I wish I could remember which program it was (Waymo?), maybe more than one, that said they have no intention of commercializing level 2 or 3 to the public because of the attention problem with complacent drivers when FSD systems are almost good enough. Inevitably humans will lose focus and/or interventions will be so rare that the human won't be ready to take over when needed. So customers won't be able to buy a (Waymo?) FSD vehicle until they've achieved level 4 if not 5. IOW no middle step(s). However, that was back a few years when everyone thought we were closer to real self driving than we are now. And clearly several automakers, Tesla, Mercedes, Ford, GM are selling level 2/3ish cars to the public now. Clearly geofenced vehicles like Waymo/AZ-SF and GM Cruise SF wouldn't be offered for sale to customers, too expensive anyway.
-
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
It was Waymo or the Google Self-Driving Project as it was called back then. Back in 2013, they considered selling a "hands free" highway autopilot system to consumer cars but decided against it when they let their employees test the autopilot system and found that they got very complacent, including sleeping, texting and applying make-up, despite camera monitoring. Krafcik, then CEO of Waymo, says that experiment was a key factor in Waymo deciding to go all-in on L4 only.
“What we saw was that our testers put too much trust in that technology. They were doing things like texting, applying makeup, and even falling asleep that made it clear they would not be ready to take over driving if the vehicle asked them to. This is why we believe that nothing short of full autonomy will do,” he explains.
“The only way to solve the problem of roadway safety, and the only way to deliver the opportunity of mobility for all, was to take the human completely out of the loop. We committed then to full autonomy, no driver monitoring, nor driver’s license, required,” Krafcik adds.
Source: Early Waymo AutoPilot Experiment Shows Why Full Autonomy Is Best
Correct. But the current Waymo robotaxis are not meant for consumers anyway, they are only intended for ride-hailing. Since Waymo is only interested in L4, it stands to reason that if they ever offer a product for consumer cars, it would only be L4 and only when costs come down.
Personally, I think Waymo could offer "L4 highway" on consumer cars at some point. Highway driving is more limited than city driving so they might be able to do L4 highway with fewer sensors. And Waymo has been testing on highways a lot. So conceivably, once Waymo has validated safety on highways, they could look to develop a cheaper sensor suite that would only be for L4 highway on consumer cars. I think L4 highway would be a very useful consumer product since it would mean that the human could take their eyes off the road and do other things during highway driving.
Last edited:
Tam
Well-Known Member
Thanks for clearing it up! It sounds like some articles are confused about what is available now "eyes-on" with the future prediction of "eyes-off".
diplomat33
Average guy who loves autonomous vehicles
As far as we know, Waymo is not working on L4 consumer cars, only robotaxis right now. Waymo co-CEO Mawakana has said that Waymo is focused on robotaxis but is open to other use cases in the future. I interpret that to mean that Waymo could maybe do L4 consumer cars in a few years after they scale their robotaxis to a few cities. Personally, I hope they do because I think L4 consumer cars are more useful than robotaxis. I just think it would make sense that once Waymo feels good about their L4 on robotaxis that they would want to adapt the tech to consumer cars. That would be a very lucrative market. I would add that a few companies like Ford and Mobileye are targeting 2025-26 for deploying L4 highway on consumer cars. Considering how good Waymo's L4 is, I think it would only make sense for Waymo to also do L4 highway on consumer cars in 2025-26 to get a piece of that market.
Last edited:
diplomat33
Average guy who loves autonomous vehicles
New SuperVision video:
Description:
Climbing the scenic 395 road in the Jerusalem hills, starting from Eshta'ol and continuing all the way to Tzuba, which involves navigating through an exceptionally narrow single-lane road that accommodates traffic in both directions. This is Zeeker001 equipped with Mobileye's latest version of the Supervision system.
Description:
Climbing the scenic 395 road in the Jerusalem hills, starting from Eshta'ol and continuing all the way to Tzuba, which involves navigating through an exceptionally narrow single-lane road that accommodates traffic in both directions. This is Zeeker001 equipped with Mobileye's latest version of the Supervision system.
That vehicle did a good job on that single lane road. We have a lot of roads like that near me. Except instead of a smaller Van coming the other way we end up with large pickup Trucks towing Trailers.
Huawei keeps advancing with ADS 2.0 update and will support 45 cities in china by end of 2023.
I really think Huawei might be leading consumer ADAS among all companies. I hope the reports of FSD being tested in china is true and it gets released by end of 2023 so we can get a direct comparison with Huawei's ADS 2.0
it will really settle the score on who is really leading consumer ADAS.
Even the biggest Tesla fans in china recognize the supremacy of Huawei's parking software. I think a showdown at the end of the year will really open up the curtains.
I really think Huawei might be leading consumer ADAS among all companies. I hope the reports of FSD being tested in china is true and it gets released by end of 2023 so we can get a direct comparison with Huawei's ADS 2.0
it will really settle the score on who is really leading consumer ADAS.
HUAWEI ADS 2.0 首发 一文读懂 2023 华为智能汽车解决方案发布会_话题文章_新出行
阿迪同学：4 月 16 日，华为在上海车展前夕举办 2023 华为智能汽车解决方案发布会，在会上，华为首发 HUAWEI ADS 2.0 以及智能座舱、智能车载光、智能汽...，来自新出行话题文章
www.xchuxing.com
Even the biggest Tesla fans in china recognize the supremacy of Huawei's parking software. I think a showdown at the end of the year will really open up the curtains.
Last edited:
AlanSubie4Life
Efficiency Obsessed Member
Wow that was painful to watch. Awful.
Not that Tesla’s would have a hope.
Just awful in general. So far to go.
Why is it so hard to park FAST? That should be the main advantage of these systems - they should be LIGHTNING quick.
AlanSubie4Life
Efficiency Obsessed Member
It took TWO MINUTES to do a simple parking maneuver. Insane!
Hopefully it is not going to take another decade to shave a minute off of that.
Yeah, that one wasn't really impressive, but there are other Chinese system demos (not from that vendor) that showed scenarios that are more impressive. Beating Tesla's autopark isn't that hard though, it's a fairly low bar.
AlanSubie4Life
Efficiency Obsessed Member
Yes, it’s an extremely low bar. The Tesla system is completely and utterly useless. When it goes off and does it on its own (coming soon allegedly, though presumably that is another case of irrational exuberance!) then it won’t matter how long it takes.
I’d love to see a demonstration of something actually useful.
It still mystifies me why it is so slow. Is this to make the human operator feel more comfortable? I assume it could do it 10x faster no problem. Or is this to limit damage in the case of an inevitable collision with an unhandled unusual obstacle (or perhaps a random human or whatever nearby who runs into the way?). But even 3x faster seems like it would be safe.
But if it is to limit damage in the case of system failure, it means we have even further to go, of course. 15 years?
Guess this thread will end up pretty long.
Last edited:
diplomat33
Average guy who loves autonomous vehicles
diplomat33
Average guy who loves autonomous vehicles
it was a difficult parking spot done on purpose to showcase how the car can park in basically any parking spot. other cars wont even be able to attempt this. Out of all autopark system tested in all scenarios, Huawei is the best and fastest and its not even close.
AlanSubie4Life
Efficiency Obsessed Member
I don’t doubt it. Yet it is still awful, which was what surprised me.
It’d be interesting to know if anyone actually uses any of the self parking systems on the market today.
I’ve never seen one that could do a better job than I could, and in reality you’ve got to be BETTER at parking than the self parking system is so that you can predict and intervene before it does something dumb and kerbs your wheels.
I’ve never seen one that could do a better job than I could, and in reality you’ve got to be BETTER at parking than the self parking system is so that you can predict and intervene before it does something dumb and kerbs your wheels.
You must not have watched the video. That was an extremely difficult parking situation, very limited front and back room with two very narrow stalls to work with. I doubt many humans could manage that exact situation in less than 2 minutes as claimed above.
The garden variety parallel parking assist (not at all what was in the video above) cars I've seen work very well and very quickly, I don't know how a human could do significantly better, and most could not.
AlanSubie4Life
Efficiency Obsessed Member
Haha. Perhaps not any significant difference in performance the first time, but subsequently humans would fly through this task. It’s not challenging with a modern car with sensors and cameras especially with 360 overhead.
It seemed like a quite straightforward multipoint turn. (Which the system made a mess of - look at its unnecessary unproductive maneuvers (0:46)!) Now if there were a car parked in that other spot, it would make it harder (both for the human and the system - human could definitely do it though). I’d like to see it do that.
diplomat33
Average guy who loves autonomous vehicles
Similar threads
- Article
