Mercedes-Benz Beat Tesla to Cheap, Hands-Free Driving
bloomberg.com
bloomberg.com
"It is important to note that they are using the UNECE Level 3 certification loophole which technically demands that the company takes responsibility, but sets the minimum intervention interval for the driver at zero seconds for the under 60km/h case. That means, as long as the car exits its autonomous mode before a crash, the driver is instantly fully responsible. It's level 2 in all but name.
There are also practically no certification or safety requirements, hence why several manufacturers have announced their gimped level 2 system as level 3 when under 60km/h, among them Honda in 2021 and BMW in 2022, all of course claiming to be the worlds first."
https://arstechnica.com/cars/2023/09/mercedes-benzs-level-3-...
> If the driver fails to take back control even after increasingly urgent prompting and expiration of the takeover time (e.g., due to a severe health problem), the system brakes the vehicle to a standstill in a controlled manner while engaging the hazard warning lights.
From the CA DMV:
> Mercedes-Benz is the fourth company to receive an autonomous vehicle deployment permit in California and the first authorized to sell or lease vehicles with an automated driving system to the public.
We get it, "Tesla Bad." They are the only automaker selling EVs at volume besides Chineses automakers (which are limited to the Chinese auto market), and they are what consumers are preferring (with the Model Y being a best seller in various large markets). You might not agree with consumer preferences, but their dollars and how they vote with them are what matter, while regulators are sufficiently tolerant of Tesla's various Autopilot driver assist flavors (AP1, AP2, AP2.5, Enhanced Autopilot with Auto Lane Change) as well as their Full Self Driving beta to not require they be pulled from the US consumer fleet.
One can be "Elon musk is not a decent human being" but also "Tesla makes EVs people will trip over themselves to buy, and we need as many popular EVs as possible to contribute towards solving for the transportation emissions component of climate change." PR pieces do not destroy demand for petroleum or reduce combustion emissions, EVs shipped do.
Regulators are also sufficiently tolerant of MB's efforts (sorry, "puffery"):
> currently certified for use in California and Nevada
Yes, MB (who I have no love for) is ramping up EV manufacture. Tesla beat everyone to the punch.
> Tesla will sell ~2M units, 4x total Mercedes Benz annual volume.
Huh, going to disagree there:
> 2022 (January to December): Mercedes-Benz Passenger Cars delivered 2,043,900 vehicles
I stand corrected, and appreciate the correction as I strive for accuracy. My stat was wrong, pulling from a quarter vs the entire year. Tesla shipped almost as many EVs as Mercedes Benz shipped entirely, vs 4x.
What does that have to do with autonomous driving? The statements "Tesla makes more electric vehicles than Mercedes-Benz" and "Mercedes-Benz has a better autopilot program than Tesla" are in no way contradictory.
https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...
Here is the regulation they are talking about: https://unece.org/sites/default/files/2022-05/ECE-TRANS-WP.2...
On page 17, Section 5.4 they discuss: "Transition demand and system operation during transition phase".
In Section 5.4.3: "During the transition phase the system shall continue to operate."
In Section 5.4.3.2: "During the transition phase, the transition demand shall be escalated latest after 4 s after the start of the transition demand."
In Section 5.4.4: "A transition demand shall only be terminated once the system is deactivated or a minimum risk manoeuvre has started."
In Section 5.4.4.1: "In case the driver is not responding to a transition demand by deactivating the system (either as described in paragraph 6.2.4., or 6.2.5.), a minimum risk manoeuvre shall be started, earliest 10 s after the start of the transition demand."
In Section 5.5: "Minimum Risk Manoeuvre"
In Section 5.5.1: "The minimum risk manoeuvre shall bring the vehicle to a standstill unless the system is deactivated by the driver during the manoeuvre."
In Section 5.5.3: "A minimum risk manoeuvre shall only be terminated once the system is deactivated or the system has brought the vehicle to a standstill."
In Section 6.2.4: "It shall be possible to manually deactivate..."
In Section 6.2.5: "In addition to paragraph 6.2.4, the system shall not be deactivated by any driver input other than those described below..."
In Section 6.2.5.1.: "Deactivation by input to driving controls..."
In Section 6.2.5.2: "Deactivation during an ongoing transition demand or an ongoing minimum risk manoeuvre. In case a transition demand or minimum risk manoeuvre is on-going, the system shall only be deactivated: (a) As defined in paragraph 6.2.5.1. or (b) Upon detection that the driver has taken hold of the steering control as a response to the transition demand or minimum risk manoeuvre and provided the system confirms the driver is attentive as defined in paragraph 6.3.1.1."
So, the car remains in control during hand-off until a driver actively takes back control. The hand-off process must last a minimum of 10 seconds where the car is expected to operate as normal. In the event that the driver does not take control, the car is required to automatically come to a stop without relinquishing control. At no point can the car automatically transfer control to the driver except after fully executing a minimum risk manoeuvre and coming to a halt, or if a emergency manoeuvre is required (at which point the emergency manoeuvre system is in control).
>On suitable freeway sections and where there is high traffic density, DRIVE PILOT can offer to take over the dynamic driving task, up to speeds of 40 mph...DRIVE PILOT available in the U.S. for model year 2024 S‑Class and EQS Sedan models, with the first cars delivered to customers in late 2023
So very limited driving scenarios (essentially only stop and go traffic on freeways) and extremely limited product availability.
https://electrek.co/2023/09/27/hands-off-with-the-first-true...
>This brings up a potential ownership consideration: You might want to keep the navigation lady’s voice turned on to warn you if you’re going to miss your exit, as one thing level 3 won’t do is change lanes.
>The system will be available on Mercedes’ EQS and S-Class (but who would want that one? it runs on gas, yuck) in the US later this year. Significant hardware is required and can’t be retrofitted, so you’ll need to get a level 3-capable car on purchase. The hardware itself won’t change the vehicle’s price, but using the system requires a subscription, which costs $2,500 for the first year. Mercedes hasn’t yet committed to prices for subsequent years.
https://arstechnica.com/cars/2023/09/mercedes-benzs-level-3-...
I think Musk's insistence on a camera only system will ultimately be a huge anchor around Tesla's neck for self driving.
I hate how this has gotten advertised as way better than a Tesla but actually it’s much much worse than a Tesla in terms of its actual ability to drive in all conditions.
Adaptive cruise control + lane keeping + automatic lane change is, from a driver perspective, feature wise 80% of the way to highway self driving but they're all limited by the requirements that they're not hands-off features. That last gap of fully letting the user remove their hands from the wheel, explicitly, and assuming liability is putting a lot more trust in the system than your 2020 Civic.
If they were as cavalier about driver safety and liability as Tesla they probably could have looser restrictions on it and just push the job of being a perfectly alert safety driver onto their customers..
They also guarantee 10 seconds between disengaging and requiring your control: meaning they're assuming liability during disengagements.
They handle emergency vehicles, using external microphones to yield right of way for lights and sirens autonomously.
_
Overall assuming liability took solving a lot of problems that laypeople keep assuming are trivial because L3 looks so similar to L2. I work in the AV industry and Drive Pilot leapfrogged a lot of offerings.
BMW is just now working through European certification for the 7 series, so their original estimate of 2025 doesn't seem far off for US markets.
Tesla apparently doesn't even want to try at this and is content with stretching the L2 rubber band as far as it can while pushing flashy demos
https://unece.org/sites/default/files/2022-05/ECE-TRANS-WP.2...
It's the second time in that much week that Ars Technica blatantly lie about what's in the primary source, maybe trusting nobody would check. I'll wait until my subscription finish, but i won't be retaking one next year.
Since all the premapped routes are highways with speeds well above 40 mph and given that a “follow car” must be directly ahead of it, this is just for highway traffic jams under ideal conditions. Useful but very limited.
Disagree that camera only will prove to be a mistake. Vision is how every car is operated and avoiding additional sensors simplifies development, reduces cost improving scalability.
The benefits of LIDAR and RADAR is you’re dealing with less data and it comes with depth so you don’t even have to figure out binocular vision.
Elon Musk didn't invent EVs or Tesla, he bought it.
Not saying Tesla is bad, but its market value was completely pumped up by Elon's tactics of turning $TSLA into a hype driven speculative asset like Dogecoin versus gaining it from achieving sales succes on its own.
There's no shortage of companies, many of which HN readers are employed by, that did the same.
I'm no fan of Elon - I cancelled a Model 3 order and went with an EV6. However you only have to drive around on the freeway for a few minutes to see the success Tesla has had, despite their stock price or how shitty of a dude Elon is.
So Musk showed up, financed it himself and he brought easily the most important engineer, JB Strauble, with him.
> hype driven speculative asset like Dogecoin
Then you made a killing on shorting it? If you knew so much better then the market?
Because Tesla stock was super high when pretty much all EV stocks were incredibly high. And all companies around that industry (and others) came back down. But its also not like Tesla collapsed to some historic lows after that.
And if you look at actual Wallstreet annalists from that time period you will see that most of the valuations wasn't based on some speculative FSD stuff. It was mostly based on their production, margin and growth.
> versus gaining it from achieving sales succes on its own.
So it had nothing to do with going from a 1000 cars a year to 2 million cars a year, and while doing that having highly positive margin in an industry where (almost) nobody else makes positive margin on EV, and at the same time also continuing to grow fast?
All of those things simply happened because of hype like Dogecoin? Are you serious?
Meanwhile, Tesla FSD is driving a million miles per day around town, not playing follow-the-leader on a highway.
I find one of these significantly more impressive than the other.
- Musk, 2 years ago: FSD will yields for emergency vehicles using microphone
- Today: Mercedes reaches L3 before Tesla passes L2 and FSD does not yield for emergency vehicle sirens while Drive Pilot does.
There's a trend on these things but I can't quite put my finger on it...
Drive Pilot uses external microphones to detect the approaching siren and manages right of way before making visual contact.
Not all drivers understand the value in the latter though.
[1] https://twitter.com/elonmusk/status/1414436167639093248?lang...
"Everyone else is stupid and I'm smart" is not exactly the rock-solid argument you might think it is.
It will never stop being surprising to me to see, on HackerNews, such baseless vitriol.
I have driven other EVs. The torque is not universal at all, especially considering I have AWD.
The very low center of gravity, combined with the torque and AWD, makes for very fun rides in mountain roads.
The autopilot is really convenient in highway driving, and is much better than the driver assist features I've experienced in other cars (Honda, Fiat, MG).
Interacting with the screen is natural, snappy and very intuitive, as opposed to most other cars I've interacted with. Being able to use the screen as an entertainment center (e.g. play youtube cat videos while charging) is fun.
Regular software updates are seamless. Hopping in the car and getting a list of new features is fun. Maybe the most fun was the time a software update shaved half a second from my 0–100 time.
The steering wheel is simpler to use (just two knobs instead of tens of buttons spread around) and the screen is much cleaner to read than other cars that try to ram too much information in too-small displays.
The audio quality is much, much better than any other car I've been in.
How long would it take to do this drive and how many interventions would I need to do?
The press release [1] seems to indicate that this is just for freeways, which I thought was the easy case that has been mostly solved (in a technological sense) for years by a bunch of different players.
[1] https://media.mbusa.com/releases/release-1d2a8750850333f086a...
I assume this ultimately is just an expensive way to relax these requirements?
Still good to see car companies push out improvements like this. Next version should be more capable, then the next version… Eventually we start to reduce the 1.3 million deaths and 10s of millions of injuries every year.
The difference is that the company is putting their money behind their promises. It is like you want to buy a product , one company gives you 5 years warranty and the other offers 0 days warranty but they show you some ads and cherry-picked stats.
The service lifetime of even an ICE car is over 10 years.
I can't read the actual article because the paywall bypasser's I've tried all use a captcha that additionally have tracking and I can't get through them.