GM Is Doubling the Size of Its Super Cruise Network In the US, Canada (engadget.com) 68
General Motors' Super Cruise advanced driver assistance network will soon double in size to 400,000 miles across the U.S. and Canada by the end of the year. Engadget reports: The Super Cruise system -- and its successor, Ultra Cruise -- relies on a mix of high-fidelity LiDAR maps, GPS, and onboard visual and radar sensors to know where the vehicle is on the road. So far, those maps, which dictate where features like Hands-Free Driving can operate, have only included major, divided highways like interstates with the big median barriers. Smaller, undivided public highways -- aka State Routes -- were not included, in part because of the added ADAS challenges presented by oncoming traffic, until now.
"This expansion will enable Super Cruise to work on some additional divided highways, but the big news is this the bulk of the expansion will allow Super Cruise to operate on non-divided highways," David Craig, GM's Chief of Maps, said during Tuesday's call. "These non-divided highways are typically the state and federal highways... that connect the smaller cities and townships across the US and Canada." These will include Route 66, the Pacific Coast Highway (aka CA Route 1), the Overseas Highway (aka US Route 1) and the Trans-Canada Highway. "if you look at I-35 which is the interstate that runs North and South up the middle of the United States, and look to the West, you will see that the Super Cruise coverage currently is just the major interstates, which is fairly sparse," Craig continued. "But in the expansion, you can see that it's just a spiderweb of roads covering the entire area. All the little townships are going to be connected now." The company said that every new Super Cruise-enabled GM vehicle will be equipped with the full 400,000-mile capabilities, as will 2021 and 2022 GM vehicles outfitted with the VIP (Vehicle Intelligence Platform) architecture. Vehicles with Super Cruise but without VIP will receive a smaller update.
"This expansion will enable Super Cruise to work on some additional divided highways, but the big news is this the bulk of the expansion will allow Super Cruise to operate on non-divided highways," David Craig, GM's Chief of Maps, said during Tuesday's call. "These non-divided highways are typically the state and federal highways... that connect the smaller cities and townships across the US and Canada." These will include Route 66, the Pacific Coast Highway (aka CA Route 1), the Overseas Highway (aka US Route 1) and the Trans-Canada Highway. "if you look at I-35 which is the interstate that runs North and South up the middle of the United States, and look to the West, you will see that the Super Cruise coverage currently is just the major interstates, which is fairly sparse," Craig continued. "But in the expansion, you can see that it's just a spiderweb of roads covering the entire area. All the little townships are going to be connected now." The company said that every new Super Cruise-enabled GM vehicle will be equipped with the full 400,000-mile capabilities, as will 2021 and 2022 GM vehicles outfitted with the VIP (Vehicle Intelligence Platform) architecture. Vehicles with Super Cruise but without VIP will receive a smaller update.
Robotaxis (Score:1, Interesting)
Interesting approach (Score:5, Interesting)
I can see a lot of problems with a self-driving car relying on external data.
Data can be incomplete, out-dated or unavailable because of connection problems.
And there is the obvious problem of vehicles with this system losing functionality when GM stops supporting it.
Tesla's AutoPilot is self-sufficient AFAIK but even that requires a paid subscription.
Re:Interesting approach (Score:5, Insightful)
I can see a lot of problems with a self-driving car relying on external data. Data can be incomplete, out-dated or unavailable because of connection problems.
I'm sure the engineers working on this have never even considered this, you should forward your concerns
Tesla gets most of the pub, but the supercruise solution is a conservative solution, and its gotten favorable assessments
Re: (Score:2)
I can see a lot of problems with a self-driving car relying on external data. Data can be incomplete, out-dated or unavailable because of connection problems.
I'm sure the engineers working on this have never even considered this, you should forward your concerns
Yes, the engineers probably considered this. They also probably considered the system being hacked and cars provided with false data that might lead to loss of life.
So of course because they've thought of this and taken steps to prevent it, we should all just rest easy and not worry; the tech sector has an exemplary record of preventing malicious hacks, and would never use weak passwords or flawed libraries for critical systems. And of course the people managing those systems will be immune to social engine
Re: (Score:2)
Re:Interesting approach (Score:4, Interesting)
Data can be incomplete, out-dated or unavailable because of connection problems.
If the data is incomplete, then that section of road will not be added to the supported network.
If the data is outdated, the next car will notice the discrepancy and report it. There is plenty of redundancy. So if an annotated road marker has been removed, there are still plenty of other annotated features.
The connection is through the cellular network. When you set a route, the data for that route is downloaded or updated. On a long trip, you will have all the necessary data cached hours before you drive a section of road.
Automated driving systems have navigated billions of miles on public roads. If any of these issues were show stoppers, someone would have noticed.
Re:Interesting approach (Score:5, Interesting)
Re: (Score:3)
It is a definite problem to tackle but I wonder if the best way to tackle it is the idea that you no longer "own" the car but it's more of a universal robo-taxi service.
We already have long track records for insurance and liability with plane, train, bus, taxi and other mass transit since there is a clear delineation of "vehicle operator" and "passenger". Most people are comfortable with all those even though they all share the same lack of self control. If the bus is driven by a human or a robot doesn't
dui liability in self driving cars? as in control (Score:2)
dui liability in self driving cars? as in control can = just haveing an app on your phone and in control can = DUI
Re: (Score:2)
how do you mean though, if i am drunk and hail a taxi today I am not responsible for if it crashes.
i summon my own car today even with all the self driving tech I am still the owner operator and responsible, even if the self driving system was in control of the car i will still get dinged with the DUI. pretty cut and dry still, i think this is only an issue when we get to the point where self driving is so markedly safer and better than humans that we shouldn't ethically let humans drive anymore.
Re: (Score:2)
The car is logging, so it will know if you did something to control it other than asking for it to take you to a destination. But it will probably be illegal to operate anything less than a level 5 vehicle while under the influence, because every other class of AV may fall back to asking you to drive for it.
Re: (Score:3)
I disagree that fewer people being harmed is an ethical problem.
Automated driving systems have killed several people, but fewer per passenger-mile than human drivers, and the public accepts that.
Perhaps having someone to blame isn't as important as you think it is.
Re: (Score:2)
My point is that I don't actually think people are comfortable with that. When that car ran over the lady crossing the road, it created a lot of concern about this stuff. If it starts happening more, people might demand that we ban them. At the very least, every incident will require an in-depth investigation by a trusted 3rd party, and we're lacking in trusted 3rd parties these days. Just look at how much the 737 max controversy negatively impacted our trust in the FAA.
Re: (Score:2)
Even though human drivers might be more dangerous
Something that doesn't appear to be in evidence outside of studies designed to give machines a mulligan when they fuck up
Re: (Score:2)
Re: (Score:2)
I agree that the number of fatalities will trend down and that's great, but there's still an ethical problem when we take the people responsible and move them so far away from the situation.
Wait, so your problem is that deaths and injuries reduce, but the legal system isn't able to pointlessly milk millions from those who can't pay and some laywers will go hungry?
Why the witchhunt for someone who is liable? That's not an ethical dilemma, that's more like the kind of thing you hear at an Evil Law convention.
Re: Interesting approach (Score:2)
If a self-driving car makes a mistake and kills someone, some big nebulous company has to pay out an insurance claim, which will be capped due to lobbying
If a human driver kills someone, the insurance claim is limited by the amount of coverage the driver (killer) paid for, why should it be any different if a "big nebulous company" is deemed responsible?
If a person's life is worth $1.5M when an individual is found at fault in an auto accident, why should the same person's life be worth more if a company is found at fault?
Is the point of an insurance settlement to 'punish' the responsible party or to compensate the victim?
Re: (Score:2)
Sure, it's not likely to lead to crashes.
But it does sound like a crappy UX, imagine Kitt telling Michael Knight to take back the wheel at random moments.
Re: (Score:2)
Danger! Danger Will Robinson!!
LOL...ok, seriously, no thank you.
I don't want my car hooked up to any cellular network, phoning home for any reason.
I've been driving more than a few decades, I'll be happy to continue doing so manually.
Re: Interesting approach (Score:2)
The connection is through the cellular network
So, Super Cruise will still be confined to major Interstate routes. Not due to mapping issues, but the lack of decent 5G connectivity annywhere not along one.
Don't get me wrong. I'm perfectly happy with you city folks being stopped at the boundaries of your invisible dog fence.
Re: (Score:2)
So, Super Cruise will still be confined to major Interstate routes. Not due to mapping issues, but the lack of decent 5G connectivity annywhere not along one.
even if that is the case, it will not always be the case.
even at that have a attainable, performant self driving system that operates on the major interstate highways, well, that's still something
Re: (Score:2)
So, Super Cruise will still be confined to major Interstate routes. Not due to mapping issues, but the lack of decent 5G connectivity annywhere not along one.
No. When you enter a route, the car caches the data for the entire route. You can drive for hours, perhaps days, outside cellular coverage.
There are very few places in America where you can drive for more than an hour without passing a cell tower.
Re:Interesting approach (Score:5, Interesting)
Tesla's Autopilot depends on the internet for routing and map data. The vehicle handles only the moment to moment part of the driving task autonomously, like all of the self-driving vehicles. They all do lane keeping and collision avoidance with their onboard hardware. They all continually phone home with their position, and get updates on route conditions and the like in realtime. The Tesla probably stores the entire route while you're navigating it like Google Maps (their routing source) does in the app.
People are acting like the GM vehicles operate only from the maps, which is dumb right on its face. The vehicle has to know where the road is in order to gracefully follow the map, it doesn't just work by GPS coordinates. Super Cruise uses cameras, LIDAR, RADAR, and GPS. Because it doesn't depend on cameras to estimate depth, it's got a much easier time of obstacle avoidance than the Tesla — which can mistake the edge of a box trailer as the horizon because it can't tell how far away it is.
Tesla has also been using their fleet to do pre-mapping. The whole idea that they're not also using high-res road maps is goofy AF.
Re:Interesting approach (Score:5, Interesting)
Re:Interesting approach (Score:5, Insightful)
Yeah, I should have noted that myself. Right now they just have RADAR on the car. Whereas Tesla used to have RADAR, then dropped it, and didn't add LIDAR to make up for it. I think that's a completely dumbshit move, personally, and the Autopilot's tendency to drive into shit at top speed bears that out. Some kinds of accident just aren't acceptable, and stuff that you can solve by spending a little more on sensors fall into that category.
When I imagined self-driving cars, what I always imagined is that among other things they would make decisions to avoid (insofar as it is possible) the creation of sketchy situations. For example, if they can maintain a better sight picture of the road by being a second or two slower, they should do that. I do that kind of thing when I'm driving, but the vehicle could be doing a lot more of it, because it never gets fatigued.
Re:Interesting approach (Score:4, Informative)
Re: (Score:3)
The reason given for dropping radar was that attempting to integrate it with the visual data was actually creating more uncertainty. More data does not always equal more information.
Re: (Score:1)
Omg,jfc, get out of here with your facts and rationality!
This is Slashdot. Such behavior is unacceptable!
You are supposed to rant how awesome Elon is and oh wait, we hate him now... you are supposed to rant how horrible Elon is and how terrible Tesla vehicles are and going out of business for racism, ethics and financial chicanery.
So, being serious for a moment... one of the scariest car moments for me (not including real crashes) was being in the back seat when the Tesla sales gal demonstrated auto pilot.
Re: (Score:3)
If the data doesn't match what the car is seeing, it signals for the driver to take over. It's the same as when Tesla Autopilot gets confused, the driver must take over immediately.
Nissan use the same system and they store all the data in the car, i.e. it doesn't depend on a network connection so connection problems are not an issue. I would have thought that Super Cruise is the same.
The Tesla system seems to be less reliable. They are trying to make it work just with cameras too, no radar or lidar. The lac
Re: Interesting approach (Score:2)
I'm trying to imagine how radar will work in bumper-to-bumper traffic, with every car around you spitting out radar signals constantly - seems like getting a reliable reading might be hard...
Re: (Score:1)
or $299/year for map updates (Score:2)
or $299/year for map updates
Re: (Score:2)
Re: (Score:2)
Re: (Score:3)
I kinda wish they would actually merge the two approaches together. As they both offer particular advantages that the other doesn't have.
Tesla's Auto Pilot and poorly named "Full Self Driving" is much more advanced and can handle a lot more conditions. For those with FSD Beta it can actually get you from point A to point B often without any real issues. However, the Super Cruise Electronic Rails can offload a lot of the hard work of making sure you are in the correct lane and on the road, and put that ext
Re: (Score:2)
I can see a lot of problems with a self-driving car relying on external data. Data can be incomplete, out-dated or unavailable because of connection problems. And there is the obvious problem of vehicles with this system losing functionality when GM stops supporting it. Tesla's AutoPilot is self-sufficient AFAIK but even that requires a paid subscription.
Imagine unironically thinking Tesla's advanced cruise control is self sufficient or doesn't rely almost entirely on OpenStreetMap. But if you do, I recommend doing your homework and not trusting it more than you would a regular cruise control because it will not keep itself out of trouble more than the others, they're just more reckless in their implementation.
Poor approach (Score:2)
relies on a mix of high-fidelity LiDAR maps, GPS, and onboard visual and radar sensors to know where the vehicle is on the road.
This system seems like it tries identify where you are within the LiDAR maps and navigate virtually. This is a poor approach because means that the second there is road work done then your car may suddenly be unable to cope.
If they are using those LiDAR maps for anything then it should be to train a neural network to deal with any situation within it. You could then automate plausible road alterations to the data and ensure you neural network can deal with those too. This would result in a far more resil
GM seems to have gotten this right (Score:5, Informative)
What GM is doing here seems to be ideal. They've figured out on which road sections the system can work exceptionally well, they've given it a non-confusing name ("supercruise" is pretty clear what it does even to people who don't study aviation), and - most importantly - they have the right array of sensors to do this safely.
By having the vehicle know which sections of road it can navigate - and which it can't - GM can be sure that the system isn't engaged in places that aren't appropriate.
Finally an adaptive cruise-control system that safety-conscious drivers would be interested in leveraging.
Re: (Score:2)
The other huge advantage of Super Cruise is that it uses a camera to monitor the driver's attention. The camera can see through sunglasses and checks that the driver's eyes are open and on the road.
Tesla does have an internal camera on some models, but it's a normal visible light camera so they can't use eye tracking as it would be incompatible with sunglasses, or even normal glasses with glare. They claim to have some kind of "driver sleeping" detection, but I haven't seen any independent testing of if it
Re: (Score:2)
Cars with cameras recording my face and every move? Cars phoning home via cellular, constantly?
No thank you...if that's what it takes for self driving cars, I'll happily keep manually driving myself and keep that shit out of my vehicle.
Re: (Score:2)
I sometimes forget that you live in a country where you have no privacy rights, and corporations basically own you.
You can't have nice things because you have no laws protecting you from this kind of abuse.
Re: (Score:1)
This is true but starting to change. A few places have started to pass privacy laws philosophically based on GDPR with different implementation details.
The average person is slowly coming around on the privacy thing. I'm guessing in about 10-15 years it'll be an election issue, all politicians will get on board and we'll just fight over the details from there.
The problem with GDPR is that it is a good first effort at creating privacy protection but the details are too lacking. I have been directly and/or
Re: (Score:2)
Does the GDPR cover that situation too?
Re: (Score:2)
You mention corporations.
While I'm not thrilled about them having my info, I REALLY don't want this to be gathered and possibly accessed by the government on any level (Federal, State or local).
Corporations can take this info and fuck with you. Govt entities can just plain fuck you UP with info.
So,
Re: (Score:2)
I drive carefully and if I am in an accident, I'll take responsibility. It will mostly likely be another driver's fault though in which case the camera will be a benefit.
There is clearly a valid slippery slope argument here, but if I'm falling asleep on an interstate, I would *like* to have my car react and prevent me from kill
Re: GM seems to have gotten this right (Score:2)
Seems to me the issue is the driver that chooses to nap with a piece of fruit on the steering wheel, not the auto drive technology. When a driver shirks responsibility for supervising their car as it hurtles down the road, they are personally responsible for whatever mayhem ensues, not the manufacturer of the auto-drive feature.
A drive can't abdicate responsibility for their car by engaging a feature on the dashboard. If the car is fully-automated, no driver, that's a different kettle of fish.
Re: (Score:1)
But it probably would not react very well to unforseen circumstances.
Re: (Score:2)
Re: (Score:2)
Tesla would do well to include curated routes and learning from instances where a human had to take over control.
Re: (Score:2)
Re: (Score:2)
Ultra Cruise (Score:2)
GM says Super Cruise vehicles are not upgradable to Ultra Cruise, so why not wait it out until 2023 and get Ultra Cruise?
No thanks (Score:3)
I find it hard to believe that so many people truly do want to pay about 1/3 more for a new car and also give up their privacy just so they can have a bunch of over complex and inherently time-bounded tech in their cars, all to cover nothing more than slight inconveniences. OR could this really be just another government initiative to remove any/all control from individuals.
I guess one of them must be true because it's getting to be almost impossible to find, let alone buy any new car that does not have all this phone home stuff.
I'm wondering when prices for older cars will start to go up in value just because that's the only way to avoid all this technocrap. It can't be long now.
Re:No thanks (Score:4)
Well believe it, its definitely the minority that care about these type of far-removed privacy issues, especially when a majority of people have yet to be in a situation that would change their mind.
The care about privacy has already been answered by the smartphone market. Does it really matter if my car is tracking me when my phone is already tracking me?
Consumers have already decided, they like the convivences these things bring and with self driving it will eventually become an ethics question when the cars become so much safer driving themselves that letting people drive becomes unethical itself. The market greatly prefers new cars with all the whizbang tech in it.
Re: (Score:2)
Re: (Score:2)
Now I do see your point but on the other hand peoples apathy with these kind of privacy issues certainly is not helped by this chicken-little attitude from the skeptic crowd. I have been hearing for years and years about these nightmare scenarios of the government or corporations being able to enact these crazy controls over our lives through the tech but very few of it has come to pass.
not to say these entities are not acting like pieces of shit at times but we need to advocate to people where they are as
Re: (Score:2)
>> but when it doesnt happened and has never happened well why are they going to believe you about anything.
https://www.forbes.com/sites/t... [forbes.com]
Re: (Score:2)
Yeah I am not going to say it's good, it's in fact very bad, but show a normal person that story and they are going to see it never affected them in a material way and cops used this to catch a heroin dealer.
It's also a far far cry from some nameless faceless oligarchy turning off your car for thought crimes. We can slippery slope that for sure but that kinda argument just slides off people. It's not going to stop the vast majority from buying a new car with whizzbang features.
These arguents just need be
Re: (Score:2)
Couldn't agree more.
Re: (Score:2)
The care about privacy has already been answered by the smartphone market.
Don't have one of those either.
Yeah, they could ping my feature phone and find which cell tower I'm near. But I can turn it off. Or leave it on and at home to establish an alibi when I visit some of my more nefarious haunts.
Re: (Score:2)
Absolutely respect that but I also assume you are observant enough to realize you are by far the minority, those who actually "walk the walk" in terms of privacy protection.
Re: (Score:2)
The "phone home" feature is really just an extension of the tracking the auto-makers have been doing for many years before it. You'd be surprised what's being recorded in the "black box" in any vehicle made in recent times.
But yes, the communications link means they can track your whereabouts as well as recording details about how hard you tend to brake or take turns, or accelerate, or ??
I think people have, by and large, become desensitized to all of it after realizing your cellphone itself tracks your loc
Slight inconvenience? (Score:1)
all to cover nothing more than slight inconveniences
Slight what? Driving is the single most mind-numbingly boring activity that most of us wish we could do with out. It is a drain on our lives and a waste of potential that we could waste in far more entertaining ways.
Forget 1/3rd of the cost of a car. I'd happily pay 3x the cost of a car if it meant I didn't have to drive it myself. As for privacy... that ship sailed long ago. It matters to Slashdotters and few people beyond that. Most commoners will happily give out all manner of personal information for a
Won't make it (Score:2, Informative)
The problem is they are GM vehicles, they won't make even a quarter of that distance because their so cheaply made these days.
Re: (Score:3)
There has "never" been a time when GM vehicles weren't cheaply made, only a time when they had really good engine and transmission designs and the vehicles were simple enough that the rest of the cheap stuff wasn't likely to cause a breakdown, only irritations like poor paint quality.
Insanity (Score:2)
Experimental evidence of the reliability of these systems needs to be submitted for peer review and published -- if it passes muster -- in the open literature. If not, cruisin' for a bruisin'.