Out of Spec roadtrips Gravity and DDP2 has some issues
- Thread starter Bdr100
- Start date
A google search would be wrong. The data does not lie. Tesla has 558K miles with the majority of them being done in austin
By comparison the miles in dallas/houston/tampa/orlando/miami are insignificant
I didn’t say “miles” I said “taxis,” as in the number of vehicles.
Tesla's registered fleet in texas is approaching 1000. They have well over 1000 if you count other states.
Cybercabs alone tesla is close to 4000 manufactured.
you take the weekly mileage and assume 100 miles per vehicle, per day (utilization of AVs is low) and you can see how many vehicles are running
tesla is running about 1000 vehicles now. If you could see videos of austin from the F1 race there were hundreds of vehicles entering and leaving COTA
There was a pretty funny picture with around 50 teslas all lined up at once leaving the dropoff area. Wait times during that day were only 5 minutes which means they were running a lot of cars
Based on the current scaling tesla is doing they will do what they did in austin in houston/dallas/san antoni. they will also unlock freeway which presumably tesla will double the geofences to some absurd sizes.
i would estimate at that point tesla will have close to 2000 vehicles by EOY based on scaling houston/dallas/san antonio as well as unlocking vegas and arizona.
Tesla also released robotaxi safety data which suggests they are ready to test in california. Because of tesla's scale of vehicles it doesn't take more than ~6 months to get approved for driverless.
- Joined
- Aug 23, 2020
- Messages
- 4,363
- Reaction score
- 5,881
- Location
- Paradise Valley, AZ
- Cars
- Air GT, Gravity GT
- Referral Code
- K9WIJHB0
I can't wait for you to tell us how much smarter Tesla Robotaxis are than Waymo because robotaxis know that they can run over plastic barricades without damage and that Waymo needs to learn this skill.
That issue was fixed
keep in mind it was fixed the next day so maybe some mapping is used there
- Joined
- Aug 23, 2025
- Messages
- 928
- Reaction score
- 544
- Location
- West central Wisconsin
- Cars
- Lucid Air Grand Touring
- Joined
- Nov 19, 2021
- Messages
- 16,251
- Reaction score
- 22,407
- Location
- Cupertino, CA
- Cars
- Air DE-P, Gravity DE
- Air DE Number
- 241
- Gravity DE Number
- 013
- Referral Code
- Q1BTN5Y3
Every issue, eventually, gets fixed. Very few are unfixable.
But, suffice it to say that Robotaxi was definitionally not 'perfect' before the fix, yes? After the fix, in order for it to be as 'great' as you imply, it would mean that it would not have such 'dumb' bugs in the future. But it does. Constantly.
And, to be clear, nobody here thinks it's useless. We just think it's not done.
You seem to think it's done. It's not.
People thinking it's "done" is what will kill people.
As if tesla is the only one with bugs? Tesla does not drive with the bumper camera and that might be a fix with v15
Tesla had a limited scale of vehicles and whatever issues they fixed with v15 is what enabled them to scale to 1000 vehicles
But really my argument is tesla seems to be improving at a faster rate. Waymo has years old issues while tesla seems to be able to fix issues much faster and the tech they have with v15 and presumably the roadmap for v16 enables things they could not do before
Last edited:
I think the issue is this is a useless metric.
3 years ago I used a toyota corolla with a weight on the steering wheel in order to trick the lane departure system to work as a lane centering system. I went 60 miles without a disengagement. The car stayed perfectly centered and it kept good enough distance with the car in front. Despite all of that, it was 60 of the most dangerous miles I ever drove.
The problem is highway is the easiest task for an ADAS system but the hardest challenge for an unsupervised system
What you need to do is perform safety tests like they do in china. Put black vehicles overturned on the highway with no lights on at night. See who slows down. Have vehicles cut in at high speed. Test accident avoidance. Test dodging pedestrians and emergency braking at 80mph
Last edited:
- Joined
- Nov 19, 2021
- Messages
- 16,251
- Reaction score
- 22,407
- Location
- Cupertino, CA
- Cars
- Air DE-P, Gravity DE
- Air DE Number
- 241
- Gravity DE Number
- 013
- Referral Code
- Q1BTN5Y3
Please show me where I said this, or do not misconstrue my arguments to suit your own. I will not respond to strawman arguments, so don't make them.
Cool. Not sure how that's related to any of my point.
Cool. Not sure how that's related to any of my point.
Your new argument, you mean. That's fine, but call a spade a spade if you're moving the goalposts.
No, that is just you seeing what you want to see. Which is fine! You're allowed to be wrong and confident about it.
- Joined
- Jul 13, 2023
- Messages
- 1,519
- Reaction score
- 2,546
- Location
- Tahoe City, CA
- Cars
- Toyota Highlander
Indeed. I worry the most when I hear someone say how much they can relax when an autonomous system takes on the driving for them. To me, that's a strong sign that they're not going to be prepared to leap in and take over when necessary. FSD fans may be most guilty of that, but I've seen people be pretty cavalier about DDP as well.
- Joined
- Apr 15, 2025
- Messages
- 970
- Reaction score
- 1,729
- Cars
- GGT, Merc E450 Conv
Yeah. I find DDP2 handsfree most useful and somewhat relaxing either in bumper to bumper, stop and go traffic, or out of city, on interstate, where everyone is traveling at similar speeds, with minimal lane changing. Using it in city freeways, with moderate traffic, cars merging, entering and exiting, etc., can be a white knuckle experience trying to trust DDP.
- Joined
- Sep 27, 2026
- Messages
- 138
- Reaction score
- 73
- Cars
- Lucid Air Pure
What does the actual data show? Are there more or less accidents compared to human drivers? That will tell if it makes driving safer. As long as it is safer than a human drivers, how does it make a difference in what the driver believes?
I never said robotaxi was done. I'm saying that tesla is improving at a faster rate currently than waymo who seems to show relatively little improvement on years' old issues.
you can see this right now. Guess who is not running on freeway for the 2nd time? Both times for having very serious failures.
The difference is that tesla improves at a faster rate, tests on millions of consumer cars and knows whether their system has reached a threshold that is safe for deployment.
on the other hand waymo uses simulation, trains on much more synthetic data, and has a sim 2 real gap that obvious does not account for the edge cases they see in the real world
This is the same company that pushed out a "fix" for blowing past school buses and then received more violations than ever after the fix was implemented. And if this is your "fix" after 10 years of driverless operation, what did the car perform like in 2021?
This forum is one big circlejerk where people only want to believe what they want to believe. The evidence clearly shows something different, but people can't accept it because it's not what they believe.
Everyone said on the tesla forum that if tesla had radar/lidar/corner cameras that they would be better at performing unprotected turns. Even to me this sounds logical. The chinese have that and they suck at unprotected turns.
Last edited:
I'm tired of repeating this but nobody understands what it takes to have a system that runs 1300 vehicles unsupervised and scaling at the rate they are doing. This is a massive hurdle that people just don't understand.
What they don't comprehend is the same system that enables these fast reactions, accident avoidance, etc. which it actually takes to run these cars is the same thing found in consumer vehicles.
The threshold of safety is many thousands of times better than what is found in other consumer cars. While you do have to supervised such a system, you barely have to pay attention to get the full benefit of safety.
With other point to point systems like Nvidia Alpanmayo they use VLA models and the like. The problem with vla models is they present as "intelligent" driving and are able to handle what appear to be complex scenarios, but the latency is so high that that the edge case performance is bad. Latency affects every task. High latency causes jerky steering decisions, can't avoid accidents, can't dodge debris on the roads, and will never brake for a pedestrian that pops out from behind a car. All of these things are dangerous and do not improve safety.
This actually reduces safety because when you're not paying ful attention, your reaction time is reduced.
Everyone is using a double standard here. Tesla at one point was very far from unsupervised, and people said that their cars would never cross the ridiculous safety threshold needed to do this with E2E neural networks and without HD maps. Now they have done it and people pretend like tesla driving is not that good.
Last edited:
- Joined
- Jan 4, 2022
- Messages
- 5,714
- Reaction score
- 11,034
- Location
- Santa Clarita, CA
- Cars
- 2026 Gravity Dream
- Gravity DE Number
- 68
- Referral Code
- AWNJLGKT
Just like people are no doubt getting tired of your repeated false claims, misinformation and shifting the goal posts all the time.
Yet another FALSE claim. I’ve seen Waymo’s on freeways multiple times in the last few weeks of driving in LA alone.
Quit your fucking nonsense and do some research.
Waymo shut down freeway as of today
Because of this issue
- Joined
- Jan 3, 2022
- Messages
- 7,604
- Reaction score
- 8,742
- Location
- San Francisco Bay Area
- Cars
- Gravity GT, XC40 P8 EV
- Referral Code
- 3OKY7YGA
It could be that nobody cares. You're repeatedly posting this in a Lucid forum, in a thread that is about Lucid's lanekeeping system.
I'm sure there are other forums where there people would have more interest in your topic.