Mercedes Temporarily Scraps Its Level 3 'Eyes-off' Driving Feature (theverge.com) 46
Mercedes-Benz is pausing the roll-out of Drive Pilot, an "eyes off" conditionally automated driving feature that was available in Europe and the US. From a report: As first reported by German publication Handelsblatt, the revised S-Class will not have the Level 3 system when it arrives at the end of this month. Mercedes was one of the first automakers to offer a Level 3 driving system to its customers when it launched Drive Pilot with the electric EQS sedan and the gas-powered S-Class in the fall of 2023. At up to 40mph in traffic jam situations on highways, Drive Pilot provided hands-free, eyes-off driving that allows the driver to look away from the road at something else, like a game or a movie.
It was big leap up from hands-free Level 2 systems -- Tesla's Autopilot and Full Self-Driving (FSD) included -- which still require the driver to be in full control, looking ahead and paying attention while the system is active. But now Mercedes says it is temporarily scrapping the feature, citing middling demand and the high production costs of developing the technology.
It was big leap up from hands-free Level 2 systems -- Tesla's Autopilot and Full Self-Driving (FSD) included -- which still require the driver to be in full control, looking ahead and paying attention while the system is active. But now Mercedes says it is temporarily scrapping the feature, citing middling demand and the high production costs of developing the technology.
I still remember 2018 (Score:2)
When fanboys were telling me that all cars would be self-driving by 2020.
Re: (Score:2)
This accelerates the timeline to FSD versus the path they were on. They're going to have better self-driving, just not using the Drive Pilot system. They couldn't do eyes off in street level driving so easily as on highways. The highway driving experience should be similar. They are implementing "Level 2++" ... "Drive Pilot Assist", that brings in street driving not just highways .. and in within a couple of years they will skip level 3 and go directly to level 4.
Re: (Score:2)
i don't see any kind of acceleration in FSD. ,FSD is 2 years away.
In 2018, FSD was 2 weeks away.
Today
Re: (Score:2)
Nobody outside of Arkham Asylum for the criminally insane said in 2018 that all cars will be FSD by 2020. We already have FSD, ride a Waymo and see for yourself. People misjudged to the time it will take to solve and distribute full self driving. They didn't misjudge that it could be solved.
Even Tesla, which has the cheapest and fewest sensors possible is almost fully level 4 FSD capable.
Re: (Score:2)
> Nobody outside of Arkham Asylum for the criminally insane said in 2018 that all cars will be FSD by 2020.
Fanboys in car forums did. I particularly remembered it because it was so absurd.
Is this another of those "even though you were there and saw it, that definitely couldn't have happened" things?
Re: (Score:1)
This is the internet.
Link please.
Re: I still remember 2018 (Score:4, Informative)
Tesla isnt even close to level 4, it can barely manage level 2 sometimes. As for Waymo- sure, on nice wide US streets set out on a grid system. Good luck in london, paris or rome with narrow streets that often only have room for 1 vehicle to pass each way at a time and lots of hand gestures and headlight flashing to indicate intent. Perhaps take a break from the tech bro propaganda occasionally.
Re: I still remember 2018 (Score:4, Interesting)
Good luck in london, paris or rome with narrow streets that often only have room for 1 vehicle to pass each way at a time
My experience with Waymo is that they're really good at navigating through areas like this. The sensors do a great job of knowing where exactly the car is, and how to maneuver around stationary things, or other traffic that's moving predictably around it.
lots of hand gestures and headlight flashing to indicate intent.
This is where they shit the bed. It's going to get confused, stop, and just completely block the road. The last time I took a Waymo, it was about a mile trip, I was just curious and wanted to try it. I was honked at three times, and as it was pulling to the curb to let me out a valet driver pulled away from the curb, which caused the Waymo to freeze and it told me to get out in the middle of Grand Ave. in downtown Los Angeles. It did helpfully warn me that a car was coming and to be careful though... After I got out it proceeded to sit there for about a minute, blocking the lane as it tried to figure out what to do. It was honked at again multiple times before it finally pulled away.
Some parts of their tech is pretty cool. But it is very much not ready for prime time.
Re: (Score:2)
"My experience with Waymo is that they're really good at navigating through areas like this. The sensors do a great job of knowing where exactly the car is, and how to maneuver around stationary things, or other traffic that's moving predictably around it."
I don't think you understand the kinds of roads I mean. How is it going to manage this:
https://maps.app.goo.gl/88wPR4... [app.goo.gl]
If there are vehicles coming the other way? Its way too long to just wait at one end until its clear. I'm afraid the tech bros just hav
Re:I still remember 2018 (Score:4, Informative)
ride a Waymo and see for yourself
I have. It spent about 5 minutes waiting to turn down a closed road that was barricaded off for a farmer's market before it finally called tech support and had a human route it to the next corner. Also have you seen the stories recently with Waymos driving down light rail tracks [people.com], through police standoffs [go.com], and onto active fire scenes [abc7.com]? They do have some impressive tech, and have good reflexes to avoid immediate accidents, but they're not remotely ready to be running loose on our public streets. Not to mention they get confused and just essentially shut down if power knocks out traffic signals [cnbc.com], and will just completely stop if someone places a traffic cone on their hood [npr.org].
Even Tesla, which has the cheapest and fewest sensors possible is almost fully level 4 FSD capable.
Lol, no they're not. They're not even close.
Re: (Score:2)
Waymos are "ready for public streets", because guess what they are already on public urban streets in cities like LA, SF, Austin, Phoenix etc. and have a better safety record per mile than human drivers. https://waymo.com/safety/impac... [waymo.com]
They made amazing progress even though so many people tried to block it in every possible way .. legislatively and by creating uncertainty to make investor afraid.
As for those one-off incidents .. I've seen humans do far worse and it doesn't even make local news. As long as
Re: (Score:2)
Almost enough of them to cause gridlock now during a power/cell outage.
Re: (Score:2)
They learn from those incidents and improve .. jeez. I don't understand you luddites need to squish something at its early seed stage. Let's revisit in 10 years and see if they have improved. It ain't right. Imagine if cars or airplanes got banned the first time they had an accident or caused a problem.
Re: (Score:2)
Well, there is this one really prolific guy:
https://en.wikipedia.org/wiki/... [wikipedia.org]
e.g. "I consider autonomous driving to be a basically solved problem. ... We're less than two years away from complete autonomy. Regulators however will take at least another year; they'll want to see billions of miles of data." - June 2016
Perhaps none of this long list of quotes is precisely what you were rebuffing. B
Re: (Score:2)
Musk was basically correct (though not because of Tesla). Waymo launched in Phoenix in 2020. My point was nobody sane ever said all cars would be FSD like what the OP claimed.
Re: I still remember 2018 (Score:2)
Waymo only operates in cities where they have mapped the streets beforehand. It is a milestone, and they deserve accolades for doing what they're doing.... but there is a fence, and the car won't drive beyond it.
We're still quite a ways from having cars that can drive any arbitrary route.
Re: I still remember 2018 (Score:2)
Elon was in Arkham??!
Re: (Score:2)
The only appropriate asylum for him.
Re: (Score:2)
Yeah that would have been a stupid thing to tell you considering we already had driverless cars roaming the streets in 2018.
Just great (Score:2)
Now when will I take my nap?
Re: (Score:2)
Nonsensical (Score:5, Interesting)
Why would the high development costs *of something already developed* mean its *roll-out* would be scrapped ?
Would the author of the summary care to explain ?
Quite Sensible (Score:4, Interesting)
They cited both middling demand and development costs as reasons for killing it.
The problem with Level 3 self-driving is that it's mostly useless. It can't control the car in all conditions. It may require the vehicle to navigate to safety and/or return control to the driver. Hardly anyone really wants that. It's not ubiquitous and reliable. You can't plan around it.
People really want Level 4, which is fully autonomous driving at human-equivalent levels in all conditions. But that's a much harder problem to solve.
Meanwhile, you're dealing with the potential liability from Level 3 self-drive without a huge increase in sales to justify it. This is a classic risk-vs-reward calculation.
I expect Mercedes-Benz to continue working toward Level 4 self-drive, but, personally, I've never seen the point of Level 3. I certainly wouldn't pay for it, and apparently I'm not alone in that respect.
Re: (Score:2)
They cited both middling demand and development costs as reasons for killing it.
As another post mentioned, development is a sunk cost. However, demand is a valid issue. Mercedes' Level 3 requires a bunch of additional hardware, including lidar. In the US, the hardware is included on the already expensive car but also requires a $2500/year subscription.
The problem with Level 3 self-driving is that it's mostly useless. It can't control the car in all conditions. It may require the vehicle to navigate to safety and/or return control to the driver. Hardly anyone really wants that. It's not ubiquitous and reliable. You can't plan around it.
People really want Level 4, which is fully autonomous driving at human-equivalent levels in all conditions. But that's a much harder problem to solve.
People would love Level 5 with no steering wheel at all. However, it's incorrect to say that the absence of Level 4/5 kills all demand for Level 3. After all, the absence of Level 3/4/5 in no way kills the demand for ADAS.
Re: (Score:2)
Re: (Score:2)
The problem with Level 3 self-driving is that it's mostly useless. It can't control the car in all conditions. It may require the vehicle to navigate to safety and/or return control to the driver. Hardly anyone really wants that.
If people understood it they'd realise that a significant portion of the population do in fact want that. L3 is a hands off with driver retaking control in 10 seconds. If you drive on the highway for any significant amount of time, L3 driving will effectively allow you to read a book, use your computer, or watch a movie. All are activities that will allow you to regain control in that time. For a not insignificant portion of the population that would mean that *most* of the driving is fully hands off and no
Re: (Score:3)
The problem was they could only really deliver level 3 as a duckling follow mode behind vehicles at consistent below limit speeds. So in jams and behind trucks. The moment your lead left it immediately needed to go to ten seconds take over. So very limited use with frequent interruptions when it was usable. Proper level 3 which doesn't require a lead vehicle would be very useful.
Unfortunately even non castrated level 3 is AI hard and if we get true AI everything changes completely, worrying about integratin
Re: (Score:2)
too many unexplained unintended acceleration events, check out some mbeq boards.
Not development costs (Score:2)
Legal costs after it inevitably fucks up at some point.
Re: (Score:2)
I also indent C++/Java code following Whitesmith's rule: https://en.wikipedia.org/wiki/Indentation_style#Whitesmiths
And I also put back the missing apostrophe in "Whitesmith's".
Because I like to live on the edge, play by my own rules, be my own boss. That's just how I roll, dudeman.
Anything short of... (Score:3)
Anything short of being able to drive me anywhere with full autonomy is more or less useless to me. If I have to pay attention at any point during the drive I might as well just drive myself and save the money for the service.
Re: Anything short of... (Score:1)
Waymo had all that in 2015
Re: Anything short of... (Score:2)
For very small values of "anywhere."
They still don't operate outside of areas that they have mapped beforehand.
Re: (Score:2)
With stopping as a fail safe and remote controllers to get them going again, not an option on the highway.
(I don't care they can't directly control steering, they are still remote controllers and necessary.)
Re: (Score:2)
Congrats your comment is silly. A significant portion of the population spend their times in situations where L3 driving would give them an entirely hands off approach for the majority of their trips. For example this technology as advertised and demonstrated would allow me to sit and read a book for 35minutes out of my 40minute commute. The idea that it is useless just because I have to spend 5 minutes driving in a situation where it was clearly advertised to not work is just absurd.
Re: (Score:3, Insightful)
Probably not. Your hands are already occupied, your feet are nowhere they need to be, and there is probably something unusual happening outside the window to put you into panic mode.
Unsupervised driving needs
Re: (Score:2)
"Useless to me" was the wording. Be more mindful before getting outraged.
Re: (Score:2)
Anything short of being able to drive me anywhere with full autonomy is more or less useless to me. If I have to pay attention at any point during the drive I might as well just drive myself and save the money for the service.
And that's why manufacturers are turning off from it.
Getting it to human equivalence, even with the terribad drivers of the US let alone to the much better but still intolerably low standards of British or German drivers is something that is very, very, very, very, very far away. AI fanbois don't want to admit it, but it's true. It's not even a question of safety, autonomus cars fail at basic decision making hence they just freeze in the middle of roads, intersections, et al. awaiting human instruction b
Level 3 Pointless (Score:2)
The benefits of autonomous driving only really come from level 4 or 5. If I can summon my car from across town or sleep in the back during a road trip, that provides real benefits to me as the owner. But sitting in the driver's seat staring at the road while waiting to take over at a moment's notice does nothing for me. If I have to be ready to take over, I'd prefer to just drive myself.
I got a Comma 3X instead (Score:2)
A rare consumer product that does exactly what it promises.
It 'just works' keeping the car in-lane, hands free, and that's it.
Under a grand, a 15 minute install, and well integrated so it feels like the car came with it stock.
Beautiful work by George Hotz and the team at Comma.ai
Dupe! Well sort of (Score:2)
This was already announced 6 months ago. We even covered it here on Slashdot https://tech.slashdot.org/stor... [slashdot.org]
Re: (Score:2)
That was Stellantis, not Mercedes