[dead]
There is absolutely no way they are remotely close to being revenue positive (unless you're asking strictly if they are charging more than the gas they're burning without counting anything else). They have like 700 very expensive taxis and a burn rate measured in billions of dollars per year.
> (unless you're asking strictly if they are charging more than the gas they're burning without counting anything else).
Yes, I assume they're asking about the operating margin. That's the number everybody cares about. If it's positive, then Waymo can theoretically become revenue positive by adding more cars and regions to their fleet.
That number needs to include car depreciation + any remote 'safety drivers' + the maintenance depot overhead + fuel and insurance + any other costs that scale with the number of cars then.
Exactly. They have revenue in the low 6 figures per vehicle annually, so they probably need to get the number of operational employees per vehicle down to around 1/10. Seems feasible, but they're probably not anywhere close to that yet.
On the other hand, HN has consistently underestimated them for the last two years or so as they've begun to commercialize. Their trajectory has been 10x growth in rides served per year for a while now. At the current scale it's hard to imagine they could grow another 10x in the next year without being close to break even on operating cost.
10x growth would be ~$1B revenue, 7000 cars. Let's assume it requires doubling head count from 2500 to 5000. That's $130K/car, $250K/employee. sounds break-even-ish.
Bird scooters operating margins are very good if you only measure specific locations on specific times of the week.
It is risky to ignore total profit and loss of the entire company
Yep, I look forward to reading good books about it in a decade or so. I think there are lots of lessons to be learned here for how to do a certain kind of difficult new thing.
I'd like to read that book too... but "how to do a certain kind of difficult new thing" in their case involved more than solid execution. It also needed an owner/investor with extremely deep pockets and technical capabilities that vastly accelerated their work, who was willing to throw huge amounts of money at the problem for years, along with favorable economic conditions (zero-percent interest rates) for long periods of time, at the right period of time. I give Waymo tons of credit for their work here, but the conditions that enabled it aren't easily repeatable IMO.
The reason it's interesting (IMO) is that other companies that also had owner/investors with extremely deep pockets and technical capabilities have been throwing huge amounts of money at it in the same favorable interest rate environment, but with seemingly a lot less success.
I think their (apparent) success probably does have a lot to do with some combination of having the deepest pockets and (I think?) being a few years earlier to get started than everyone else, but still, it also looks like they had a better strategy, and I'm pretty curious how that worked.
It is fascinating how Google simultaneously provides a great lesson in how to leverage their primary competency of AI, as well as how not to (losing to OpenAI in LLMs).
> Waymo is very tight lipped about their operational costs, I would really love to know how close they are to being revenue positive. Recent moves by Waymo suggest they believe they are getting close.
What specific "moves" by them indicate they are close to being profitable? All the industry research I get indicates they are still negative to the tune of $1B+ a year.
They're probably at a point where scaling is at least slightly profitable. Otherwise they wouldn't be scaling.
Scaling out now helps along their regulatory moat, being the first trusted operator in a given market means they get a seat at the table for setting the rules
Considering just how many of their cars I see driving around Atlanta covered in sensor arrays every day, I would certainly hope they’re making real progress.
> non-techy
He studied mechanical engineering and worked in that field for a while. Do we agree that there is other technology than software?
Techy does not mean technology
I also understood it to mean the "move fast, break things" attitude, that many techy VC startups have.
So did my father, but I would not consider him "techy".
Anyone knows whether there are fundamental differences in the approach waymo took vs tesla, or are they following the same algorithms and techniques, with more polishing or experience in the "small details that makes all the difference" ?
This an explanation from a layman on the auto side.
Tesla has focused on attempting to solve all cases using vision and generalize it as much as possible. The pro is you would be able to drop FSD capabilities anywhere in the US and it will work. The con is the tail of edge cases take significantly more effort then the first 80%.
Waymo instead of being generalized, works off of a detailed map for each region they drive in. They have a complete expectation of every detail on those roads and so then only need to account for the dynamic unknowns on the road. The pro is that they are getting to hands off driving sooner. The con is that anywhere its deployed needs detailed mapping and for that mapping to be kept up to date.
Time will tell which strategy pays off.
> Waymo instead of being generalized, works off of a detailed map for each region they drive in. They have a complete expectation of every detail on those roads and so then only need to account for the dynamic unknowns on the road.
This isn't entirely true. Everything at Waymo is built to be generalizable. Engineers and execs at Waymo have said to multiple times, as recently as yesterday [1].
They are able to drive just fine without up-to-date maps as well. It's an assumption built in to the system. The cars are also self-mapping. Ultimately, maps are just another input to the general driving software. The same driving software is deployed across all their cities.
[1] https://www.youtube.com/watch?v=d6RndtrwJKE
A map is really just a data cache, I've been surprised so many technical people see it as a flaw. It seems like an important feature to me!
A cache is exactly how I explain maps to people. They are saving precious compute cycles not computing the same static objects over and over again, by N number of vehicles.
A con does not have to be a flaw. But I would see it as a tradeoff. With all of these companies its hard to sometimes see beneath the curtain of what they promote.
Are they deployed into any cities where they are not actively mapping?
They test in many cities (Buffalo, Tahoe, Bellevue, NYC) . But they won't deploy to any cities which they haven't mapped because they consider it critical to safety.
So the answer is no and that's why I outline it as a potential con. They are unable to just drop deployment into a new city, they as a matter of history, map out the entire city and keep an active map of the driving territory. I have no idea who wins in this dog and pony show but I think its a valid potential con and a easy way to see some of the decision differences in companies.
Does Google Earth/Maps/StreetView not count as mapping? I was feeling certain that all that infrastructure and data was invested into Waymo. It would seem that Google had a huge advantage from running mapping/camera cars for years and years, before SDC were even a glimmer in their eyes.
The mapping that waymo uses is much more detailed than streetview, the biggest difference being that it includes lidar data. I wouldn't be surprised if they combined the two for some of their latest data updates, but the pre-existing streetview data is not obviously enough to give an advantage.
> But they won't deploy to any cities which they haven't mapped
And therein lies the fundamental flaw with Waymo's approach...
It's not a flaw if that's what is required to make it work. So far only Waymo works at L4 autonomy, so they are vindicated.
As compared to what, Tesla's FSD that tries to crash the car every 20 miles or so?
Obtaining a detailed map doesn't seem difficult. This is the same company that created Street View. E.g. SF has ~1000 miles of streets [1], that would only take 50 hours of paying for manual driving to map all of SF at 20mph.
1. https://data.sfgov.org/City-Infrastructure/Miles-Of-Streets/...
Never said it was difficult but that their driving fleet relies on a lidar map of the region it serves. Its a definite trade-off so I would list it as a con but it does not mean one is better than the other.
Elon said a couple of days ago Tesla cars might never be able to self drive.
https://futurism.com/elon-musk-realizes-all-teslas-self-driv...
And probably because of this strategy, by the time their system is successfully deployed in several cities, the technology to make fully self driving has become much more accessible and they can relatively easily add that incrementally.
I don't see it as a significant dis-advantage for Waymo.
With modern drones and UAV, getting millimeter level details on infrastructure should be easy and cheap.
I would expect such maps for every corner of the world in next 5 to 10 years.
The other pro is cost: cameras are much, much cheaper than LiDAR. Tesla is making the bet that they can close the performance gap between cameras and LiDAR faster than the cost of LiDAR will come down.
I wonder if this will ever be a significant factor. How much can a LiDAR setup cost? 4k? Maybe 2k if built in-house at car manufacturer scale in the long term? Is that significant when buying a whole car?
if i am interested in reading more about this, can you point me to some resources?
I’d advise you to ignore lay explanations of the space - outside of the industry most of the discourse about self driving cars is poisoned by Elon’s deceptive presentations and his followers who parrot what he says.
If you want a grounded explanation of how Tesla’s stack works, follow @greentheonly on twitter. He’s a Tesla reverse engineer who regularly posts about the software that’s actually running on the car.
If you want an explanation about how real AV companies stacks work, I’d read Sebastian Thrun’s robotics textbook - then imagine what’s outlined in that book but with ML plugged in to a ton of spaces throughout the stack. This is also similar to how Tesla’s stack works, btw - greens just good to follow because a lot of people refuse to believe Tesla isn’t running some kind of “LLM but for driving” fully end to end black box model.
Tesla won’t launch a robotaxi anytime soon because they can’t use remote support or HD maps - although I think they’ve been stepping up their mapping efforts. Even the demo at Universal studios a few weeks ago was HD mapped - per @greentheonlys twitter.
I worked in the space for years and have seen the internal of both a traditional robotaxi company’s stack and Tesla’s.
For reference, Sebastian Thrun led the Stanford team that won the Darpa (self driving car) Grand Challenge in 2005, and then joined Google to lead Waymo (then called the Google Self-Driving Car Project), among other accomplishments.
NOVA episode: https://www.youtube.com/watch?v=vCRrXQRvC_I
Unfortunately the industry like many others are poisoned on many ends. I would caution taking anyone's opinion too strongly including this one.
Google "waymo dolgov" on YouTube, there's a recent lecture and a recent interview by Dmitri Dolgov. In the lecture, he talks about their architecture.
Tesla chose to recklessly endanger lives, killing at least 50 people by releasing an unsafe beta test. Their goal was to capture market share from Waymo.
This is the thing that drives me nuts: wantonly ignoring regulations designed to protect consumers, and claiming that it was going to save more people than it hurt, in the long run. Tesla's participation in the market has not and will not make self driving arrive earlier than if they had refrained from participating. All it has done is given them an opportunity to compete for market share, and engage in pump and dump scams by misrepresenting their progress to naive investors.
People will tell you they're fundamentally different, but they are in fact the same. First, there are two different, independent aspects: Perception and Action. Seeing the world, vs taking action in that world.
For perception Waymo uses more sensors than Tesla. It uses lidar to construct a 3D scene of what the car is seeing, while Tesla uses SLAM-like techniques with their cameras. What people are missing is that these are only a small part of the perception problem - lidar returns a monochromatic 3D scene so it cannot see labels, markings, read signs, lights etc. LIDAR simply doesn't carry the necessary information needed to navigate the world, thus it is a secondary source of info. Cameras in motion do carry all the information needed, so the big difference between them is only one part of the overall perception stack.
Once you have constructed a labeled, accurate 3D scene (whether by lidar or SLAM), "action" is the same and there is no difference between Tesla and Waymo here. They both have to learn how to drive safely using the same information, so it's going to be a lot like LLMs where there's difference between LLAMA/Claude/GPT, but they're also all kinda the same thing.
The fact that you don’t know the difference in capabilities between cameras and LiDAR means you shouldn’t really be commenting. LiDAR allows you to also ‘see’ things in conditions where weather is bad visibility is limited. Cameras cannot allow you to do this.
Going through this thread, it is mind blowing how people let their fanboyism for others talk nonsense. So many Elon lovers here. Can all those Elon lovers just jump in the backseat of a Tesla and turn on FSD (which is an oxymoron) and jump on a high way or something? The world would be a better place. I can’t wait to see their faces and a the drivel they will come up with next year when Elon announces delays and other bull**t fit why he can’t launch what he promised. It’s going to be hilarious.
Tesla is taking a minimalist approach to every important factor: Specialized vs unmodified production cars; back seat in-vehicle UI + app vs app-only for sitting in back; multiple sensor types vs camera-only; intensive mapping of a service area vs crowdsourced; only Waymo-owned vehicles vs private cars seconded to the service.
The biggest small detail is that Waymo's expansion is gated on achieving performance goals like very little need for remote supervision because Google won't hire a building full of remote monitors. Tesla claims this will be possible "by next year."
Tesla is betting the farm on vision only L4 AD actually being possible and so far have not done much to prove this is the case.
Fundamentally different - people will flock to LiDAR vs Vision when debating this, but the more fundamental difference is that Waymo uses HD mapping + remote support.
Thanks, i was expecting something beyond just the lidar thing. Can you elaborate more about mapping + remote ? I've never tested neither a tesla autopilot nor a waymo (european here).
edit : also, are there differences in the core algorithms ? Tesla seems to be full AI / ML. Is waymo the same ? ( as the company is older i wonder if they haven't built more things manually)
Neither of them are “full AI/ML”, they’re both traditional robotics systems with ML used for detection/prediction/planning at certain steps. Elon will sometimes say something about moving to a “new ML stack”, but Tesla reverse engineers regularly look inside of what’s running in the cars and that’s not the case at all.
Contrary to what other people in this thread are saying, the remote support isn’t remote direct driving of the car - essentially what will happen is that if the car finds itself in a situation where it’s unsure of how to proceed and it’s safe to stop, it will pause for a few seconds and wait for a remote operator to clarify a situation for it.
A good example might be road construction - if the car detects new road construction work that doesn’t match its map of the area, and its onboard systems determine that it’s not sure how to proceed through the construction with confidence, it will send what it thinks the top five likeliest ways to proceed to a remote operator. The operator then selects the proper path (or says that none of them are proper). The car will then follow the path presented by the operator, but actual driving behavior /collision detection / pathfinding is still determined locally. Think of it like ordering a unit around in StarCraft.
You can actually see this behavior in the car when it runs into a difficult situation. It tells you it's asking for assistance or something similar, and pauses for a few seconds.
I also made the mistake of assuming the remote operator drives the car but if you watch Waymo's technical videos, it's clear that the AI is in control of the car all all times and the remote operator is just doing near real time labelling of what the car is seeing.
> Tesla reverse engineers regularly look inside of what’s running in the cars and that’s not the case at all
How are they doing this on the software stack? Any references?
Same way reverse engineer get access to internals of other devices - a bunch of tricks. :^)
In one case, greentheonly realized some fraction of Tesla’s cars are shipped out of the factory still in dev mode, with debug mode enabled and increased privileges. He found someone with a car like this who was down to helped and swapped part of his cars hardware with their car, and from then on was able to get a much better view of what was running on his car.
Unfortunately twitter is awful to search and a lot of his info is buried deep in old threads, but a few (old) examples to illustrate that he regularly does this.
Visualizing the outputs of the models running in the car: https://x.com/greentheonly/status/1404164587927314435
Tesla’s dev debug menus circa 2020: https://x.com/greentheonly/status/1336467014727110656
It’s been a while since I’ve worked in the space so I haven’t followed green as closely.
Wow, thanks!
Waymo has been at the forefront of cutting edge AI/ML since the beginning: https://waymo.com/research/
This is a good read: https://www.understandingai.org/p/elon-musk-wants-to-dominat...
This talk by Waymo co-CEO is even better (especially from 33:50): https://www.youtube.com/watch?v=s_wGhKBjH_U
It's a fallacy to really compare them directly. Waymo has active cars in service that are available to the general public. Tesla right now just has promises that they have been unable to deliver on for almost a decade now, as their approach is pretty radical.
Yes, Elon puffery is silly, but so is throwing the comparison in favor of your preferred team by precision-selecting a metric that does so.
Driverless miles: Waymo 20M, Tesla 0
Self-driving miles: Waymo 20M, Tesla 2000M
Miles driven with assumed legal liability: Waymo 20m, Tesla 0
I don't think Tesla will ever operate a self-driving car with no driver where they are legally liable for crashes.
What's a "self-driving mile"?
Only metric that counts is driverless miles. That's the true test of autonomy.
It's a mile driven by the car, not by a human. An easy concept to understand, unless you don't want to understand it.
If you want to play the "ultimate goal" game, the ultimate goal is to do it all profitably at scale -- and Tesla is way ahead on that front, which is why their fleet self-drives almost as much per day as Waymo's fleet has ever driven.
Time will tell, but anyone who can't make a case for both "Tesla wins" and "Waymo wins" scenarios is a fanboy with deeply compromised thought processes.
If miles driven by cars under human supervision counted, Toyota cruise control would clock the highest. Autonomy is a binary: it’s either driverless or it’s not. There’s no need to invent terms such as “self driving mile”.
Tesla may be profitable, but nowhere close to a working solution. So how far ahead are they really?
Tesla's system is so fragile it needs a human ready to take over at any second to prevent a crash. Compared to Waymo Tesla FSD is like a kid using training wheels.
Touched on it to your original question but they require cars to drive through their service regions on a regular basis to keep a very detailed map of the environment. Not just the 2d map of a representation of the entire environment. On the remote part, Waymo will encounter trouble and they have support to take control of the vehicle remotely.
Is the mapping a valuable byproduct?
Google maps is famous for their human driven cars that drive around with sensors. Is there overlap here?
to finish your comparison:
tesla uses cameras + local support (safety driver)
I imagine they'll try to switch to remote support at some point.
People get hung up on the vision stack and mapping. Obviously, any self-driving car needs a way to perceive the world around it, and Tesla’s approach is very different from Waymo.
The software that actually does the driving, ie turns the wheel and works the pedals, is very similar between the two companies. Tesla took a big step in Waymo’s direction earlier this year when they replaced their optimization based path planner with a neural network.
I would highly recommend watching this side by side comparison video of the experience: https://www.youtube.com/watch?v=Kswp1DwUAAI
I would recommend watching this video, where the same YouTuber is using FSD which takes an exit and needs intervention to avoid running directly into a divider.
https://youtu.be/LbCGAN6Pk_c?t=2026
I will never understand how Tesla can get away with releasing such a broken product on the market and charging for it.
You seem to be stating this as some kind of gotcha.
1. Since that video, FSD 12.5.6.1 has been released, and v13 (which is what is rumored to have been used on Cybercab) is just around the corner. It is completely disingenuius to point to a video 7 months old (on FSD 12.3!) and insinuate that it is representative of the current experience.
2. In FSD, interventions are in your hands. With Waymo in the same situation, your fate lies solely in the remote operator watching your vehicle and how quickly they are able to react. FSD is obviously not perfect but the rate of interventions plummets with every new major release.
I have videos of safety critical interventions in v12.5. Here's one of it attempting to drive straight through a roundabout at 50 miles per hour:
https://www.youtube.com/watch?v=b1XagBTmpgw
Your understanding of remote support is entirely wrong. At no point can a remote operator "drive" a Waymo. They confirm or change plans when the car gets stuck -- that's it.
If you look at the rate of progress over time you see a monotonically improving system that has no apparent halting of improvement. Likewise you have examples of competitors making progress as well in different areas each converging to a pretty, what appears to be, inevitable conclusion of full autonomy.
No one knows when that happens. But it feels pretty certain it’s happening.
I’ve been using FSD for 5 years now. It’s gone from glorified cruise control to something I generally don’t need to intervene with on city streets in that time. Will it improve that fast over the next five years? I doubt it. But it doesn’t have to because the residual problems are much fewer if harder. At this point, especially given the rate of AI improvement overall, I am confident in that five years those problems will largely if not entirely disappear.
Do I take a nap in the back seat? Of course not. Should it be marketed in its condition? I don’t know. But I do know the joint probability of me making a mistake as the attentive operator and it making a mistake while in control is significantly lower than either alone. The fact it makes mistakes at times is obviously concerning to me as a driver, and the fact I also make mistakes actually doesn’t concern me nearly as much as it should. However - I catch its mistakes, and it doesn’t make mine. Why is it rational to be more upset about the machine making a mistake than a human? It’s not - but humans are taught logic and are never rational.
It's impressive software for an ADAS.
But it's not a robotaxi. Even its level of sensor and compute redundancy is not ready to be a robotaxi. Nothing shown at the Cybercab event changed that. They go for form over reliability every time.
With HW3, they ate their redundant compute node because they underestimated the compute required for the task.
Now even with the redundant node utilized for non-redundant purposes, that doesn't seem to be enough as they are finally admitting HW3 will never not be "supervised".
And then the many years of lying about its upcoming readiness. There are websites out there where you can find all of Musk's quotes about it being just around the corner, or their current generation of vehicles all becoming money-making robotaxis with a little software update worldwide.
There's no indication at all they'll break out of the 100-120 miles per safety disengagement they currently sit at (community tracker, Tesla themselves doesn't publish reviewable safety data).
You being happy that your car can finally make zero intervention trips is NOT the standard necessary for taking the driver out of the seat.
I can see no path of how Tesla's current L2 system can become a L4 system as capable as Waymo.
These comments on that video perfectly capture my reaction:
> This is FUD. Who or what speeds up to 55 mph to enter a rotary? How is it that the posted limit and presumably map data indicates 55mp seconds before the rotary? What do you expect FSD which is training on humans using vision and the maps to do? I saw a dinky little rotary sign AT the rotary. I'd slam on the brakes or have an accident too.
> Why would the car come to a stop? I don't see a stop sign, and most roundabouts are yield and I don't see another car blocking your way. Why enter a roundabout at 55? You are wrong, not the car or FSD. You don't know the correct way to drive a roundabout.
FSD mimics human behavior. If you are speeding into a roundabout at 55 mph, you are the one in the wrong, not FSD. It's honestly kind of incredible the ridiculous lengths people go to try to discredit FSD.
That YouTube commenter you quote pretty clearly did not pay attention to the video.
> FSD mimics human behavior. If you are speeding into a roundabout at 55 mph, you are the one in the wrong, not FSD. It's honestly kind of incredible the ridiculous lengths people go to try to discredit FSD.
That's just rephrasing the YouTube comment. Try watching the video yourself. Particularly watch test #7.
Here's a summary:
• The car is on a highway, traveling at normal highway speed of 55 mph. There is no visual indication that there is a roundabout somewhere up ahead.
• After traveling ~3800 feet there is a sign that indicates a roundabout and says the roundabout speed is 15 mph. The roundabout is not yet visible.
• The car continues at highway speed past another sign ~600 feet past the first that also shows that there will be an upcoming roundabout. The road starts curving after that sign, and the roundabout starts coming into view ~600 feet further down the road.
• The car continues approaching the roundabout at highway speed until the human intervenes. He tried to give the car as much time as possible for FSD to decide to slow down. In some of the tests he waited long enough that when he did hit the brakes he had to brake very very aggressively to slow down to 15 mph before entering the roundabout.
Even if it does not have that roundabout in its map and did not read the signs so it is not expecting a roundabout there shouldn't is see it as a sharp bend in the road that should not be taken at highway speed and slow down?
I wish we could shut down the road and put a crash dummy in the drivers seat and see what happens.
> > This is FUD. Who or what speeds up to 55 mph to enter a rotary? How is it that the posted limit and presumably map data indicates 55mp seconds before the rotary? What do you expect FSD which is training on humans using vision and the maps to do? I saw a dinky little rotary sign AT the rotary. I'd slam on the brakes or have an accident too.
This is a laughable hot take. "seconds before".
Watching the video, it starts with him at 43mph, and he drives at 55mph for FORTY SECONDS before encountering the roundabout.
All these clowns saying "Oh, in the real world he'd have slowed down for that roundabout".
No. He wouldn't have started slowing down two-thirds of a mile away (40 seconds at 55mph). This is a garbage argument.
Are you serious? That driver “testing” FSD doesn’t seem to understand basic rules of the road, like that you are not supposed to stop at a roundabout.
You're not going to be taking that roundabout at 50 mph. That why, significantly before the roundabout, there's a 15 mph roundabout warning that FSD completely ignores.
Almost crashing the car IS a gotcha for any vehicle purporting to be autonomous. Especially when Musk seems to be betting Tesla on it working vastly better than it currently is.
It's apples-to-oranges comparison.
One is doing it completely driverless and has to get it right every single time. The other has a driver ready to intervene and just needs a single intervention-free drive.
Waymo uses LiDAR and Tesla uses optical cameras. At least that is my understanding as a Joe Schmoe in the Midwest.
The biggest difference is the presence of lidar in Waymo vehicles which means they have an accurate physical representation of the world vs the vision only based approach that Tesla is taking.
iirc the big thing is they use LIDAR/RADAR in addition to camera sensors, Tesla is hellbent on only using cameras.
Also, Waymo works.
So you’re going to ignore all the data Tesla is collecting, and it’s purely AI approach? Very biased comments here, I get it, everyone hates Elon, but let’s not lie to ourselves.
No one's ignoring it. They're saying that Waymo currently works, and Tesla Self Driving does not currently work.
My Tesla drives me around the city autonomously every day and you're telling me it doesn't
Waymo is restricted to very specific pre-mapped areas. It is not a scalable solution.
For all their faults, one thing you can't say about Google is they don't know how to scale. Prior to Google Maps, Street View would have seemed impossibly expensive, but now we take it for granted. If they need to do a LIDAR run with a car over every public road in the US for them to deploy it, it seems like they could just do that. All they'd have to do is add LIDAR sensors to the existing Street View imaging fleet and continue running them.
If it turns out LIDAR 3d mapping is needed, how much is camera-only data worth?
If on the chance Lidar is needed, which I know it's not because my car drives me in every type of situation I've thrown at it, then camera only data is still valuable because it gathers so many crazy edge cases.
Tesla is aiming for a camera only method, whereas Waymo is going for a camera+LIDAR approach.
Waymo's latest hardware uses 13 cameras, 4 LIDARs, 6 radars and at least one mic. Def not early-optimized. I have not found a BOM for the whole package. That's a significantly different idea of how much, and how varied, sensor input it takes.
and high def maps, which may or may not be up to date
One major difference is that Waymo went all in on lidar while Tesla did not.
to my knowledge, Tesla has gone the computer vision route where they are solely relying on cameras and algorithms, while Waymo went the way of more traditional LIDAR and other scanners to close achieve the safe full self drive.
The disadvantage of using the LIDAR and full sensor stack is largely price.
The advantage is that it works.
So does Tesla’s. I use it daily. From home going through a busy city, onto a major highway with rush hour traffic, into a downtown area to work. It can do this without me touching the wheel or pedal for the entire length of the drive. I have a hw4 S plaid and it’s made dramatic improvements over this last year. I’m blown away at how good it is (also blown away by waymo).
The difference is you are in the driver's seat paying attention at all times and ready to intervene. That's the expectation set by the system.
It's the fundamental difference between partial autonomy and full autonomy.
Waymo also has drivers. They are just remote and they intervene when needed. Waymois is also bound to highly mapped roads in a few cities.
Don’t get me wrong waymo is impressive, I use it as much as possible. The future looks amazing. Do you also agree Tesla’s self driving is impressive?
Well, we know exactly how Waymo's remote operators help out: https://waymo.com/blog/2024/05/fleet-response/. They can't prevent accidents in real time like the Tesla drivers do and can't "control" or "drive" the vehicles.
Tesla FSD is impressive for a driver assist system. But that's all it is — a driver assist. They need orders of magnitude improvement to match Waymo's performance and go driverless.
Is it reliable enough that you don't need to supervise it? What's your estimate of the miles per intervention?
It still needs to be supervised for the edge cases, but the standard city roads and highways are a solved problem. I think some of the complex roads where you have to quickly cross two way traffic that doesn’t stop can be difficult, I don’t use fsd in that situation, it’s even hard for a human. Sometimes I’ll give it a nudge when it’s being too safe. There’s a construction area that I hit which would have caused the car to take a non optimal path, so I take over there on a regular basis, those issues do get fixed though. That’s about the only issues I have. It can now do things like drive down my long private unmapped driveway without issues.
My work is about 10 miles way in the Seattle area. I can go to and from with zero interventions until I get to my works parking garage
You kind of have it correct, but Tesla is using vision, AI, and huge amounts of data. It’s like the chat-gpt of autonomous driving.
The data is the most important part, to solve real world driving everywhere, you need huge amounts of data for all the edge cases. Tesla has millions of cars on the road gathering this data, vs a couple of thousand for Waymo
Data quantity is useless if the data is of low quality. You need to be able to judge the car's performance in simulations to guide training. Elon admitted in the latest quarterly this is a huge problem for Tesla -- they have to do many millions of miles of simulations to compare two models. Higher fidelity data would cut this number by many orders of magnitude.
You either didn't understand what Elon said, or are deliberately misinterpreting what he said - I listened to the earnings call myself. He said it's taking longer to train the models because the miles between interventions is getting so large that it takes a while to see which model is better when they're comparing different models. It's not a "huge problem", it's a good problem.
3rd party testing has Tesla at 13 miles between intervention. Even Elon only promises 10,000 miles between interventions later this year: https://x.com/elonmusk/status/1842029594570006992
> Tesla has millions of cars on the road gathering this data, vs a couple of thousand for Waymo
Except that Alphabet has been mapping and scanning for years, since before Waymo. And, Waymo vehicles are on the road while waiting for a fare, so they can use that time for mapping, while Teslas are reliant on where their owners go.
Fundamentally different.
Tesla thinks they can do everything with nothing but cameras. Everyone else is using multiple sensors to give the computer additional info to verify what the vision is sensing. At this point it seems extremely unlikely Tesla will ever produce fsd with vision only. Probably why they added radar back into the model s refresh. It is currently not enabled afaik.
I find it fascinating the crew of Tesla fanboys on this site that run around downvoting anyone who isn’t drinking the koolade, but never have anything factual or constructive to add to the conversation.
There is literally no indication at all that Tesla is right and every other expert in the industry is wrong about vision only never working. We’re a decade in and they still can’t prevent phantom braking that a 2010 ford Taurus didn’t suffer from.
> fundamental differences
No. A Tesla with a driver and a Waymo have a lot in common: "When self driving cars don't actually drive themselves"
https://www.nytimes.com/2024/09/11/insider/when-self-driving...
There isn't a big story to this. Does Waymo share details about how and when its vehicles are driven remotely? Since the answer is no, you can only assume that it is pretty often.
Really kudos to Zoox for turning a profile about them into this unveiling of Waymo. I fell for Waymo's sleight of hand. Aspirationally, we really do want self driving cars.
Waymos are NEVER "driven remotely".
Occasionally there are problems of being stuck (at 0 mph) due to a lack of aggressiveness or the refusal to violate a rule, or being in a loop the car can't figure a way out of. Often times this is just a safety precaution, at this early stage, of wanting a human in the loop to confirm a plan, but sometimes it's because the system isn't robust enough to properly make the plan and I'm sure that's where a lot of the development effort is currently pointed -- reducing these events to as close to zero as possible. That's the long tail. It never really quite goes away.
> when its vehicles are driven remotely?
They aren't. The car asks the remote operator about its options, the operator suggests an option, the car proceeds. Nobody is driving in realtime over LTE, that would be insane.
> Nobody is driving in realtime over LTE, that would be insane.
To be pedantic that's not really a good argument against the possibility of remote driving. All of Waymo's service areas are inside the coverage areas of AT&T's 5G+, Verizon's 5G Ultra Wide, and T-Mobile's 5G Ultra Capacity networks.
Just because they figured out an efficient way for the cars to be driven remotely doesn't mean that the cars are not being driven remotely. This ventures a bit into a subjective sense of what is self driving, and clearly, Waymos overstate their self driving ability through a variety of very clever, very powerful sleights of hand.
Waymo shared that their disengagement rate is 17k miles per disengagement meaning they are driven by human test drivers once per 17k miles, across all causes. It doesn't include assists by humans because that's not driving: it's merely remote humans answering a question about where to drive in case of a novel obstacle on the road. You'd be nuts to think remote humans actually drive: the internet infrastructure is not designed for guaranteed low latency.
I also wouldn't trust NYT reporting on anything related to Big Tech: they have a well-documented pattern of anti-big-tech biased reporting.
> That's not driving: it's merely remote humans answering a question about where to drive in case of a novel obstacle on the road.
Listen to yourself.
It’s the difference between Star Craft and Doom.
Google Maps tells me where to drive. Does that count as driving? If so, Google has achieved self driving decades ago.
> Does that count as driving?
In this comment, you are saying the verbatim words "where to drive" to mean the navigation that Google Maps does, but then you used the verbatim words "where to drive" to mean resolving questions about obstacles. So I guess, yes, resolving questions about obstacles is indeed remote driving, and navigation questions from Google Maps is not remote driving. But everyone already knows that.
The sort of obvious definition of self driving means no human intervention whatsoever, which Waymos also fail. So I don't know. Why doesn't Waymo use Google Maps instead of people to tell the car, "where to drive"?
Hang on a second. You said that the disengagement rate is not zero. But also that remote humans never drive. So which is it?
The disengagement rate is calculated when a human test driver is in the vehicle and takes over driving. It's a specific term from autonomous vehicle testing. If a self-driving vehicle has no human drivers in it and it does not know how to drive, it simply pulls over if possible and stops. That's not among the definition of disengagement.
This has happened multiple times in San Francisco. You can find videos of driverless Waymo cars getting stuck. You can also find a hilarious instance where a stuck Waymo was driven by the emergency personnel after it became stuck.
What a terrible business, past, present, future, any way you want to stack it. Once whatever executive sponsors of this science experiment age out, you can expect this company to no longer exist. At this point, it's basically a redirection of wealth from Google shareholders to small group of people who are running a very expensive, medium size taxi cab company that needs a constant stream of outside capital to keep running.
The fact that they decided to use expensive Jaguars for their fleet just shows they are openly laughing in the faces of Google shareholders. They know how preposterous the entire venture is. At least they have had the foresight not to kill anyone (yet).
Big Tech is largely set of interlocking duopolies.
* Ads: Google/Meta
* Cloud: Amazon/MS
* Mobile: Google/Apple
* Desktop: Apple/MSFT
* VR: Apple/Meta
There's also smaller players and sometimes strong third players (eg GCP), but by and large there are no single winners. Waymo and Tesla are looking like a good bet on the self-driving duopoly. I think a lot of people are skeptical on Tesla because "LIDAR", but lidar is largely irrelevant. It is only used in 3D scene reconstruction, it has nothing to do with actual driving which is the hard part.
> lidar is largely irrelevant. It is only used in 3D scene reconstruction
Source? AIUI lidar is actively used as a sensor/input to detect absolute distance between things, which seems like 80% of what you need to drive safely.
Lidar produces a monochromatic 3D scene, which you can also extract with cameras using SLAM techniques. But these scenes simply do not carry the information needed to navigate the world - you only get one colour, so you cannot see signs, markings, lights, anything. LIDAR is thus a small, secondary input to the problem of "perception".
Once you've accurately perceived the world, the hard part is the action, and the judgement needed to drive safely. It doesn't matter how you got your labeled 3D scene (LIDAR or SLAM), you have to figure out how to navigate it safely and this is more like 90% of the self-driving challenge. If you look at videos of the latest FSD in action (12.5) perception is not an issue. Teslas see everything just fine, they don't disengage because they didn't see a car, but because their judgement is not human level yet.
Isn’t lidar how you avoid driving into the side of a white truck on a bright day when the sun is positioned just so?
That is one way, yes, there are others. But if you find my writing unconvincing, here's something you can try: Watch videos of FSD driving and note how often it fails to see objects like cars, trucks, people etc. FSD has 99 problems, but depth perception ain't one.
Can you stop talking nonsense. There is so much difference between cameras and lidar. Weather and fog being one clear example. It’s unbelievable what people will do to suck up to elon.
Completely agree. It’s frustrating how every thread on self-driving devolves into the same argument about cameras vs. LIDAR. Everyone points to Tesla FSD disengagements as evidence, when the vast majority of disengagements are caused by the route planning, not the vision stack.
I’ve watched countless hours of FSD footage and have FSD on my Model 3, and I’ve never seen it incorrectly map the position of a vehicle. Both Tesla and Waymo seem to have reliable perception stacks, the actual decision-making is the hard part, like you said.
I'm really glad that Google/Waymo isn't the only player in this space. Everything Google invests in is ultimately to benefit their ad business. If Tesla wasn't also competing here, the future would be robot taxis that relentlessly bombard you with ads from start to finish and we would have no choice but to tolerate it.
Alphabet != Google
Happy to see self driving companies raising rounds. Waymo is a leader, no doubt.
I recently took a Waymo ride in Scottsdale, AZ and it was amazing, I will definitely be taking Waymo over Uber wherever available.
This will cover another 4 years of R&D. They have a very long way to go before the economics of their business make sense. They are pretty far from making money on a per ride basis. They'll have to show significant per ride profitability before they can scale.
They could give the rides away for free, and it would not change anything. Somebody willing to risk $1b on Waymo doesn't credibly care about the kopecks they are earning on rides today. If they did, they would just sit on the sidelines, and wait until it is a public company, and buy post lockup expiry crash priced shares then.
Got a source on that? Their former CEO said they're likely already profitable in San Francisco.
I wonder what Waymo sees as the limiting factor to expanding to the top 24 US metropolitan areas in the next 12 months: affordable vehicle availability (e.g. tariffs on Zeekr), all-weather performance in northern climates, regulatory limitations?
I've heard that just acquiring and setting up the real estate and setting up depots takes a while. After all there aren't a bunch of robotaxi hubs sitting around in cities waiting to be bought or leased. That's where Tesla could have an advantage (if they could get robotaxis working), since in theory they could bootstrap using their supercharger network.
I haven't read much about Waymo depots, but from what I can tell they are : empty parking lot + chargers. If so, that seems like there would be ample options for this in almost any city. Amazon built a charging network in short order (https://news.ycombinator.com/item?id=40083622). Perhaps, their peak (kWh) charging requirements are higher than I would guess.
Nope, Waymo depots are secured garages in industrial zones.
I've been to two in the Phoenix area (Lost & Found FTW).
They are secure facilities near freight train tracks, and their neighbors are logistics companies. I saw Waymo vehicles coming and going, but I did not spot their point of entry into the building. There was no "big parking lot" full of Waymos or chargers visible from the outside.
The visitor/guest entrance is just a security desk and a waiting room with about 3 chairs. Even during business hours, the front door may be locked. You announce your purpose and the security guard admits you. I was once told that I'd need to wait outside, no problem: the weather was fine, and picnic tables were nearby. Once an employee retrieved my Lost & Found items, I was sent on my way.
I wish this wasn't private investment. Would be great if smaller investors had more opportunities to get involved
there is, $GOOGL stock
What's the exit here? They've raised over $10 billion at VC multiples, valuation around $50 billion. Do they expect this to be a $1 trillion company, even then the return would be a modest 20x.
I feel when that much money is at stake and such a high exit is expected, this could distort incentives. I feel like investors would prefer to set their money on fire for some tiny probability of that 20x+ return rather than look to actually make a modest profitable business
Do they expect this to be a $1 trillion company, even then the return would be a modest 20x.
The ideal end game for self-driving taxis is replacing personally-owned cars. Instead of anyone having a car of their own they'd just order one for any journey they make. If ordering a taxi is easier and cheaper than owning a car, and potentially faster and safer because it's self-driving, it's a no brainer for anyone who sees a car as a cost rather than a recreational vehicle.
If they manage to do that then Waymo will be significantly more valuable than a trillion dollars.
As long as US keeps the Chinese manufacturers out..
Baidu Apollo now has a $28K self driving vehicle [0], I doubt Waymo is ever getting close to those numbers. They might replace some US personal cars, but they won't be able to compete globally.
edit: Waymo wanted to use Zeekr vehicles before, I guess that ship has sailed now - they will have to get by with their new Hyundai contract [1]
[0] https://cnevpost.com/2024/05/15/baidu-apollo-launches-6th-ge...
[1] https://waymo.com/blog/2024/10/waymo-and-hyundai-enter-partn...
> I doubt Waymo is ever getting close to those numbers
Why? In theory, Waymo should be able to copy everything Baidu is doing, e.g. manufacture in a low-labor-cost country like India.
This is mainly about the technology, that's the hard problem. Optimizing the costs is something that has been done many times before and we know how to do it.
At some point the technology is good enough, then the cheapest car wins. Waymo doesn't have vertical integration necessary to rapidly iterate and it can't afford to take risks - it's still not operating their commercial service on the highways, not because the tech isn't there, but because of the liability and PR damages if there's a fatal crash (much more likely at highway speeds).
Baidu can have their test car kill a pedestrian every day for the next 10 years if necessary, that's not exactly an even playing field.
Waymo is providing the self-driving AI, it's currently tied to a specific hardware, but it will become adaptable to a wide range of hardware. It will be used in both cheap and expensive cars. If the technology is good enough, it will work with cheap cars too.
They will start using highways with regular passangers probably by the end of this year. Maybe Baidu can launch on highways with a 10x higher accident rate, but I don't think it's a significant advantage to the business.
When operations are scaled up and routine, robotaxis will have a short life, probably 3 years or less. Even at $100k/car every 3 years, that's probably 10% of LTV for the cars.
Uber knows this. They just did not have the money or tech Alphabet could bring to bear.
> When operations are scaled up and routine, robotaxis will have a short life, probably 3 years or less.
Can you elaborate on this? Do you mean due to wear?
Yes, wear: Robotaxis will probably go farther in a day than full-time rideshare drivers, maybe by 2X or more. Plus sensors and on-board compute will advance quickly. Waymo is on their 6th generation. 10 sets of tires in that time would not be surprising. That many miles also bumps up against battery lifespan.
But aren’t most cars on the road at the same high traffic times?
I need a ride a 9am and 5pm just like half the city. People keep talking about fleets being in demand and your Tesla becoming a taxi while you sit on your office but wouldn’t everyone else’s also?
$1T is about 6X Uber's market cap. If you can do Uber with more capex, less opex, 6X seems very do-able.
Didn't Uber raise over $30 billion from VCs before going public? It seems likely that they could continue for many more years at the current burn rate or even higher.
Google controls/dominates search, much of the mobile phone market, OS on many devices, some of our programming languages, build tools, infrastructure, maps, ads, video, email, and now transportation? Anything I’m missing?
Google (and Amazon, and probably a few Chinese players) have multiple revenue add-ons they can use with robotaxis.
[dead]
Telling that this comes right on the heels of the Robotaxi unveil. Google sees a threat.
This was first announced in July: https://techcrunch.com/2024/07/23/alphabet-to-invest-another...
Pretty sure Waymo sees very little threat from Tesla, who don't even have a single driverless mile on public roads.
You are being disingenuous, "summon" could be considered as fully driverless, so there are a couple of miles. They are getting there for sure, just not at the similar level, which is expected because customers do not care about it that much.
This is silly. The conversation is about L4 autonomy. Summon isn't fully driverless by any stretch of imagination. It requires the driver to be within a few feet and hold down a button in the app (essentially a kill switch). Moreover, the liability is still with the driver.
Disingenuous would be trying to counter an argument that Tesla has no driverless operations on public roads with a suggestion that a parking lot party trick counts.
average miles between intervention in similar driving scenarios would be a better metric. I’m struggling to understand how anyone thinks the robotaxi thing is competition though
There's a few other metrics people should care about:
"As of June 2024, there have been forty-four verified fatalities involving Autopilot and hundreds of nonfatal incidents."
Have there been any for Waymo? None that I can find.
Waymo is currently providing over 100K driverless rides a week.
Tesla has "concepts of a plan" to do something similar --- about where Waymo was in 2016.
https://www.forbes.com/sites/bradtempleton/2024/10/11/tesla-...
> In reality Tesla FSD is still in very primitive shape.
From the article. Have the people saying this actually tried the most recent FSD build?
I used it yesterday, and it reacted flawlessly to road construction that blocked a lane, then had cones demarcating a lane that didn't follow existing road markings. Then it saw that a driver wanted to turn left from parking lot while approaching a stop light and made space for the driver to turn out. It also flawlessly executed a merge involving turning into a suicide lane and waiting for traffic to pass.
It's very good at adapting to changes and can work on roads its never seen before. There is no other production car that can do what FSD can do currently.
Does this mean it's ready for full autonomy tomorrow? No, but I could use supervised FSD today if I wanted to drive 1600 miles from Austin to Toronto with hands mostly off the wheel.
The pace at which FSD is now improving is quite remarkable. In the long run it's clear that this approach is going to be WAY more flexible than Waymo's, because Tesla has so many more vehicles on the road collecting so much more data, and throwing vast amounts of compute at vast amounts of data is basically the best way to see rapid improvement for the foreseeable future.
But sure, "concepts of a plan".
No, but I could use supervised FSD today if I wanted ...
Musk has invented his own oxymoron --- "supervised Full Self Driving" --- which means not really full self driving at all.
Your car is ready to be a robotaxi --- as long as you are behind the wheel to supervise it.
If a driver is required, it's not a robotaxi is it?
It drives itself with zero control input from me, hands literally off the wheel and feet off the pedals. That's the definition of "full self driving".
It's bizarre how people are so adamant about this when they haven't even tried it. Every passenger I've shown this to have had their minds blown.
So sit in the back seat and let the car drive and my mind will be blown --- if you don't get arrested either before or after the crash.
Waymo actually does this today. Musk has plans to do this --- someday. It's mind blowing that you can't see the difference.
Would you have sat in the back seat? How about in a Waymo?
Could you drive at all in a Waymo outside of a strictly geofenced area in a handful of US cities?
How about in a Tesla?
They are approaching this from two fundamentally different directions.
Could you drive at all in a Waymo outside of a strictly geofenced area in a handful of US cities?
Absolutely not. But how many people want to hire a taxi to drive them cross country? Near zero.
They are approaching this from two fundamentally different directions.
Correct. One has a working robotaxi. The other has some marketing concepts.
This is exactly the point and I'm consistently astounded by how many allegedly tech-first people on HN just do not get it.
It's crazy that when I walk to the taco stand down the street in my sleepy little neighborhood I'll see waymos just hanging out waiting for a ride, and if I wanted I could just hop in one and be robot chauffeured around the city.
It's funny how the future arrives and intermingles with the present. Stuff I could only dream of as a kid, things that seemed impossibly far off are here right now.
I'll probably have a Rosie in my house before I die and that's pretty incredible.
Are the cars still teleoperated though?
If you mean "driven or steered remotely": They were never teleoperated. I'm assuming that would be completely reckless given the nature of radio networks and possibly because of regulations. The car will call a service once it can't make a decision (this is only when the car is fully stopped), and a human will decide to send out a driver or make a decision in the moment.
They never were. They get occasional labeling / decision making support from humans.
It’s funny, my Tesla drove me 30 miles yesterday and I didn’t have to do a thing. People in this thread have no idea how good FSD has gotten
You're still in the driver's seat ready to intervene. Others are doing it every 13 miles: https://arstechnica.com/cars/2024/09/tesla-full-self-driving...
Waymo's remote operators are still sitting in an office intervening and as a passenger you don't even know that it's happening
Except they can't make the kinds of safety critical interventions Tesla driver's in the seat can make i.e. prevent accidents. That's just physics.
We know exactly how their remote operation works: https://waymo.com/blog/2024/05/fleet-response/
If my Tesla swerves into oncoming traffic with no warning and I get in a crash, who is responsible?
If my Waymo ride does the same, who is responsible?
Yeah, so? You don't think they're going to keep improving it until supervision isn't required?
13 miles per intervention after 8 years of development suggests it’s not happening anytime soon.
These guys are all talk and no action. They wouldn’t dare risk their life in the backseat while fsd is turned on but happy to be a keyboard warrior for Elon. It’s hilarious.
Yup, this. I use FSD almost every day now that it's available for Cybertruck, and it's already quite excellent. It pulls off human-like maneuvers that are kind of gobsmacking.
Then jump in the backseat if it’s gobsmacking and just as good as a waymo. Tell it to drive you to the other side of the city while you sleep in the backseat. Put your money where your mouth is.
I can understand the average person not realizing what Teslas can do today, but I'm highly disappointed that so-called "tech" people on this message forum have no idea.
See the thing is they do and you don’t. If you don’t know and understand why waymo is far ahead of elons lies and will never catch up with its current tech direction and strategy for supervised fsd (which is an oxymoron) then you should only be disappointed in yourself. If you’re so confident of the tech go jump in the back of a Tesla and let it drive you 100kms. Let’s see if you live to tell the tale. I’m happy to also do the same but in a waymo.
The Robotaxi is a PR stunt which doesn't look remotely close to be a threat to Waymo right now.
lol ... right
Robotaxi hasn't deployed anywhere. There are lots of other startups ahead of them in terms of progress.
Waymo isn't afraid of a threat. Waymo is dominating the market they created. This is just another chess play to lock up the TAM.
Robotaxi will use the same stack as other Tesla vehicles, which is deployed on millions of vehicles. The <$30k price point and lack of controls/supervision are the only things that truly differentiate it from a Model 3.