It's a bummer if true but after all that time and money the design either works or it doesn't. No sense throwing good money after bad.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Autonomous Car Progress
- Thread starter stopcrazypp
- Start date
-
- Tags
- Autonomous Vehicles
diplomat33
Average guy who loves autonomous vehicles
If the report is true that Ultra Cruise is cancelled, I am not surprised. Ultra Cruise was typical legacy automaker approach: announce a concept, do nothing and then cancel it when it is too expensive. And Ultra Cruise was always iffy IMO. GM was going to put a ton of expensive sensors on the already expensive, low volume, luxury, Celestiq, map every road and then somehow develop software to do 95% of driving? That's a terrible plan. And 3 years later, nothing.
I think the real question now is what will GM do next. They say they will refocus on Super Cruise but Super Cruise is highway L2. Then what?
Personally, I think it would make sense for GM to partner with Mobileye. Mobileye already has a wide portfolio of products from L2 to L4. And adding SuperVision would not be difficult. And they could deploy L2 hands-off that works everywhere, with a clear road map to add Chauffeur for eyes-off later.
I think the real question now is what will GM do next. They say they will refocus on Super Cruise but Super Cruise is highway L2. Then what?
Personally, I think it would make sense for GM to partner with Mobileye. Mobileye already has a wide portfolio of products from L2 to L4. And adding SuperVision would not be difficult. And they could deploy L2 hands-off that works everywhere, with a clear road map to add Chauffeur for eyes-off later.
diplomat33
Average guy who loves autonomous vehicles
IMO, Ultra Cruise was a terrible design to begin with. The plan was to basically put a ton of sensors on the already expensive, low volume Celestiq, map every road and then somehow develop the software to do 95% of driving but only L2. It is like they were trying to do everything and ended up doing nothing.
First, if you intend to do L2, you don't need a ton of sensors since the human will supervise. For L2, vision-only is good enough. So packing an expensive car with a ton of cameras, lidar and radar for only L2 makes no sense. And if you intend to do L4, then put all the sensors on the car, and design L4, don't restrict the system to L2.
Second, restricting your fancy L2 system to just a low volume, super expensive luxury car makes no sense either since it means very few customers will benefit from it. It is better to put the L2 on affordable, mass volume car that will benefit a lot more customers.
Third, how was GM developing this software that was supposedly going to cover 95% of driving? AFAIK, GM was doing the software in-house with some collaboration with Cruise. Maybe they were hoping the Cruise software would be good enough? But then when Cruise got shutdown, they probably realized the Cruise software would not work and their in-house software is probably far from ready.
So at the end of the day, they had an expensive car that nobody will buy, overpacked with sensors that L2 does not need, and software that is not ready. No wonder the project got cancelled.
Last edited:
Daniel in SD
(supervised)
Yep, never was going happen. I still don't believe FSD beta will ever have competition in the US.
Tech politics. "AI" is hot right now (translation: bubble) purely as a buzz-word, and getting $$$ investment or a stock boost simply involves a press release saying you are incorporating AI in your (insert random product here).
I am slowly approaching the time when I will foreseeably need to give up my license, due to deteriorating vision. Despite various oft-repeated claims on TMC, Tesla FSD is the only system that I can buy on a consumer car, that comes anywhere close to a possible solution.
Mobileye Chauffeur on e.g. Polestar 4 looks like 2026 at best - I'm very interested but not holding my breath. Various Chinese demos are interesting but don't seem to be purchasable in China, much less in USA anytime soon. GM Ultra Cruise had some brief exciting press, but appears to be back on the shelf. Most other systems are not even L4 aspirational, as in they might be L3 but only on the highway in limited conditions, and even that is mostly press-release fodder.
Anyone not already working on the engineering of an L4 car is not going to be in the market until probably 2028 - under a aggressive program.
Yes, Waymo works but no one ever talks about it being a model for personal cars; even if it came to my city, it's a ride service and it's unclear that would be dramatically preferable to Uber/Lyft/taxi.
The above is the current true state of affairs, and is at the forefoot of my thinking whenever I see "expert" analysis declaring Tesla to be trailing the AV industry.
For sure there are problems, of Tesla's own making, that may prove insurmountable for near-term L4 (IMO mostly around camera placement choices and lack of some relatively inexpensive cleaners or ancillary sensors). But they keep getting closer, and the fact that I can participate and evaluate and discuss makes it so much more compelling.
I was at CES last week and certainly AI is the word of the year. I would say that it's actually not just an investor-attracting buzzword boost, but more like a protective requirement to keep investors and customers from running away if you don't talk AI in your plans and promotions.
diplomat33
Average guy who loves autonomous vehicles
From old legacy OEMs, no. But there are plenty of FSD beta like systems out there that would be competition if/when they come to the US. For example, Xpeng's XNGP and of course Mobileye's SuperVision. Baidu is also testing a vision-only system with occupancy networks, much like FSD beta, and close to public beta. And I am probably leaving out a bunch. To say they will never come to the US is not realistic. It might be a couple years but they will eventually come to the US. For example, Mobileye SuperVision is slated to come to the US by 2026. Now, I am not saying that these other systems will be better than FSD beta, maybe they won't. But they will come to the US at some point and when they do, they will be similar and close enough to be competition.
Waymo continues to impress. Obeys hand signals from police
scottf200
Well-Known Member
Aside: general
I was forgetting the differences so I asked BARD and here is what it said.
Me Q: what is gm's ultra-cruise vs super cruise
BARD - A: Both Ultra Cruise and Super Cruise are advanced driver assistance systems (ADAS) developed by General Motors (GM), but they offer different levels of capability and target different scenarios. Here's a breakdown:
Super Cruise:
- Functionality: Offers hands-free driving on pre-mapped highways. It controls acceleration, braking, and steering within its operational boundaries.
- Technology: Uses LiDAR, radar, and high-precision GPS for lane centering and obstacle detection.
- Availability: Currently available on several Cadillac models like CT4, CT5, Escalade, and XT6.
- Limitations: Works only on mapped highways identified by specific lane markings and infrastructure. Driver still needs to pay attention and be ready to take over if needed.
- Functionality: Aims for hands-free driving on more than 95% of public roads (highways, city streets, rural roads). It handles stop-and-go traffic, lane changes, and even turning at intersections.
- Technology: Builds on Super Cruise with additional sensors like a forward-facing camera and expanded data processing capabilities.
- Availability: Not yet released. Scheduled for launch on the 2024 Cadillac Celestiq, with rollout to other models planned progressively.
- Advantages: Offers significantly wider operational range compared to Super Cruise, potentially enabling more practical day-to-day hands-free driving.
- Coverage: Super Cruise is limited to highways, while Ultra Cruise aims for most roads.
- Features: Ultra Cruise offers more advanced capabilities like handling intersections and stop-and-go traffic.
- Availability: Super Cruise is currently available, while Ultra Cruise is just beginning its rollout.
- Super Cruise is a good option for highway driving and offers a taste of hands-free technology.
- Ultra Cruise, once fully launched, has the potential to be a more comprehensive and practical hands-free driving system for everyday use.
Remember, neither system is considered fully autonomous driving. Drivers are still responsible for paying attention and being ready to take over if necessary.
diplomat33
Average guy who loves autonomous vehicles
I am sorry to hear about your vision. It sounds to me like you are kind of betting all on Tesla because you think they are the closest.
Yes, I know all this. Both are L2 systems. Ultra Cruise was supposed to be GM's "FSD Beta", a L2 system that works on city streets, and not be limited to highways like Super Cruise. It looks like that will not happen now.
My questions are: what will GM now do after they cancel Ultra Cruise? GM was hoping Ultra Cruise would be their L2 hands-free system that works everywhere. So without Ultra Cruise, GM has no L2 beyond highways. Will GM try to expand Super Cruise to beyond highways? Will GM just be content with L2 highway? Will GM try to make Super Cruise better at highways, maybe get Super Cruise to "eyes off" on highways? Will GM look to another supplier for L2 beyond highways like Mobileye?
That video doesn't show any understanding of hand signals. It's coincidental, just like various Tesla videos on the same topic.
Ignore the arms. There are other indications Waymo is using to go (pedestrian has moved out of the way, and the intersection cleared). If you watch very carefully, Waymo actually starts to move before the hand signal is made to go through the intersection.
High-confidence detection of a hand signal (and not a random swat at a fly) would require a couple full arm swings without any other indicator.
Waymo did the right thing at the right time, just not for the reason stated. If the officer ordered Waymo to turn right from the same standing location I'm 100% certain it would not have made a right turn in that specific situation.
This doesn't say Waymo doesn't understand at lesat some hand signals in other situations.
Why would GM Ultra Cruise be permitted to operate hands-free while Tesla's FSD is not? Is Tesla's hands-on requirement a holdover from earlier times when people were doing dumb things with the system? If GM owners do stupid things, will Ultra Cruise inherit the same requirement?
Rain and fog can be severe enough that human drivers must choose between "minimal risk condition" of parking off of the roadway, or continuing to drive with extraordinary risk of collision. IMO, certifying that your autonomous vehicle can drive in rain or fog does not include situations that would be impossible for most, or all, human drivers.
GSP
scottf200
Well-Known Member
It was general info for others as I indicated.
Good questions about what GM will do now. Rumor is they are switching to Mobileye <--- OK I just made up that rumor.
Was GM planning on getting tech from Cruise? Yikes.
diplomat33
Average guy who loves autonomous vehicles
Hands-free or not is really left to the manufacturer with the caveat that if they screw up, regulators like NHTSA can come try to come after them. So if Tesla felt like FSD beta were reliable enough to be hands-free but with robust camera monitoring, they could do that. Of course, if FSD beta got into an accident regulators could say that Tesla needs to strengthen the camera monitoring or go back to hands-on. The latest recall was basically NHTSA telling Tesla that they need to increase the nag because they feel drivers are not paying enough attention. So, GM was free to make Ultra Cruise hands-free but with a strict eyes-on requirement if they wanted to. But the same would apply to them as well. If Ultra Cruise got into accidents, NHTSA would force GM to do a "recall" and increase nags or impose hands-on.
Absolutely correct.
Daniel in SD
(supervised)
willow_hiller
Well-Known Member
Correct, but somewhat beside the point. The last example we had of Waymo entering the MRC from fog in April 2023 was not dense enough to stop humans from driving. We know this by the photos of the event, and by the description that the AVs were impeding the flow of traffic from human drivers that were still driving in the fog: https://www.sfchronicle.com/bayarea/article/san-francisco-waymo-stopped-in-street-17890821.php
We know that since then, Waymo has improved their ability to handle fog of some density, but we don't know how dense of a fog it can handle.
It's because GM had the camera system designed as the primary way of attention detection, while the torque on wheel was the primary way for Teslas (cars didn't even have cabin cameras at the start, and of the cabin cameras, only recently was IR even added).
Thanks.
Looking up GM's Ultra Cruise description, I find this:
I wonder what minimum requirement NHTSA will set. They're apparently happy with an IR-lit camera right in front of the driver. Will they allow an offset camera in daylight, and/or an offset camera with IR-lighting at night?
Similar threads
- Article
