Motor Trend is running a piece on the systems in the recently released Mercedes-Benz "Drive Pilot 95", https://www.motortrend.com/reviews/mercedes-benz-drive-pilot-95-first-drive-review
Here are a few of the details I found interesting:
By the end of this year, pending final certification from the authorities, German customers will be able to order an upgraded version called Drive Pilot 95, which, under certain operating conditions, will allow their S-Class and EQS models to self-drive for an indefinite period in the right lane of autobahns at speeds of up to 95 km/h (59 mph).[The earlier version from 2022 only worked in congested traffic up to 65 kph (40mph)]
[...]
Why has it taken so long to implement a software tweak? Well, both the German legislators and Mercedes-Benz, which assumes legal responsibility for the functioning of its vehicles while they are operating in Level 3 autonomous drive mode, are cautious. The Silicon Valley 'move fast and break things' approach doesn't work for them.
[...]
In addition to the parking sensors in the front and rear bumpers and the 360 degree cameras in the rear view mirrors that are fitted to many Mercedes-Benz models, Drive Pilot equipped cars have multi-mode radars at each corner, a front-facing long-range radar and a lidar unit behind the grille, a stereo camera at the top of the windshield, a regular camera facing rearward through the backlight, and a moisture sensor in the front wheel well.
The rear-facing camera is used to detect the flashing lights of emergency vehicles approaching from behind, though the 'Hey Mercedes' voice activation microphone in the cabin will pick up the sound of the sirens even if the vehicle cannot be seen. The moisture detector, which measures the sound level of the spray from the tire on wet roads, is used to determine whether rain and spray could interfere with the camera, radar and the lidar systems.
[...]
In simple terms, the key difference between the original Drive Pilot system and Drive Pilot 95 is the latter will now operate autonomously at Level 3 for an indefinite period if the Mercedes-Benz is in the right lane of the autobahn and is following traffic traveling at no more than 95km/h. Without that traffic, which can be up to 1000 feet ahead, the system will not activate.
This is where the trucks come in: The traffic on German autobahns that most consistently conforms to that pattern are the swarms of semis that are constantly crisscrossing the country. "The trucks are generally limited to 80km/h (50mph)," says Drive Pilot test engineer Jochen Haab," but they usually travel at about 90km/h, and up to 95km/h on downhills."
Drive Pilot 95 could operate without having to follow traffic, Haab says, but making that part of its operational design domain provides an additional safety redundancy: If there is traffic ahead, and it is moving, the car knows for certain the road ahead is clear without needing to process more data to double check.
A highly precise positioning antenna mounted in the roof enables the car to know, to within a fraction of an inch, exactly where it is in terms of its absolute position, its relative position, and its position correlated to carefully measured landmarks on an HD map built from data collected Mercedes-Benz engineers who drove every single mile of Germany's 8,196-mile autobahn network in both directions and in every lane.
Still not a fan of Level 3 which requires the driver to be ready to accept a handoff--but this system gives the human 10 seconds to take control.
Many other interesting details in the link.
This seems to be the definitive answer to one of the original questions about self-driving:
"Mercedes-Benz, which assumes legal responsibility for the functioning of its vehicles while they are operating in Level 3 autonomous drive mode..."
(Score: 3, Insightful) by Runaway1956 on Sunday September 29, @01:51PM (11 children)
That's the problem with Musk's ideas. He relies on a small suite of similar sensors. Spreading the computer's sensors across multiple spectra is guaranteed to make the vehicle safer. I'm sure they will still screw things up from time to time, but that redundancy means it will screw up less often.
Are all those sensors costly? Well, sure, but how costly is a serious accident? Why on earth do we want to trust our lives to a vehicle with fewer senses than we humans have? Give the car super powers. Multiple super powers. Make them as infallible as possible. And, things will still go wrong sometimes.
That is exactly the way it should be.
“I have become friends with many school shooters” - Tampon Tim Walz
(Score: 5, Touché) by gnuman on Sunday September 29, @02:08PM
Worked well for Boeing! /sarcasm
(Score: 5, Insightful) by canopic jug on Sunday September 29, @04:08PM (3 children)
Why on earth do we want to trust our lives to a vehicle with fewer senses than we humans have?
Furthermore, level 3 still requires the human driver to be ready to take over if the system requests or if there's a failure [capitalone.com]. In other words, you are on the hook for incidents despite not actually being in control. That's a big NOPE from me. However, those level 3 cars will still sell millions of units because "there's one born every minute".
Money is not free speech. Elections should not be auctions.
(Score: 1, Disagree) by Anonymous Coward on Monday September 30, @02:12AM (1 child)
Ditto, this has also been my objection to Level 3 -- the instant-handoff to the person, the moment the system can't deal with the situation. However, tfa includes this paragraph, which makes the handoff with this Mercedes system sound much better than I anticipated:
10 seconds seems like it's plenty for me to re-gain situational awareness and pick up the driving task.
(Score: 2) by Deep Blue on Monday September 30, @09:12AM
IF it has 10 seconds to give you. Once it hits black ice in a corner and can't handle it, then it's more like 0,10 seconds.
No thanks. Level 3 is not enough. I want it to handle all driving if it has any of that capability, then i'll have some time to sleep.
(Score: 2) by sjames on Wednesday October 02, @03:49AM
I vaguely remember an old silent short where a model T is running out of control and the steering wheel comes loose. The driver hands it to the passenger who desperately attempts to air steer the car before it crashes.
I always picture that when I hear about the automation handing control back to the driver.
(Score: 3, Interesting) by aafcac on Sunday September 29, @06:30PM (1 child)
Part of the problem is American regulations are insufficiently concerned with safety. Part of the problem is that idiots are willing to pay for the cars that are known to not be ready and then drive them like they're full self-driving cars. Or, I guess ride in them like they're self-driving. And part of it is that Musk isn't going to go to prison for his death machines.
Personally, I've had the opportunity of driving around in a 10 year old Ford and honestly, partial systems can still make a huge difference. Being able to set my cruise control to what people are driving and then just hover my foot over the brake in case I need to brake makes things so much easier. I can concentrate on the road rather than maintaining a safe distance as the system generally handles that for me. But, the parallel parking system is great, not only does it handle that on it's own, but it identifies the spot, so I can still manually parallel park, but I know that there's probably going to be enough space when I try.
(Score: 1) by Runaway1956 on Sunday September 29, @09:25PM
We should be grateful for the idiots, actually. Let them find out all the ways that self driving cars can go wrong, so that in 20 years, the rest of us can have more reliable cars.
“I have become friends with many school shooters” - Tampon Tim Walz
(Score: 1) by pTamok on Sunday September 29, @09:38PM (3 children)
The target for the self-driving car is provably to screw things up less often than humans, or perhaps, more accurately, have the economic and health costs of accidents less than those of human drivers. It's not 'zero accidents', but better outcomes than human-driven cars.
Considering my current car's ability to both ignore speed-limit signs (legal, but of a style not in the recognition database), and misread other signs (not speed-limit signs, but misrecognised as such signs, sometimes dangerously so) as being valid speed limit signs, we have a long way to go.
(Score: 2, Insightful) by whibla on Monday September 30, @07:16AM (2 children)
I have long thought that, if we actually want to have autonomous cars, we should be seriously thinking about modifying the built up environment through which they drive - and road signs would be an essential element of that. Why rely on a car's camera to read the speed limit from the sign, when we know computer vision is ... temperamental? Wouldn't it make more sense for the sign to tell any passing car "the speed limit is ..."?
(Cue arguments that they'd be too easy to spoof / override / etc. Before you do that however, ask yourself: would it be easier than walking up to a current speed limit sign and gluing a different number over it?)
(Score: 2) by Deep Blue on Monday September 30, @09:36AM
Exactly. Cameras just are't trustworthy enough. The infra needs to be changed to make this autonomous driving actually work. The speed data can be encrypted, so it'll be less difficult to change them, and you can add a feedback system where a change in a familiar speed limit would make the car signal that back to the infra for checking.
But obviously the cars need to be able to handle situations which are sudden, like morons running or biking in front of the car suddenly, potholes and ice and water on the road. Obviously the cars can then communicate the road conditions to other cars, which will lessen the probability of a serious accident.
(Score: 1) by pTamok on Tuesday October 01, @09:48AM
I've pondered this briefly, and have thought that road signs should have associated balises [wikipedia.org], which are transponders used on railways, but the engineering is non-trivial.
As a different solution, how about QR-codes on road-signs? If computer vision has such difficulty with signs made for humans; how about signs made specifically for machines? More information could be incorporated, such as speed limits that are different for different times of day or days of the week in a fixed pattern (ie. suitable for permanent display - I don't expect all roadsigns to be e-ink or LED) or for different types of vehicles (the laws are there now, but drivers of vehicles with different (usually lower) speed limits are meant to know the actual limit in force), and other restrictions/modifications. It would also stop car-vision systems confusing the small pictures on some vehicles that display the maximum speed the vehicle can be driven (or towed) with actual speed-limit signs. QR-codes are reasonably robust to graffiti, dirt, and partially obscuring foliage. The QR code could incorporate the road the sign is on, so that when there are parallel roads with different speed limits, the in-car navigator can ignore signs posted for roads that the vehicle is not travelling on. QR codes are easily checked with any smartphone, too. If there is a difference in the information between the human-readable roadsign and the QR-code, then legally, the human-readable one would apply.
I'd also like my car to show me a picture of the last sign it 'recognised', so when it confuses a 'lorries not allowed to overtake' sign with a 60 or 80 km/h speed limit sign (which it does, routinely, when the signs are the LED type), I can double check, and override the spurious warning, rather than having a spurious speed-limit sign flashing in my peripheral vision until it 'recognises' the next sign - which can be a loooong time on a motorway.
(Score: 0) by Anonymous Coward on Sunday September 29, @04:04PM
https://www.motortrend.com/features/mercedes-benz-chairman-ola-kallenius-interview-ice-ev-offense [motortrend.com]
Interview is conducted in a big Merc, with DrivePilot active, Ghostbusters on the center screen, and popcorn.
(Score: 5, Funny) by Rosco P. Coltrane on Sunday September 29, @07:19PM
I'm old enough to have learned instinctively to distrust and stay clear away from any software that ends with 95 because it's totally insecure and unstable. It's a sort of long-term psychological scar thing.
(Score: 2) by bart on Sunday September 29, @08:24PM (7 children)
Everyone is going at least 130 or often a lot faster, except the trucks. At 95 you're definitely not keeping up with traffic, but causing issues behind you.
(Score: 2, Insightful) by pTamok on Sunday September 29, @09:28PM (6 children)
Nope. The busy parts are extremely congested and speeds are low. 95 kph is more than adequate for nose-to-tail traffic jams.
As as the article points out, this works only in the lane furthest from the central reservation - it's designed to travel with the heavy goods vehicles (HGVs) , and needs a vehicle less than 300 metres ahead to activate. I'm not sure what the HGV drivers will make of it.
It is not designed for people belting down the autobahn as fast as possible.
(Score: 3, Interesting) by aafcac on Sunday September 29, @10:46PM (5 children)
Which is really the right call. They'll gather real world data at 95kmph and the cap can probably be raised with a firmware updated it that turns out to be more conservative than necessary. An increase of 10 or 20 mph in increments probably wouldn't radically increase the risk for the people in and around the car.
I think people often times forget that the worst bits about driving a car are mostly minor things like driving in stop and go traffic. The only thing that I wish my Ford Fusion could do is handle the accelerator at under 12mph as that's the bit that I personally find the most taxing. It is predictable enough that I know it's going to kick back to me, but I'm not gutsy enough to make it actually stop me completely as it kicks me back before that would happen.
(Score: 1) by pTamok on Sunday September 29, @11:58PM (4 children)
Oh yes. I really want an adjustable 'crawl' mode for parking in unreasonably tight spaces. I'd love a setting where I could make the maximum speed of the car 5 km/h with the actual speed the result of an exponential function of the accelerator pedal position. My current car has some kind of offset on the controls so that nothing happens when I touch the accelerator lightly at standstill, but when I depress it further it leaps into motion like a startled rabbit, so I have to brake, and there's a specific low speed where the car decides I want to stop immediately rather than crawling asymptotically more slowly towards a particular position. It's like a deliberately programmed backlash/hysteresis , and it's as irritating as a very irritating thing.
(Score: 1, Informative) by Anonymous Coward on Monday September 30, @02:56AM (1 child)
> ... when I depress it further it leaps into motion
This is called tip-in, at least in USA. It goes back many years, from what I've heard it was developed as a way to make a low powered car feel powerful...on a first test drive from the dealer's lot. You barely touch the throttle pedal and the car leaps ahead--to unsophisticated people this gives the impression of great power.
Once you (or some other sucker) buys that "powerful" car, you eventually learn that it's a fake, since pushing further doesn't get you much more.
Worst case I experienced was a VW Passat c.2000. The amount of tip-in changed depending on recent driving style. Example: I leave my quiet street and accelerate hard to mesh with high speed traffic on the big road. Then I do it again when getting on another big cross street. I travel fast for a mile or two until the first traffic jam. Now I want to inch along with the car in front of me in line. But I can't, because VW engineers have decided that I must be engaged in sporty driving and they have flipped me over to the throttle map/look-up-table (in the engine control software) where tip-in is extreme.
Many years ago a good and very sharp friend owned a few VW products. While he kept coming back for more, there were a number of "features" similar to this which he classified as "sadistic German engineering."
(Score: 1, Informative) by Anonymous Coward on Monday September 30, @02:59AM
Reply to self. The original tip-in was done mechanically, the cable from the throttle pedal wound around a cam on the carburetor throttle shaft. Changing the shape of the cam determined throttle opening angle vs pedal position.
(Score: 0) by Anonymous Coward on Monday September 30, @04:07AM (1 child)
What are you yammering on about. How does any of that apply to the post your responding to?
(Score: 2, Informative) by pTamok on Monday September 30, @10:08AM
The off-topic replies can sometimes be more interesting and informative than the article itself.
That's one of the things about this place that makes me keep coming back.
That, and that there are some knowledgeable and experienced people here willing to share.
There are also some that are not, but the signal is strong, and the signal to noise ratio sufficiently low for good information transfer.
(Score: 2) by jb on Tuesday October 01, @06:22AM
So presumably if the driver of the vehicle in front decides to drive off a cliff (or bridge, etc.), all the Mercedes in the queue behind it will follow suit?