Mercedes-Benz wins approval for Level 3 autonomous driving on the Autobahn
ttnews.com
ttnews.com
It's gotten so bad that I'm switching off the lane assistant when I'm approaching construction areas.
I've talked to a guy at the car dealership, and he said "my mother complains about the same thing, it's a known issue".
I really, really hope they've managed to fix that for the autonomous driving models, or at least auto-disengage before construction areas. (And if they fixed it, can I get a software update please?)
Sure, youtubers can video it for fun and drink the kool aid but actual drivers will just turn it off after a hairy situation. Well, I will anyway.
Maybe this is why Tesla are touting their ridiculous robot ... it's a potential pivot or alternative use of the (no doubt) impressive AI tech behind FSD which, they've now realised, will take much longer to be properly driver ready.
Maybe FSD's fate is similar to the space program. The eventual secondary technology it spawns is more useful than its original purpose. Because I won't be going near these autonomous driving solutions for many years.
Yeah, give me a Dalek-style home robot instead for now.
Which is to say... this is mostly a stunt. Teslas are literally driving people around now[1], and the rest of the industry feels they need to do something. Announcing a "SAE Level 3" product, no matter how constrained, at least gets them marketing hits like this that look like an advantage.
[1] No joke: mine takes my kids to and from school reliably. FSD beta isn't finished, but it's really, really good. There remain some path planning and confidence glitches that force me to disengage every dozen miles or so, but in 500 miles since I got it I've yet to see the car attempt anything genuinely unsafe. Mostly it just annoys other drivers by refusing to enter traffic.
If Tesla is also able to get approval for this (which might be hard if they push OTA updates that completely change the way the software works all the time and can't guarantee there won't be regressions) then they should by all means do so.
I mean, sure, it's an advantage for people who actually need to drive in exactly those conditions.
But it's not a engineering advantage of the system. Fast forward a year and look at where Daimler will be and where Tesla will be. Who's going to get to a L3 system on general roads first? Who's going to be running L4 driverless vehicles first?
Which is to repeat: it's just a stunt. It's a way to "sound like" they're "ahead" in this area of technology, when clearly they aren't. In fact this car is doing more or less exactly what Teslas were doing in the first version of Autopilot more than six years ago. And note that in all that time, those Teslas haven't been hitting anyone in slow traffic jams on well-maintained limited access highways either.
> those Teslas haven't been hitting anyone in slow traffic jams on well-maintained limited access highways either.
weird then that Tesla, despite showman Musk at the helm, hasn't been doing the simple stunt of putting their money where their mouth is, getting L3 permission for their system and actually taking the risk for something that is "never happening" instead of finding ways of blaming the driver every time. It would be an easy way of providing an additional actual useful capability to their users on top of what they already have.
No, it's not some massive technical leap, but "ok, we'll take the blame for our system failing" is still a big legal step.
Sounds pretty hand-wavy. Autonomy is about defining conditions in which a self driving system can safely operate, also known as Operational Design Domain (ODD).
Of course, ODD as a concept is foreign to Tesla FSD because it is "50% of the time, it works every time" i.e. you don't know when it works and when it doesn't. It's a YOLO driver assistance system with a misnomer. Not sure how many "real problems" that is solving and definitely won't be L4 anytime soon.
> I mean, sure, it's an advantage for people who actually need to drive in exactly those conditions.
> But it's not a engineering advantage of the system. Fast forward a year and look at where Daimler will be and where Tesla will be. Who's going to get to a L3 system on general roads first? Who's going to be running L4 driverless vehicles first?
> Which is to repeat: it's just a stunt. It's a way to "sound like" they're "ahead" in this area of technology, when clearly they aren't. In fact this car is doing more or less exactly what Teslas were doing in the first version of Autopilot more than six years ago. And note that in all that time, those Teslas haven't been hitting anyone in slow traffic jams on well-maintained limited access highways either.
If it's a race to see which company can develop fully self-driving cars that work in all conditions in the next 5 years than maybe Tesla could be "ahead" of Mercedes-Benz (although it seems to be way behind other companies like Waymo).
However, if we are actually decades away from fully self-driving cars that work in all conditions it is much more useful to have cars that can be trusted to operate without human supervision in limited but well defined conditions, because it is much more useful to be able to have a car that allows you to do something else other than focus on driving SOME of the time than a car that assists you more in some ways but requires supervision ALL of the time.
If Tesla's approach precludes them from making self driving technology work reliably under limited conditions like this, it doesn't matter whether their self driving technology works better (but not reliably) in other conditions, because in practice it won't be as useful as more limited but reliable self driving systems.
Some countries are already requiring the approach Mercedes-Benz is taking here for regulatory approval and based on the number of accidents Tesla vehicles have been involved in its possible that the US could adopt the same approach.
If this happens it will not be useful at all to have a more advanced but incomplete and unreliable self-driving technology in the short term (but perhaps it will be possible to keep developing it without shipping it in cars until it reaches a sufficient level of reliability).
Nope - it's a new system powered by LIDAR and cameras and only works on pre-mapped roads, comparable to what Tesla and Waymo are doing (but erring on the side of caution).
The 60kph limitation is temporary because that's what the regulator was comfortable with.
Lane assist is a different tech stack and continues to exist alongside this for regular driving.
https://www.daimler.com/innovation/case/autonomous/drive-pil...
Drive Pilot doesn't seem to be doing any of that. It's just a lane assist package: it will drive straight, in its marked lane, behind another vehicle, and that's all it will do. Maybe it will someday, sure. But it's not exhibiting these features anywhere, nor is Mercedes claiming that it has them. Am I missing something?
You can see it drive around cars stopped in the middle of the road, read street signs, switch lanes...
I don't think any other than waymo has solved this yet. My telsa is good but the amount of times I have to take control from it is way too high. Snow is another area where autonomous driving sucks at the moment, and as a Canadian, man that makes my Tesla's self driving not useful 5 months of the year.
Waymo seems to be way out front and everyone else seem to be fighting for second here. Though I'm very happy that we're atleast at this stage and not where we were 10 years ago.
I just view it as we now have working lane keeping and adaptive cruise control for most new cars these days and that's a solid win.
I'd say Cruise is comfortably second. They've made very good progress in SF and I believe they are doing driverless rides there for employees.
> Also curious to know about Waymo’s performance in the snow because I haven’t heard much about that.
Me too, i know almost nothing about their ability in snow.
When you experience it, it's clear they have a mature system that's learned from driving millions of real world miles (and billions in simulation). And if you follow their research/tech stack closely, there's no doubt they are years ahead of everyone else. I'm not sure how successful of a business they will be, but their tech is solid.
As far as driving in inclement weather, Waymo is pretty transparent that it's very much a work in progress. They claim the 5th gen Jaguar I-Pace vehicles (the 4th gen Pacificas in AZ don't do as well) will be able to handle rain/snow/fog better due to upgraded sensors, but it's one of the big challenges remaining for them.
Waymo (and others) are very clear that geofencing and prior knowledge of roads are necessary for safe operations. Their safety standards are extremely high because there’s no driver for fallback — the vehicle is responsible end-to-end. Geofencing allows them to validate extensively and that means when they say it works, it really does.
I think the mapping argument is pretty overblown at this point. Most of them have figured out how to do it efficiently and keep it up-to-date.
[1] https://blog.waymo.com/2020/09/the-waymo-driver-handbook-map...
On the other hand, I have no expectation that it is doing anything autonomous, and this is backed up by the way it gets upset if I take my hands of the steering wheel. And unlike most other cars I hear about, it isn’t content to simply touch the wheel, it wants a hand on either “ten” or “two” specifically.
For my n=1 needs, it’s fantastic: When I commute during rush hour, it handles stop-and-go traffic for me, and lets me sit back and enjoy an audiobook. It does that so well that I’ve completely lost the anxiety/aggression that formerly would have had me trying to figure out which lane is moving faster and so on. Now I just relax. Yes, I am ‘still “driving,” my hands stay on the wheel and my eyes stay on the road.
But I’m more relaxed, and that is more important to me right now than whether I’m relaxed while driving or relaxed while being driven. Yes, for things like construction (and a few other cases I’ve learned to handle) I disengage it or simply override its choices with the steering wheel, accelerator, or brake.
But it’s a big win over old-fashioned cruise control. I never felt like cruise control lowered the stress of highway driving the way Pilot Assist does.
Mind you, none of my experience changes the fact that “Pilot Assist" is nowhere close to autonomous driving. I wouldn’t even look at it and say it’s something you can iterate on until it’s autonomous. It is what it says on the tin: Something that makes my drive easier, but doesn’t replace me as a driver.
In the airbag era, shouldn’t this be nine and three?
Volvo steering wheels are set up for 9 and 3, but you are supposed to hook your thumbs (9 and 3), not rest your lateral palms (10 and 2).
See the diagram here:
https://www.popularmechanics.com/cars/a9343/forget-10-and-2-...
If I put my hands right on 9 and 3, those bumps don’t come into play at all, but I suppose I could hook my thumbs around the wheel while resting my fingers on the backs of the button-festooned spokes, which are at 9 and 3 themselves.
So yes, I can put my hands at 9 and 3 and ignore the bulges at 10 and 2. Or I can put them at 9:30 and 2:30 and rest my thumbs on the bulges. We’ll see how it goes unlearning 10 and 2.
The difficulty jump from "Mostly straight line highway driving" to "good fucking luck city streets" is extreme
Look at driving fail videos from Germany and the ones from Russia and the USA, you'll regularly see something they show as "huge mistake OMG" in Germany being unremarkable (if not business as usual) when compared to others.
TBF, now I'm not in Germany but when I was, driving on the autobahn usually involved religiously following the signs of speed limit and no speed limit(the limit-free coverage on the autobahn is patchy).
I wonder if the self driving systems asses the road type to determine the speed they should go. Is there a possibility for a autonomous car to speed up to 120kph around a school or in a residential area if the sign is missing or someone puts a bogus sign?
it also takes the curvature of the road into account and slows for the traffic, but there's always a small empty town with straight roads...
it also sometimes messes up the steering when going through a construction site with yellow lines, so I really hope that they improved this system heavily in the last year since I bought my car.
But even a lot of Germans don't know this, I'm afraid.
Yes, because this is not the case.
https://www.bussgeldkatalog.de/geschwindigkeitsbegrenzung-au...
This is, in fact, the legal standard in California! The actual speed limit is the minimum of (posted speed limit plus a substantial fudge factor) and (a subjectively safe speed for the road conditions). i.e. if the sign says 65mph, and the road is winding and a bit icy/rainy so that 45mph is the highest safe speed, you can be ticketed for speeding if you drive 60mph.
The legal standard, like many legal standards, is around reasonableness. Regardless of the posted limit, you may never driver faster than is safe for the current conditions.
And it is indeed borne on the judgement of the officer. Who else would ticket you?
Not entirely related but yeah, depends on road conditions and signage.
If you can legally rely on the speed limits in the maps, this should be much easier to do.
Recognizing speed limits seems to be a technical solution to an organizational problem.
Standardization is necessary for good automation.
E.g. there are narrow streets in citys with cars parking on both sides. The official speed limit is 50 km/h. If you drive at this speed you straight up murder any pedestrian coming out behind of a car. If you do this you'll be pleaded guilty, because your speed was not appropriate to the situation. You even have to learn this exact situation when you make your drivers license.
An other funny example. Where i lived was road with a bend, where an ordinary person should maybe drive 50 km/h. It was borderline questionable to drive through there with 70 km/h and a good car. Directly before the bend the speedlimit changed from 70 km/h to 100 km/h. There where a lot of accidents, because even germans don't understand the meaning of german speed limits.
I’ve seen a similar differences in another border regions, e.g. Switzerland-France or France-Germany. It almost seems like there are two schools of thoughts that different countries subscribe to.
Prague, on the other hand, is oversaturated with signs. I guess someone makes money on them, because some are clearly superfluous (Segway bans in distant neighbourhoods where no one ever drove a Segway; Segways were only a bane of the touristy center).
I didn't want to imply this. I was just talking about the task of finding the speed limit. Getting from there to a reasonable speed to drive at is a different issue and much more complicated, yes ;)
It's probably the easiest task for an autonomous system. So if your car can't do that, then probably it's safer to drive it manually.
"Das Ende einer streckenbezogenen Geschwindigkeitsbeschränkung oder eines Überholverbots ist nicht gekennzeichnet, wenn das Verbot nur für eine kurze Strecke gilt und auf einem Zusatzzeichen die Länge des Verbots angegeben ist. Es ist auch nicht gekennzeichnet, wenn das Verbotszeichen zusammen mit einem Gefahrzeichen angebracht ist und sich aus der Örtlichkeit zweifelsfrei ergibt, von wo an die angezeigte Gefahr nicht mehr besteht. Sonst ist es gekennzeichnet durch die Zeichen 278 bis 282."
The translation of this is: If a speed limit is combined with another sign indicating a reason for the speed limit (e.g. construction work, dangerous curve, train crossing...) the speed limit automatically ends without any additional sign.
So basically the AI has to understand why the speed limit is there and decide when it has passed the dangerous part of the road.
When starting to drive the car has to recognize if it is inside the city (50km/h or 30km/h limit depending on location), outside the city (100 km/h) or on the autobahn (no limit). This is simply impossible without having a map. You could save the state from the previous drive. But what happens if the car gets towed from the autobahn into a city. Will drive through the city without any speed limit?
I don't think this is simple at all.
It's a road without speed limits. Mistakes might be less likely - at least they are for humans, not sure about an overly cautious AI model - but the cost of a mistake is very high.
I think you missed the context there.
Some roads (at least here in Italy) also have a minimum speed requirement when there is no traffic, which goes higher the more you go to the left lanes, but that’s another story.
Anyway I think that reading speed limits from road signs is quite stupid and frankly useless, most of the time the speed you need to be driving is contextual and the same applies to a lot of other dynamics of driving.
Well done Mercedes, one step at a time raising the bar once you manage to do “simple” things right.
What reasons other than speed are there for keeping bicycles off fast roads? Motorcycle can be very nearly just as small and vulnerable as bicycles, but they can keep pace with cars and consequently they're allowed on every road.
Same reasons why you cannot drive small motorcycles on highways here in Italy, even if they can easily reach 110km/h they are not allowed on even second tier highways.
Maybe it's different in Italy, but in America the restrictions on bicycles really seem to be about top speed, not acceleration.
Overall the reasons why you cannot drive bicycles on highways are many, but when you put them all together it mostly fall under one term: safety.
What I wanted to say on my original comment is that trying to put an autonomous driving car in a urban situation is not about checking the speed limit from a road sign, looking at a ground stop or other road signals. If you want a safe autonomous car in an urban center it must be aware of the context much more than “this is a stop sign” or “speed limit is 50km/h”, that’s why I praise Mercedes for trying to solve the solvable problems first and then move to more complex ones.
My only source on this is a drivers training handbook.
That being said, driving laws, as far as I know, always say that you need to adapt your speed to the local conditions in order to drive safely, but you can never exceed the speed limit. Many (most?) jurisdictions have laws against dangerous and reckless driving and I expect it is possible to be found to be driving dangerously even when staying under the speed limit.
Where there is a minimum speed, which is also of course conditioned to current road conditions, it is also for safety. For instance, on a motorway where the traffic moves at 70mph it is dangerous for everyone if a vehicle decides to move at 30mph...
In the UK you’ll fail you driving test if you drive below the speed limit, eg 60mph in a 70mph zone.
Hard to find an authoritative source, but this seems to break it down pretty well: https://www.drivejohnsons.co.uk/learning-centre/faq/driving-...
'Appropriate speed' is definitely a potential minor (or major) error on your driving test (and you are marked as failing for either a single major or three or more minors in the same category).
Generally, it's up to the examiner's discretion whether speed is 'appropriate', but often lenience will not be more than 10% of the speed limit (unless there are other factors, such as weather, road surface, visibility, etc. which would make it unsafe to travel at the posted speed limit).
Sounds incredible until you witness it yourself.
I'd say that most cars drive at around 130-140 km/h in the center lane, and 170 km/h isn't unusual. This requires people to keep the left lane free, something I yet have to witness in North America.
But beyond that, the Autobahn is really boring to drive on. It's just like any other highway: almost straight, and completely isolated from the world around it.
If everyone is driving the 'recommended' speed of 130kmh on a German motorway then it is a severe risk if you drive slower. Road users need to be predictable. Trucks or motorhomes are slower, but you expect them to be slower. You see them from the distance and plan ahead. But a normal car that drives 60 when all others drive 130-140 is a massive danger as other drivers do not expect it to do so, so will lose precious seconds to react. You can get fined for driving unnecessarily slow.
In Germany you also cannot merge onto the motorway unless you have a certain speed already, and you cannot enter the motorway at all if your car can't do an appropriate minimum speed (or you have special transport papers and warning lights).
Do you happen to know the speed? I drive an Aygo (tiny, 3 cylinders) and I’d love to know how close I can get.
So every truck on the roads is a severe risk?
Different traffic laws. In Germany, there's only a lower limit on the Autobahn, and there's an advisory speed limit that in precedent cases you must be able to maintain if your vehicle and road (and weather) conditions permit but in practice is somewhat overrun, which is legal.
Most states in the US penalize overspeeding, but also penalizes underspeeding on state and interstate highways if car, road and weather conditions otherwise permit. While counterintuitive, in highways (especially interstate ones) there's an expectation of a "belt" of vehicles that maintain more-or-less the same speed: different speeds of vehicles should be on different lanes (the innermost being the fastest lane otherwise posted in signage).
What? Where is the legislation that says that?
That could be by design, or the factory fitted alloys having a different circumference to the factory steel wheels.
I peg my cruise control at 76.5 and overtake all day long because other road users are oblivious to their actual speed.
If you are doing 40mph but everyone else is doing 65-70 then you cause a hazard for people as they have to take evasive action to go around you, then others have to brake to avoid hitting the people going around you, then a ripple braking wave starts behind you, then a BMW driver rear-ends someone half a mile behind you and suddenly you've caused a multiple pile-up where someone died today, because you were going too slow for the conditions.
But I seriously doubt that driving below the speed limit has a measurable increase in risk.
I would bet the opposite: That driving below the speed limit decreases risk of serious accidents.
One reason for that is that rear end accidents where both cars are going in the same direction are typically much less serious than head on collisions. So even if slow drivers increase the risk of rear end accidents (which I doubt is significant), they reduce the impact of head on collisions much more leading to a net positive.
People should drive slower, it's always safer. (Apart from dumb maneuvers like brake checking...)
Some subsequent research affirms this curve, others call it into question. My own, informal read is that the stronger research generally affirms it.
Reducing the rate of harmless crashes is no good if the rate of severe crashes increases.
Also, all crashes are relevant. Fatalities can’t happen when there aren’t crashes.
The speeding crisis is mostly a moral panic.
"Would this person have died if the cars were driving slower?"
And simple physics tells us that almost always the answer is "No".
If you try to argue that driving faster is safer, you need to provide strong evidence. Statistical correlation over a time frame where cars and roads have been made much safer doesn't really convince me. Even a correlation between driver speed and collision risk is meaningless -- eg. people with poor vision generally drive slower, but are still more dangerous than drivers with good vision.
(I'm not saying that driving fast is unsafe in general. On a well maintained road with good visibility you can drive pretty fast with minimal risk. All I'm saying is that, all other things being equal, driving slower is almost always going to be safer)
It is not simply the case that anti-speed crusades result in safer roads. The USA’s 55 mph experiment a was a failure in that regard, and Vision Zero campaigns aren’t delivering on promises.
We really need to focus on crash elimination, which is not a hyper focus on speed. Crashes that don’t happen can’t injure or kill.
But I really doubt that slow drivers are inherently dangerous. Especially since a lot of road users are slow and can't drive fast (big trucks, tractors, mopeds, bicycles, RVs, cars with trailers, etc). So the focus must be on making sure that drivers are attentive and have good visibility, not telling slow drivers to step on the gas.
Sometimes, the right thing is for slow drivers to speed up and go with the flow.
If someone drives very slow, there's probably a reason why they are going so slow. It could be:
- they are distracted
- they are tired
- they are lost and don't know where to go
- they are intoxicated
- they don't see well at night or because of glare
- they are inexperienced
I don't think telling these people to go faster is going to make anyone safer.
An inability to read an argument without imputing your biases onto the person making it makes what you say a bit useless.
Many of our highways are posted at 55 mph or 65 mph. Car traffic is typically 70-80 mph on those highways.
Ignoring everything that is happening around you, even if you are under the speed limit, is an incredibly unsafe way to drive.
People are driving contextually, it's the limits that are not keeping up.
They're not blindly targeting the "speed limit" and throwing all caution to the wind.
(of course, they're ALSO often throwing all caution to the wind, but that's another story).
But once you’ve explicitly told the driver they don’t have to pay attention, you don’t have that excuse anymore.
No, only 6 years ago. In 2011 they hadn't even started selling the Model S.
For a given set of parameters (car/driver/environment) it gets to an accident every X miles causing Y$ in damage. This Y$ is paid out by pool of insurance money which a group of owners have already paid for.
Self Driving car is no different. In fact, the Car manufacturers will have more data to ascertain the 'blame' and fewer accidents every X miles and they can actually insure the drivers directly at a much lower rate. Car manufacturers will pay the much lower Y$ from this insurance pool
"More self driving capabilities means fewer crashes per mile which means they're cheaper to insure" hasn't really panned out with Teslas.
In 2018, Teslas had three times the fatality per mile rate of similar luxury cars: https://medium.com/@MidwesternHedgi/teslas-driver-fatality-r...
Teslas are also ludicrously more expensive to repair, and often parts availability is so bad that cars are "totaled" or the owner gets stuck in limbo - the insurance company says "well, we're happy to pay out, when you get the repair done..." and Tesla body shops and service centers have months long backups.
Result? Teslas, despite getting into fewer crashes per mile than the average, are much more expensive to insure than other cars.
The full self driving "beta" looks to be even worse, from sampling a couple of youtube videos of it "driving."
The average age of a car in the US is 12 years old, by the way, which significantly skews things away from all the cars that have ADAS features)
I'd imagine insurance companies have something similar internally when calculating premiums, as a $1,000 20yo Toyota involved in an accident could cause the same amount of damage as a $150,000 Mercedes with this tech to others.
That would be the part that would become cheaper as they will see over 1,000,000 miles the Toyota would have a higher rate of incidents than the Mercedes which has all these safety features (if it does what it says).
If your insurance also covers damages to your own vehicle, of course that's going to be more expensive, as repairs to the Mercedes will be much more expensive than the Toyota.
Insurance is great when you can't handle the cost of replacement.
That's why you don't buy insurance every time you buy a $100 electronic appliance. Now scale your money to a rich dude's money, you'll get the idea
I hear ya but at some point that's what administrative staff is there to handle.
Average age of a car in Britain is 8 years and it’s only recently got this high. I got given a 10 year old car a few years back and I paid to have it taken away by the time it was 12 years old because it was costing too much in repairs and I realised I would be spending the same money on car payments on a new car.
They'll only accept that if they can charge a subscription fee to car owners, and car owners are legally required to have an up to date subscription. There is no way manufacturers will take on the burden of insuring their cars voluntarily.
The human in the front seat of a L3 car isn't the driver. They're the scapegoat.
https://journals.sagepub.com/doi/abs/10.1177/154193121360142...
Do you have a link that leads to that actual study?
Found it, though unfortunately I couldn't find any non-paywall links.
https://journals.sagepub.com/doi/abs/10.1177/154193121360142...
Polar opposite of what Tesla is doing. If they can't match this, that's the whole German fsd market gone for them, possibly whole EU, as they will harmonize the Requirements.
I am really happy that they have the gusto to push things forward, if it weren't for them we wouldn’t have an EV race, a autonomy race, a rocket race.
This may be interpreted as the purchaser accepting known risks from a product, but no such consent exists from those around the car. Safety standards exist to protect people outside the car as well as inside, and may not be unilaterally waived on the part of those inside the car. Even assuming that the purchaser is fully informed about the risks when given consent, and is not misled by Tesla's erroneous use of "Full Autopilot" to describe something that requires continuous human oversight, not even that shred of consent exists from pedestrians struck/killed by self-driving cars (e.g. link below).
https://www.newsweek.com/tesla-lawsuit-model-x-autopilot-fat...
What's the risk profile of an empty autonomous car driving 20 mph on local roads to my doorstep, vs a fully autonomous vehicle?
Or, would you be willing to pay higher premiums or even extra tolls and taxes in order to drive yourself?
I think this will eventually be a question that fails over into economics while people are arguing about it. Insurers will be in a good place to influence the future of automated driving.
Humans as a group can't even not drive drunk, still. Individuals are not predictable either. Our taste for control will have to be worked around. (Hopefully in ways that allow us more time to...for example drive in VR, on a better course, with an even more stimulating feeling of presence, while we are driven around...)
Most people are willing to pay for things they enjoy.
It's also possible that a human driven car could use its autonomous driving system as a safety system to intervene if it anticipates a crash.
DUI test else engine does not start.
Speed limit as per road and weather conditions no car needs to do 70mph in the city or 150 mph anywhere.
Alertness check and monitoring , if not alert, car stops and that's it.
In cities where taxis are more ubiquitous, why isn’t level 5 simply replaces the taxi drivers with computers no? Then it’s not really a new paradigm, just that (a) traffic might flow better and (b) places where taxis aren’t cheap and plentiful (most of the USA, much of Western Europe) get to have an experience similar to a middle class lifestyle in a developing country.
Allowing high speed cars anywhere (even on the twisty "empty" roads) is the main reason the public realm is now objectively dangerous.
Driving should have been integrated in the culture as a solemn activity, only reserved to those with impeccable integrity and high moral standing, a true privilege in which the driver assumes responsibility for the lives of others, and accepts accountability commensurate with this duty.
Ending somebodies life, injuring others, causing damage, those are the inevitable by-products of careless driving, of driving for sport.
I'm exaggerating for emphasis. But driving as a hobby should be reserved to the racetrack. Just like shooting guns as a hobby is reserved to the gun range.
This obviously doesn't quite translate, but the amount of damage you can do with a car is massive, even at that speed. Also consider the weight, eg when rolling over a pedestrian.
I don't think you can set something loose in traffic that relies on everyone else correcting for its mistakes.
My point is, if I see a car coming, I'm not going to cross the street and let it hit me. That's the reason parents teach their kids to look both ways before crossing the street. Some idiot driver, human or otherwise, may not be controlling their vehicle in a predictable manner. At 20mph, as a pedestrian I can make a good decision about crossing the street.
I feel much safer about an empty driverless car at 20mph on local roads than one full of passengers at highway speeds.
(20 mph is entirely made up, it could be higher or lower in relationship to actual human reaction times)
Cars sit idle 80-90% of their useful life. As an auto manufacturer I don't need to achieve L{n} autonomous to make money, I just need wheels moving not parked.
Wheels moving at 20mph on local roads, in-between paid trips with human drivers, is a huge market and with a lower risk profile than L4 or L5 autonomous. The fat margins here are in making idle capital more productive.
The biggest problem with most self-driving cars is that they are easily fooled by various images, since they have extremely simplistic image classifiers to work with, rather than hugely specialized mamallian visual cortexes. Speed of processing is one aspect, but not nearly the only one. The Tesla that ran over that cyclist crossing that street at night wasn't missing reaction time, it simply didn't recognize the cyclist as an obstacle for whatever reason.
Now as far as I understand, lidar mostly fixes these false negatives (in good weather conditions), but at the cost of more false positives - and here speed could be a bigger help.
Even so, L4 and L5 would require planning skills and understanding of other traffic participants and road conditions that are still beyond us, at any speed. Note that the CEO of Waymo has plainly stated that L5 is not achievable with current sensor technologies.
Perhaps at some extreme low speeds (probably closer to human walking speed than 20mph) L2 could work as L3, but I think that's about it.
In my understanding, cars that drive more break faster, mileage being the dominant factor, not age.
So whether you have 10 cars serving 10 people, breaking after 30 years, or one car serving 10 people where you have to buy a new one every 3 years - seems about the same.
In the former model, car makers even have the opportunity to get people to buy new cars before they'd technically have to, which sounds like more money to me.
For society at large, the latter model seems better.
This can't possibly be a good thing. Nevermind if it works as advertised or not -- just from the very definition of what's claiming to do.
I know it certainly sounds like a good thing -- it will relieve the stress and fatigue of these slow-moving situations, leading to better reaction times, etc.
But just think about for a second. When you're at the wheel, the most you should be doing on the side is talking to people, and occasionally checking your GIS. You just should not be engaged in other attention-demanding activities like reading emails, surfing the internet, or watching movies. Period.
What's going to happen is: someone's leg's are gonna get crushed, because (foolishly or not) they walked between cars in a slow-moving traffic situation (like they always do). Because, you know, a human driver would have seen them. And then the same debates will start up all over again, completely missing the point. It's not the overall statistics of this happening, or that the driver didn't respond to the flashing red light that must have gone off in time (because they were zoned out, of course).
It's that they're pushing this inherently risky technology out there (which by definition cannot be fully validated until subjected to extensive real-life conditions), not to save us from drivers who are temporarily incapacitated at the wheel, not to warn us from approaching objects we might miss ... but to let us watch Netflix while driving.
The purpose of the system is basically to take over in bad traffic, doing everything for you, until things are good enough to drive again yourself.
This makes a certain amount of sense. Obviously the slower the car goes the more time it has to react to situations. It will also cause less damage if it makes a mistake.
Driving on the highway isn’t bad at all, especially with radar cruise control. I think I would rather have a system capable of handling bad traffic for me totally.
But I think driving 60 on the right lane is also not the scope of the system - it’s about handling traffic jams.
Does this phrasing seem unusual for others? It made me look to see if this was a satirical news site.
I’m not quite sure how it’s related though?
When you start reading your email it’s probably a bit unreasonable to call that anything other than your primary activity.
I think it was a big mistake that Tesla has given up on sensor fusion.
It may indeed have more sensors, but that doesn't mean it's more sophisticated.
Why would there be absolutely no way? People keep forgetting that Tesla is actually a small newcomer in the automotive sector.
They were ahead of the peak because established automotive companies slept on the market, but that does not mean these behemoths just gonna keep sleeping on it and leave it to Tesla without even trying to compete.
Expect to see more of this happening over the next decade, these companies might be slow to adapt, but once they get momentum behind their mass it will be very difficult for Tesla to keep up with them.
If I'm not mistaken the law requires a 10 second take over period before it disengages.
For the moment Tesla can't or won't do the same. Not in Germany, Not in America. Nowhere.
In the end we will see who is better (it might be Tesla, or another company, or all of them in close succession).
But right now Tesla is saying and implying a lot of things but still requires the driver to be in full control at any moment (wink-wink nudge-nudge), while Mercedes decided that there are situation where they got this.
But Tesla insurance, safety score and FSD beta will likely change that
I just think taking on liability is a big deal. Probably several lawyers at MB needed some fresh air when they were asked to formalize this policy.
Tesla obviously aims for an all or nothing approach. It's not before a solid level 4 that they want to take this leap.
I wonder whether the market (and regulators) will force their hand before they are good and ready.
Where as Daimler are simply limiting liability through engineering and constraints.
Unfortunately Tesla will be dead in that future.
It’s already happening in California. Young FAANG engineers respect Teslas less, compared to 5 years ago. And a higher % of techies are considering Audi, BMW, and Porsche EVs.
It seems to me that we are looking at a maturing market where several brands have released products which are quite good.
Judging by the numbers Tesla is doing extraordinarily well and is still growing strongly.
Unless one believed the Elon Musk groupy theory in which Tesla would be the only car manufacturer on the planet it's not really surprising at all.
Kind of stupid that this headline implies Mercedes is beating Tesla when Autopilot obviously supports speeds above 37mph. It would be like if a car manufacturer offered a package called "Full Self Driving" that wasn't actually level 5 self driving.
That’s what Mercedes system can do.
Tesla forces drivers to apply resistance to the steering wheel and the interior cabin camera (and cabin infrared camera starting 2022) will ensure the driver is looking up.
You mean like something called ‘autopilot’ when you still have to pay attention all the time?
Comparing autopilot to level 3 system is like saying that fly is more advanced organism than a dog, because it can fly.
Autopilot is a drivers aid system that will actively try to kill you as a part of a design (like any other level 2 systems). Level 3 system goal is not to kill you.