Whew-- lots to cover here- consolidating replies to all 3 of your posts.
Why are you so tied to exactly how something is in one moment?
Because it's the moment people keep inventing fictional FRAUD AND SCAM narratives for.
So pointing out what happened, when, is directly relevant.
It's not surprising you wouldn't appreciate this since you're on the other side of the discussion.
Sure in the beginning it was vision plus frontal radar, and now its vision only. What difference does that really make when Vision is the bulk of it?
Because vision was
not the bulk of it back in late 2016
Radar was.
Tesla from late 2016 said:
After careful consideration, we now believe it (radar) can be used as a primary control sensor without requiring the camera to confirm visual image recognition. This is a non-trivial and counter-intuitive problem, because of how strange the world looks in radar.
Then, later, they realized no- it can't.
So they decided vision-primary with radar backup would be "the" solution.
Then, later, they realized- no, it can't.
So now they're trying vision only.
The system has
fundamentally changed multiple times at a very basic design philosophy level.
So the narrative they "knew" they'd be where they are today, still at L2, is
provably false.
THAT is why points in time matter.
Furthermore you know Elon isn't a Sensor fusion proponent like myself.
[
Again this is outright false.
Elon is a "best solution for the problem" proponent.
For
most of the entire history of AP that included sensor fusion.
For SpaceX it
still does
But for this specific situation they eventually found sensor fusion too limiting, and have moved to something else.
Fairly recently in fact.
So your trying to pretend it was "always" the plan from the beginning just isn't' gonna work in the face of actual facts.
So it really doesn't make one lick of difference.
See above- it makes
all the difference[/B]
Or did you not notice the word "timeline" in the title of the thread?
So yes, the order, in time, which thing happened makes a great difference.
Just like what Tesla tells the dmv about the current iteration of FSD Beta today doesn't mean Tesla doesn't have the full ambition of having FSD as a L4 system eventually. That its the end goal.
...what?
"FSD" the end product isn't what is discussed in the CA DMV docs.
City Streets, which some call FSDBeta, is. And is
explicitly only, ever, intended as L2
That's
not the same thing
As to EAP it doesn't really matter that in my experience it sucks, and in your experience its great. What matters if how useful it is for the majority of the people, and it doesn't seem to work all that well.
Again, your basic claims are inaccurate.
EAP is far more than smart summon, and I'd suggest you'd be hard pressed to find any threads on here in recent couple of years where "the majority" of people don't think it works well.
I don't think this is a case of getting the timeline wrong.
This is a case where Elon didn't tell consumers what the game plan was.
Case in point.
If the plan changed, fundamentally, multiple times, you can't ALSO claim he "knew" this is where we'd be now.
With the rockets he used his own money to iterate the design over and over until it worked. It was a completely different kind of problem though where they had control over their own destiny.
With Self driving cars no one has control over their own destiny because the public has a say in what's allowed on the roadway.
... what?
Rockets are
more regulated than l2 driving systems today.
In many states even L5 driving systems are
less regulated than rockets
The one it's missing?
Recognizing a vehicle parked partly in its lane/not fully pulled over to the shoulder.
What L3 ODD? Traffic Assist L3 or up to 80mph freeway L3?
I guess we can add "owners manual" to the list of basic reading on the topic you haven't done
The ODD for NoA is limited access divided highways (ie using on/off ramps, no intersections or cross traffic, no oncoming traffic).
I don't believe the current version qualifies because
[*]Phantom braking -> Tesla Vision version is really bad
On city streets code it's annoying (though it's usually only a 1-3 mph slowdown in recent versions)
I experience virtually none on highways using the legacy wide release NoA code though... pretty much only time I get hard braking there's nothing phantom about it... (for example if I'm in the right lane approaching an on-ramp merge lane and someone is getting over in front of me- the car might brake hard there but hardly "phantom" the cause is right in front of your eyes.
[*]Lack of Smoothness during low speed traffic -> Can't see the point of releasing something too uncomfortable to actually use.
Me neither.... good thing they didn't. Again no issues.
[*]Too many lane change cancelations triggered by phantom vehicles.
You can adjust how aggressively it chooses to make these.... though even then the only lane aborts I can recall experiencing in any recent version are when a car is approaching at speed from the rear after cresting a road rise or something.
[*]Out of date Navigation giving incorrect information to the NoA system
This isn't a system limitation at all- it's a mapping problem.
[*]Doesn't respond to blinkers from cars ahead changing lanes. I often have to cancel to let them in
The braking incident I mention involved his blinker being on to merge in front of me.
FSDBeta visualizations actually SHOW the turn signals being seen/understood.
[*]Bad logic in determining appropriate times to change lanes before an exit. It will often put the blinkers on when there is car right next to me when it could have easily gotten over when there wasn't anyone there
FINALLY a thing that has actually happened to me
That said- this is pretty easy to train better in the long run- and not a "isn't capable of the task" thing, it's a "could do the thing better" thing.
[*]Doesn't seem to have good logic on when the rear camera is dirty. Heck it doesn't even seem to use it.
It does use it of course. I'm not sure what "logic" you're looking for it to have regarding any dirt on it?
Then again, the primary rear-relevant sensors would be the side ones, since they see approaching cars in the lanes you'd be changing into.
[*]No debris recognition/avoidance.
Tesla’s Full Self-Driving Beta has impressed with its capacity to navigate city streets, and it now looks like it can...
electrek.co
That's in the beta of course, but you can expect that to come to highways with the single stack in V11- so it's a largely solved problem.