Stellantis Shelves Level 3 Driver-Assistance Program (reuters.com) 70
Stellantis has put its fully developed Level 3 driver-assistance system on hold due to high costs, technical hurdles, and weak consumer demand. Reuters reports: As recently as February, Stellantis said its in-house system, which is part of the AutoDrive program, was ready for deployment and a key pillar of its strategy. The company said the system, which enables drivers to have their hands off the wheel and eyes off the road under certain conditions, would allow them to temporarily watch movies, catch up on emails, or read books. That Level 3 software was never launched, the company confirmed to Reuters. But it stopped short of saying that the program was canceled.
"What was unveiled in February 2025 was L3 technology for which there is currently limited market demand, so this has not been launched, but the technology is available and ready to be deployed," a Stellantis spokesperson said. The three sources, however, said that the program was put on ice and is not expected to be deployed. When asked how much time and money was lost on the initiative, Stellantis declined to say, responding that the work done on AutoDrive will help support its future versions. [...] Stellantis said it is leaning on aiMotive, a tech startup it acquired in 2022, to deliver the next generation of the AutoDrive program. Stellantis declined to say when that program would be ready for market or if it would include Level 3 capability.
"What was unveiled in February 2025 was L3 technology for which there is currently limited market demand, so this has not been launched, but the technology is available and ready to be deployed," a Stellantis spokesperson said. The three sources, however, said that the program was put on ice and is not expected to be deployed. When asked how much time and money was lost on the initiative, Stellantis declined to say, responding that the work done on AutoDrive will help support its future versions. [...] Stellantis said it is leaning on aiMotive, a tech startup it acquired in 2022, to deliver the next generation of the AutoDrive program. Stellantis declined to say when that program would be ready for market or if it would include Level 3 capability.
I can't see insurance companies (Score:2, Troll)
A self-driving car will follow every rule to the letter and it will refuse to speed.
I do wonder if you will see higher speed limits though. Most speed limits are set intentionally low with the understanding that everyone is g
Drank the Flavor Aid (Score:2)
Allowing us to drive much longer. Once self-driving cars are common and assuming our wider civilization doesn't collapse for a laundry list of reasons then insurance companies are going to make it prohibitively expensive for you to be able to take your car off self-driving mode. A self-driving car will follow every rule to the letter and it will refuse to speed. I do wonder if you will see higher speed limits though. Most speed limits are set intentionally low with the understanding that everyone is going to go 10 mph over. Hell I had to do a long drive today and it been a while since I did a really long one so I was driving slower than usual to be careful but found I was still 10 over while people were passing me on the freeway
Allowing us to drive much longer. Once self-driving cars are common and assuming our wider civilization doesn't collapse for a laundry list of reasons then insurance companies are going to make it prohibitively expensive for you to be able to take your car off self-driving mode.
You'd have a point if this were a fantasy world, but in the real world, if you can't drive as well as glorified cruise control, you shouldn't have a license in the first place. These glorified cruise controls drive about as well as a teenager with a learners permit and ADHD at best. On a good day. And now that tech companies and car companies are all in on generative AI, you're seeing the peak capabilities, it's all downhill from here until gullible people (you included in this case) come back to their s
I don't think you understand (Score:3)
Insurance companies like any business want predictability. They want to know XYZ is going to happen so they can plan for it and structure their pricing around it.
Taking squishy humans out of the equation and replacing them with unfeeling machines would be a huge leap for predictability Furthermore following all the rules will drastically reduce the scale and the damage from accidents. They don't care if there are
Re: (Score:2)
It's not about how well the cruise control drives it's about how predictably it drives.
Oh, if that was the criteria, insurance companies would never cover them. Damn things think trucks turning left on an expressway is a billboard and attempt to drive under.
Re: (Score:2)
A self-driving car will follow every rule to the letter and it will refuse to speed.
You really should watch that video of a Tesla plowing right through a fake wall.
Which also brings me to: Have you ever actually priced insurance for a Tesla? It's horribly expensive. It's like the insurance companies know you're likely to be distracted by the touch screen UI or eating a meal with the self driving enabled, and wind up upside-down in a ditch.
By the time self driving tech works well enough to deploy on a large scale, you won't own a car anyway. You'll just summon one via an app and insuran
Those are just pump and dump stock scams (Score:2)
I suspect that they don't handle bad weather which is why they're in Arizona because it rains like once a year there but that just means you have people take over when it rains and you have sensors to detect the rain
Re: I can't see insurance companies (Score:1)
Re: (Score:2)
hahaha no. (Score:5, Insightful)
There is essentially unlimited demand for this technology when it works.
There is no demand for the technology if it does not work.
What they are telling us is that it does not work.
Given that it's Stellantis, it comes as even less than no surprise.
Re: (Score:2)
There is no demand for the technology if it does not work.
That was my first thought as well. "We spent ungodly amounts of money to get this to work. It's ready to ship, works perfectly, but we're not going to release it because nobody wants it" just doesn't pass the smell test. I'm going to guess it's more along the lines of they spent ungodly amounts of money to get it 95% done, realized it was that last 5% that was going to double their budget, and then shelved it.
Re: (Score:2)
Re: (Score:2)
Re: (Score:3)
Why expend that budget when there is another company out there very close to FSD likely to get it done (already demonstrating it in use)
I'm not sure Tesla is likely to get it done. Maybe, maybe not. In theory it's definitely possible for a camera-only system to be as safe as a human driver, but that's not enough. It has to be safer than a human driver before we'll accept it, and I don't know if Tesla can get there -- and they obviously haven't reached the lower bar yet.
Note that I own a Tesla Model S and use FSD (supervised) all the time. It's marvelous. I especially love it for long road trips because it's basically flawless at freewa
Re: (Score:2)
Re: (Score:2)
Yours sounds worse than my current car (2025), but about the same as my previous car (2020). I think the difference isn't the software but the hardware. More CPU horsepower.
It is good enough for me to use all the time, and it's fairly smooth (unlike the 2020), but, still it's definitely not good enough to be unsupervised.
Re: (Score:2)
There is essentially unlimited demand for this technology when it works.
False. A technology working does not generate infinite demand. You left out cost. Cost is an important factor here, especially since their system was only slated for inclusion with top of the line EQS models which start at around $125k in the USA, and it wasn't clear that trim included it at that price either.
They were relying heavily on China to make money on this as high end luxury sedan sales cratered around the world. The Chinese government recently put the brakes on that for all car makers by putting a
Re: (Score:2)
Not only have they demonstrated it working plenty of times
You fell for the demo?
The downside in Germany was they were only approved for operation at up to 95km/h which is a non-starter for many people looking to use this since the Level 2 driver assist systems was approved to 150km/h. America, Canada and the UK did not place this condition on them.
They only even claimed it worked up to 37 mph [stellantis.com], and further only claimed it would work up to 59 mph. Nobody had to place this condition on them, they placed it.
Re: (Score:2)
I did not fall for anything, they made a demonstration which worked as shown and they have a certification for their system from multiple regulators. People with far more knowledge on the matter have said it works and even certified it to be activated on public roads.
What do you have? I mean other than a baseless hate boner and ignorance?
Re: (Score:2)
What do you have? I mean other than a baseless hate boner and ignorance?
A citation and logic. Both are pretty neato in my book.
What do you have besides insults and appeals to authority which don't even come with the identities of the supposed authorities?
Re: (Score:2)
Except your cite doesn't match your claim and your logic also failed miserably.
What do you have besides insults and appeals to authority
An "appeal to authority" is only a logical fallacy when the authority is presumed by the person making the claim. I'm not appealing to authority. I am stating that authorities that define requirements for road safety have declared the system works, you on the other hand have just blindly posted claiming it doesn't. If you want us to trust you on this then it is in fact you applying the appeal to authority fallacy, pointing to act
Re: (Score:2, Troll)
you on the other hand have just blindly posted claiming it doesn't.
I posted a link. You haven't posted shit. The link directly contradicts what you said. Put up or fuck off.
Re: (Score:2)
There is essentially unlimited demand for this technology when it works.
This is exactly why the BC Skytrain is so popular (and Detroit could also leverage this if they bothered to expand the Peoplemover, which was basically built with surplus Skytrain equipment).
Re: (Score:2)
I've often said that we've had the tech for self steering since the 1800s, and that really reduces the tech needed for self driving, as you have pointed out here.
I still will go ahead and spend my karma pointing out (as I have done previously) that we could be building PRT on an ultralight rail to make use of these technologies on a scale similar to existing automobiles, and even keeping the automobile companies in the loop on it, but big oil won't have it because it makes it easy to cut out the fossil fuel
Re: (Score:2)
Re: (Score:2)
Cars can function if you have shitloads of road, there's no reason why car-sized vehicles can't work if you make the track a lot cheaper. The Vegas loop doesn't work because it uses shitty cars on tires on roads in tunnels. Morgantown PRT has too-expensive track requirements that you can't conveniently mix with other forms of transport.
Re: (Score:2)
Cars can function if you have shitloads of road, there's no reason why car-sized vehicles can't work if you make the track a lot cheaper. The Vegas loop doesn't work because it uses shitty cars on tires on roads in tunnels. Morgantown PRT has too-expensive track requirements that you can't conveniently mix with other forms of transport.
Cars don't work because you need shitloads of road if you want them to work in the city. They don't scale. This same problem is why PRT and the loop fail. You need roughly human size vehicles for individual humans, or enough room for everyone. Car size vehicles are farm equipment for rural life.
Re: (Score:2)
Level 3 means that you need a driver ready to take over on short notice, but unlike level 2 they don't need to be constantly monitoring the situation. They can do something else, as long as they can stop doing it and come back to driving in a reasonable time, say 10 seconds, when the car deems it necessary.
For example the car could handle itself in low speed traffic queues, but once it gets back to the open road it can require the driver to take over.
I'm not sure that there is great demand for that. Fully s
Re: (Score:2)
They should make it easy for comma.ai to plug into their cars and pay comma.ai to support their cars. Sending liabiities to them.
Re: (Score:2)
Everyone on ArsTechnica told me 10 years ago we would all be driven by autonomous vehicles and human drivers would be obsolete?
Did they lie?
Not me, I was telling everyone that it wouldn't
happen for at least 20 years. I am impressed
at how far it's come, but I think it will still be
another decade before it really works.
Possibly even another 20 years.
As for lying, that would be Stellantis.
They are "shelving" the technology
--- by which they mean cancelling their
project and getting out of the business --
not because it WORKS, but because
THEY COULD NOT GET IT TO WORK.
"No demand for self-driving cars."
That excuse is beyond absurd.
"No demand for self-dr
Re: I can't believe it (Score:2)
Elon lied to you
Re: (Score:2)
That said, Tesla is making some progress. They have their small Robotaxi fleet in Austin, and are now increasing the number of vehicles. (From maybe 20 to 30). In Texas they do have a "sa
Re: (Score:2)
Did they lie?
They didn't know what they were talking about, they lied about their powers of prognostication, or both.
Self-driving cars may one day obsolete human drivers, I even believe that they potentially could in my lifetime. But I don't think it will happen soon, and I don't think it's a particularly worthwhile goal either. Instead we should eliminate most of the need to drive, and get down to lighter vehicles (think SxS or NEV) for most of the remaining (rural, recreational, emergency) use. We could delete a lot o
Good call (Score:3)
Another stupid thing is using the speed limiter, you can be driving 60mph on a main road and all of a sudden the car slams on the brakes reducing speed to 30mph because the camera has picked up a sign in a side street while driving past. Even when it is working properly it is very rough, unlike a human driver that sees a lower speed sign up ahead and eases off to smoothly slow down while approaching the lower speed section, the automatic system waits until it reaches the sign and then brakes hard increasing risk of being rear ended.
Seems more complex than necessary (Score:2)
...reducing speed to 30mph because the camera has picked up a sign in a side street while driving past...
Are you sure it reads road signs? That seems vastly more complicated and much less reliable than simply getting the speed limits using GPS coordinates and a map. Every car I've driven that has a speed limit display seems to use this approach and while it does mean that you have to keep the maps up to date I'd think that would still be much more reliable that reading road signs for reasons similar to what you mention. In fact, you will have to use GPS to know how to interpret the signs: Canadian speed limi
Re: (Score:3)
Re: (Score:2)
Forgive the pedantry, but GPS does not know speed limits, it just provides location - to varying degrees of accuracy.
Speed limits are part of the mapping data loaded into the SatNav/in-car computer/etc, and can be inaccurate, incomplete or just out-of-date - sometimes all three.
Re: (Score:2)
Re: (Score:2)
Are you sure it reads road signs? That seems vastly more complicated and much less reliable than simply getting the speed limits using GPS coordinates and a map.
I googled "stellantis level 2 driver assistance reads road signs -stla" (I added that last on there to avoid getting a shitload of stories about the cancellation this story is about) and the top result is about the Jeep Compass [stellantis.com] and how it can read road signs. Learn to internet, bro.
Every car I've driven
Why do you think the cars you've driven are relevant?
So if you have to have a GPS map to know how to read the signs why not just use it to get the limits too?
Because speed limits can be changed faster than the database gets updated.
Re: (Score:2)
Re: (Score:2)
GPS is sufficiently accurate to determine which country you're in, and so whether the signage is in mph or kph.
It is sometimes insufficiently accurate to determine whether you're on a 70mph road or the 20mph road than runs parallel to it, which can be mildly irritating when your SatNav routing keeps recalculating as it switches between the two, but is massively dangerous if it's connected to your brakes.
Map-based speed limits are also problematic if, for example, a municipality drops a limit from 30mph to 2
Re: (Score:2)
Re: (Score:2)
Yes, they do read traffic signs. Even temporary ones.
Still, stellantis car software is absolute crap. Never again.
Re: (Score:2)
If a human driver drove like that he would lose his license.
Yet these cars are allowed on public roads.
Re: (Score:2)
So you based your judgement on the fact that you have experience with a different car, without the feature in question, without the hardware in question, in a different model class, bought at a different time.
I'll run with this line of thinking: my dog is stupid, my dog is also a mammal and so are you therefore you are stupid?
No you're probably not stupid but you did make a very stupid post.
Re: (Score:2)
Re: (Score:2)
if Stellantis shows an inability to get the basic driver assistance features working well
But you're begging the question. Where have they shown this? You have provided no evidence that you have any knowledge or experience with the system under question. On the flip side several government safety agencies have. Their system is approved and certified as level 3 in multiple jurisdictions in the world. That is safety agencies have declared the system actually works.
Any experience you think you have is invalid, this has nothing to do with any system you have ever driven in any Mercedes car. Fundamen
Re: (Score:2)
There seems to be a long history of this kind of thing with Stellantis. I've never owned one, but I've rented them numerous times, and I find regular old cruse control doesn't work well either. It shifts up when it should shift down, and vice versa, causing major speed swings on either side of the target speed.
Re: (Score:2)
Or am I to gather that it has now been "innovated" into a spectacularly complex fuckaroo that doesn't work with a shit by comparison?
Re: (Score:2)
How is it even POSSIBLE to have cruise control not work properly?
By adding a proximity/range sensor to stop inattentive drivers rear-ending a slowing/stopped car ahead.
Mercedes added "RADAR-controlled cruise control" to some cars years ago, and it worked very well, because they did what it took - time/money/people - to get it right.
I suspect that other implementations suffer from cheapest-possible components and inept implementation, perhaps even using the parking sensors to double as long-distance range-finders.
Re: (Score:2)
Simple: by optimizing for cheap, rather than optimizing for quality. There are plenty of decades-old technologies that are still implemented cheaply and poorly, even though the problems have been solved long ago. Something as simple as a serving spoon for your kitchen. If you buy that spoon at the dollar store, the handle will probably fall off before long. There's no reason a spoon's handle *should* be at risk of falling off. But if you want quality, you have to pay for it. Some car makers don't want to do
Level 3 (Score:2)
When the SAE defined the levels for AVs the immediate feedback was that L3 in particular didn't seem like a great idea, as the manufacturer has the liability for accidents, yet the driver is supposed to take over when requested.
Re: (Score:3)
Re: (Score:2)
That's not what L3 and L4 means. You don't need to be better in emergencies, you need to be better in circumstances. E.g. highway driving. That's what this is. Effectively fully unattended driving on the highway with a handover to humans at the end of the highway trip. The system is designed to handle scenarios in that subset and hands back to the driver with significant advanced warning.
There's enough of an argument to say that if for example Tesla geofenced their FSD to just highways they'd probably get L
Re: (Score:2)
L3:"Level 3 vehicles have “environmental detection” capabilities and can make informed decisions for themselves, such as accelerating past a slow-moving vehicle. Butthey still require human override. The driver must remain alert and ready to take control if the system is unable to execute the task."
L4:The key difference between Level 3 and Level 4 automation is that Level 4 vehicles can intervene if things go wrong or there is a system failure. In this sen
Re: (Score:2)
Re: (Score:2)
Not really. There's a defined handover period. Unlike Tesla's bullshit they tried with the L2 (they claimed it wasn't Autopilot problems when Autopilot disengages 1 second before an accident) L3 certification requires defined hand-off times to drivers. In Mercedes certification the time was 10 seconds meaning the L3 system needs to handle all cases that a driver can't take control of within 10 seconds. This is also why the system is limited to specific scenarios where they know the system can handle, and yo