skitown
Member
Alright, my roads are a bit worse than the above post from @daktari. After multiple attempts I got AP1 to engage, but there was no point as I needed to disengage pretty much immediately.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
.48 feels like AP2 finally passed AP1
- Thread starter jimmy_d
- Start date
-
- Tags
- Autonomous Vehicles
Alright, my roads are a bit worse than the above post from @daktari. After multiple attempts I got AP1 to engage, but there was no point as I needed to disengage pretty much immediately.
Dunno if you remember but:
1-Amnon Shashua showed EyeQ3 acing just that very condition. We don't know if Tesla crippled it.
2- somebody reported in 2015 after hacking the car that in fact AP1 could handle a lot more than we think it can.
Ah... that joke was funny AR. You have to admit he got you pretty good there.
Who's up for some christmas rap? Shiiiiiiiit who am I kiddin' I'm bored and need a hit of a second easter egg fast.
skitown
Member
skitown
Member
One thing I noticed tonight that was fascinating. I mean, I realize I'm late to the AP1 party just getting my used Models S this past summer - but it was fascinating tonight - my car was reading snow tracks as if they were painted lines. So it was engaging and driving on autopilot on totally unpainted roads that now have lots of snow on them. These are roads that it wouldn't normally engage on in the summer (or with no snow). I mean it was frightening and my wife wasn't stoked as it was steering like a drunken sailor
but still an interesting case where it was actually better on the opposite end of the spectrum than ideal conditions.
lunitiks
Cool James & Black Teacher
Engaging AP1 on an unmarked road is one thing - almost never possible.
(I say almost because somethimes my AP1 thinks that it sees lane markings when it’s actually seeing a stripe of snow (no @buttershrimp not that kind of snow).)
But what I find amazing is how good it keeps Autosteering without any kind of lane markings or co... snow at all. Which it often does if I engage autosteer on a marked road in the first place
(I say almost because somethimes my AP1 thinks that it sees lane markings when it’s actually seeing a stripe of snow (no @buttershrimp not that kind of snow).)
But what I find amazing is how good it keeps Autosteering without any kind of lane markings or co... snow at all. Which it often does if I engage autosteer on a marked road in the first place
I like jokes that are sufficiently poignant. I like a good touche of course. This one IMO missed the mark.
This one implied if only I drove more the I'd have California-level mapping or conditions for EU in my car.
Obviously a false premise. One can consider that funny perhaps if they actually don't believe in the regional difference angle. To me it is a frustrating reality that seems to make exchanging experiences here really hard, because you get sunny state people and U.S. mapped/conditions experiences that are very different from snowy RoW.
Yeah, fair point. It is well known that nobody really uses full EyeQ3 capability in production yet. And of course it seems unlikely that Tesla would invest in AP1 improvements too much either anymore...
@AnxietyRanger, what do you think of my four main videos on my site now? www.buttershrimp.com Can your car do this in AP2 in Europe? I have a theory as to why these roads may or may not be possible. Please look at the first video closely in the second half the one where everyone accuses me of having nearly died.
Lol all of a sudden I'm making good points eh? Now that I say something positive about Mobileye and its capabilities I'm making "fair points." Honestly I would appreciate it if you don't even respond to my posts. That would be a great favor - I don't want to hear from you. I'm not the only one. Just knock it off. @Doug_G enough people behind the scenes are telling you what's going on. Merry Christmas everyone.
Time to get back to the family. Plenty of time for this weirdness after Christmas. So long folks.
@buttershrimp I'll look into it, I've used recent versions on motorway only.
Anyway, please don't take my frustrations the wrong way. I do appreciate you making the videos and as posted recently (How is .40/.42 AP2 doing for you?) I totally agree the latest AP2 version I have is driving nicely. I also recently concluded it likely AP2 now handles tight curves and poorly marked roads better than AP1 (from multiple reports of others). It is progressing and I think your videos are nice testament to that.
Personally I had the first good experience already on .36 and it has only gotten better since.
My frustrations are the at times massive differences in anecdotal experience posted and I've been trying to speculate on the reasons. Different regional mapping (Tesla Autopilot maps), markings (subtle differences in training?) and conditions (different snow?) would seem like a plausible explanation.
Other ideas?
Well, I'm just wanting @an
More just wondering if you have the sort of rapid lane shift in that video?... have you tried anything remotely similar to that sudden jog in the road?
Not that I remember. I've used AP2 on divided roads given the bad experiences with early versions and now winter, but I'll promise to report back once I've done so. Still on .46.8 though.
The head on collision in the video? Do you have any situations like that where there is that opportunity to turn and follow lane markings or get hit? The other thing I can't explain is the up and down nature of the experience.
I appreciate your comments about the videos. But now I find myself scouring to find clues in the code.
I will try to find some. So far the places where I drive AP2 weekly are divided roads (motorways and divided non-motorways) that don't have that kind of lane changes. The fact that the ground is a mushy goo makes me a little apprehensive of how accurate my reports would be anyway. But I will look into it and get back to you once I have something.
The thing is, I have stuck to comparing AP2 on the divided roads because it has been my understanding that is where it is recommended for and where early successes can be expected. Unlike some think, I haven't been out to make AP2 look bad. Just looking for a good motorway experience at this stage, would be good enough for me - and since .36 we've been getting there.
For other types of road, I need to activate myself...
Someone give me feedback.
Let me know is you find anything similar to that near collision in my video.
No I appreciate the feedback... I'm trying to find edge cases and places that it struggles. Right now, @brkaus give me a very difficult road to catalog and it is going to keep me occupied.I will try to find some. So far the places where I drive AP2 weekly are divided roads (motorways and divided non-motorways) that don't have that kind of lane changes. The fact that the ground is a mushy goo makes me a little apprehensive of how accurate my reports would be anyway. But I will look into it and get back to you once I have something.
The thing is, I have stuck to comparing AP2 on the divided roads because it has been my understanding that is where it is recommended for and where early successes can be expected. Unlike some think, I haven't been out to make AP2 look bad. Just looking for a good motorway experience at this stage, would be good enough for me - and since .36 we've been getting there.
For other types of road, I need to activate myself...
Let me know is you find anything similar to that near collision in my video.
Similar threads
