It's remarkable how often these significant limitations are ignored.
The difference between SAE Level 3 versus Level 2 is liability, not functionality. Conceptually, it would be relatively simple to create a Level 3 system that only worked in parking lots and never drove over 5 MPH. And yes, such a system would be "Level 3," which naively sounds better than "Level 2" because the number is higher.
But you could compare such a system that works only in parking lots against Tesla's Supervised FSD Level 2 system which controls the vehicle at prevailing speeds on all city streets and highways, executes left and right turns, stops at traffic controls, executes U-turns, parks, and everything else. Tesla doesn't want to assume liability yet because they are still iterating fast. The last few builds have been released about two weeks apart.
The functional domain of a Level 2 system can be significantly greater than a Level 3 system. And that is the case when comparing Tesla Supervised FSD versus Mercedes Drivepilot. It remains commendable for Mercedes to take on liability, but we should not kid ourselves. Doing so is a play for cheap media wins and punchy-sounding milestones like what we see in this article. It's not actually moving autonomy forward in any meaningful way when compared alongside Waymo and Tesla FSD. The scope of the Mercedes system is simply too narrow.
No one’s going to take liability if their technology is not ready for the conditions they want to work in. And no one’s creating systems that only work in 5 mph parking lots just to claim the L3 title. So comparison to hypothetical systems aren’t useful.
Only one of them allows you to take liability.
I would take a car with reliable, narrow L3 and reliable broad L2 (especially when the L3 is reliable in the most frustrating driving conditions to encounter) over broad pseudo-L3 any day.
Waymos are incredible, but Teslas are a much worse tradeoff than what MB is proposing here in my opinion.
The Mercedes system will absolutely not follow navigation routes, change lanes, execute right and left turns, execute U-turns, stop at traffic controls, and so on. All of which FSD will do.
A broadly capable Level 2 system that is suitable for 99% of driving scenarios (with supervision) is way more interesting than a tightly constrained Level 3 system that is only suitable for <1% of driving scenarios.
No it's not. The difference is functionality, proof that the vendor believes they've achieved that difference is most clearly communicated via liability.
> The scope of the Mercedes system is simply too narrow
Producing systems that perform reliably under the prescribed conditions is called "good engineering."
I would guess the reason for not wanting to assume liability is not their update cadence, but the same reason that made Mobileye fire them as a customer.
More range and more driving conditions just don’t matter here if I can’t read a book on the way to work then it’s not self driving just advanced cruse control.
Every dollar spent on private cars reduces that infrastructure, and that's not just my own thinking, but also that of car manufacturers, as evident by their lobbying to derail public transit projects.
Further I don’t live in a city so my local government has zero interest in public transportation beyond school buses. We just don’t have the density where it makes any sense. When I did live in a city the issue wasn’t drivers it was the inability to scale roads for local demand. Self driving cars don’t change anything about that equation.
...those same Mercedes cars that have Level 3 also have a much more "powerful" and flexible Level 2 system built in, but this requires to keep a hand on the wheel and pay attention to the road (mentioned towards the end of this video, at around 6:15: https://www.youtube.com/watch?v=ZLz95Gw7g8c)
I originally thought autonomous cars would greatly reduce costs but considering the tens of billions dumped into these companies, they can't charge lower prices as they have to recoup investments.
> As of the latest available information, Waymo is not yet profitable. Alphabet’s Q2 2023 earnings report indicated that its “Other Bets” category, which includes Waymo, had revenues of $285 million and an operating loss of $813 million. This was a decrease in operating loss from $1.34 billion in the second quarter of 2022, but still significant. Over the last five years, “Other Bets” has generated $3 billion in cumulative revenue but incurred $20 billion in operating losses, with Waymo constituting the bulk of these losses.
I just don't see the end game. Are marginal rides not profitable even at Uber fates? Uber for instance was losing money because they were taking rev from profitable cities to give incentives and build up drivers in the new markets. But Waymo doesn't have that problem. Is it just GPU spend that they can eventually stop?
The business model is pretty simple. Robotaxis can operate much cheaper than humans and simultaneously provide a better service. Competing at the same cost as existing services makes them massively profitable. You can unlock additional efficiencies through things like fleet management, not supporting unprofitable cities the way Uber has to, vehicles designed for longer service lives and so on. Most companies plan to use that market (and closely associated ones like food delivery) as a profitable beachhead to scale down the technology to the point where it can be applied more ubiquitously in things like consumer vehicles. Some are more vocal about this than others.
Self-driving cars seem like a poor choice of market, since the task is very complex and there are also many people worldwide who are willing to work as drivers.
The sweet spot would be a task that is easy to automate and also doesn't have much human competition. Obviously the OG automation case of assembly lines fits this. LLMs can occupy this space since there are probably more people who can drive a car than can write grammatically correct natural language. I would put self-driving trains in this category as well but that's an outsider's perspective.
In general, it's a mistake to think that selling price has anything to do with the costs. The price is simply what the market will bear. If people are paying Uber $x for a service, waymo would be stupid to price the same service much lower than that. If their costs are much lower, they would rather make that much more profit. If the rider wants a small piece of that pie they can invest in the business (via stocks in the case of Google).
They don't need that kind of bad publicity when potentially the licenses they operate under could be pulled given enough political pressure.
I wouldn't consider current pricing indicative of what it might ultimately be, consider that if aggressive enough google could divert a big portion of the countries car spend to themselves.
Waymo driving in SF is way better than human. It's not timid either.
This Mercedes post is marketing - it's not really what people consider autonomous given the constraints.
> "Drivers can activate Mercedes's technology, called Drive Pilot, when certain conditions are met, including in heavy traffic jams, during the daytime, on specific California and Nevada freeways, and when the car is traveling less than 40 mph. Drivers can focus on other activities until the vehicle alerts them to resume control. The technology does not work on roads that haven't been pre-approved by Mercedes, including on freeways in other states."
The above makes it nearly useless and far less useful than what Tesla has been shipping for years imo.