Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

This site may earn commission on affiliate links.
Texas football? They LOVE their football, will book out hotels in the entire city when there is a home game.

Texas football? They LOVE their football, will book out hotels in the entire city when there is a home game.
Thought maybe was that with say homecoming or something like that. Checked no game that weekend.
 
That voxel object detection is in the FSD 10.2 stack. You can see this crosswalk post visualized in voxels in one of the subsequent tweets in that thread:


I don't think there's anything wrong with perception. It can probably see the obstruction, it just doesn't yet know how to respond to it. The way those construction cones were laid out would have required the car to cross the double yellow.
Well, disagree. If car saw object it should break, not accelerate still. Even AEB should break in this case imho. remember the ufo quote Elon gave in the Sandy interview months ago ? im really disappointed to see it doesn’t slow when uncertainty rises :(
 
I always assumed these were caused by options activity large enough to move the needle. So if someone was holding a significant amount of 860 strike ITM puts and exercised them all at once, the NASDAQ might temporarily record the ask price close to 860.
That's always been the general assumption, I just never see it for other companies. Some day I will find us the answer!
 
  • Like
Reactions: willow_hiller
Whats going on in Austin next weekend? My son invited us down for when he gets engaged. So went to book hotel room and the cheapest place is a Motel 6 $313 per night. Next is $659 per night. Just crazy. Could a lot of Tesla employees be coming down for some serious testing and launching of Model Y production?

Your likely culprit
 
I'd expect Kato 4680s to Austin would be the flow for the first year.
Super rough numbers:
10GWh at 300Wh/kg = 73 million pounds (plus containers) / 40,000 lbs per semi = 2,000 semi loads/ year or 6 per day. 3,400 miles round trip, 60 hours at 60 MPH, call it 3 days. 18 semis full time.
At 10 MPG, that's 340 gallons a trip * 2,000 trips = 680,000 gallons @ $3.3 = $2.2 million in fuel costs or roughly $20 per vehicle produced.
Thanks. I think at the end you meant to suggest a fuel savings of $122k per year per truck. These numbers also suggest 378k miles per year per truck. A savings of $0.32/mile.
 

Your likely culprit
Formula 1 race :)
 
IDEA:
Why can't our cars take pictures of potholes, broken signs etc and automatically email the appropriate road maintenance department ?

Not a lot of code to do it.
Indeed. Tesla could also produce detailed maps of road hazards, poor markings and other anomalies. These maps could be used in a wide variety of ways, but most importantly it could create political incentives for local road maintenance authorities to do a thorough job. Local voters will have an objective criteria to compare local roads with the rest of the country. Auto insurers would also take an interest in these maps.
 
Check AAPL today. The stock tanked AH so I assume a few contracts got executed. There are like six of those spikes.
actually Apple tanked due to a Bloomberg report right after market closed saying they are going to have to reduce iPhone supply significantly due to chip shortages.

Apple Inc. is likely to slash its projected iPhone 13 production targets for 2021 by as many as 10 million units as prolonged chip shortages hit its flagship product, according to people with knowledge of the matter.

The company had expected to produce 90 million new iPhone models in the last three months of the year, but it’s now telling manufacturing partners that the total will be lower because Broadcom Inc. and Texas Instruments Inc. are struggling to deliver enough components, said the people, who asked not to be identified because the situation is private.
 

So yeah Ford......so much for that huge demand you're seeing :rolleyes:

The longer the "chip shortage" excuses go on, the more it's apparent it's a demand problem for legacy auto.

FORD knows that in 6 years, they'll be broke, and the ICE resale market will have melted... :p

Advice: STAY AWAY from Ford Bonds.

Cheers!
 
I've been using the beta now for a couple days, having taken 10-15 drives around the city with it. I'm honestly a bit perplexed at how badly it handles certain situations. It fails to act normally in what seems like a very uncomplicated situation. Then other times it performs really well in what seems like a complicated situation. In general, based on the visualizations I would say that it seems to understand the world quite well, but makes poor decisions based on that data. It's really quite amusing to have it sometimes start a right turn and then stomp on the brakes in the middle of the intersection with no other cars oncoming, but then also on a random unprotected left turn find an appropriate gap between oncoming cars and begin accelerating before the first car passes to perfectly and smoothly maneuver through the gap.

Despite my complaints I do see a clear path to success here though at least as far as assisted driving goes. I look forward to seeing improvements as updates are released.
So, it drives (and makes decisions) like a 16 year old not raised on a farm.
 
@The Accountant @Artful Dodger @Curt Renz @Papafox

I’m just going to fall right on my sword — I’ve been name dropping. I’ve been referring folks to TMC and specifically telling them to check you guys out.

I just drove from Southern California to New Orleans to Tuscaloosa Alabama and now I’m driving home. I’ve been telling folks about you guys all across America. I’m hoping forgiveness is easier than permission.

Oddly, I ran into a fellow Southern California TMC’er at a Supercharger in Meridian Mississippi. We maybe growing exponentially, but it’s still a small world. A shout to Mace; very cool plaid Model S.

I'm just happy to be in there somewhere. ;)

im-somebody-now.jpg


Cheers!
 
Well, disagree. If car saw object it should break, not accelerate still. Even AEB should break in this case imho. remember the ufo quote Elon gave in the Sandy interview months ago ? im really disappointed to see it doesn’t slow when uncertainty rises :(

We both agree that the driving policy failed in that situation; but we can't know for certain that the perception stack failed. Given what we know about how the perception works, those construction signs are within the realm of perceivable objects, but we can't be positive the car saw them because it didn't seem to react to them.

But still, I would advocate that we analyze FSD's performance on perception separate from it's performance on driving policy wherever possible. Many people think that FSD is impossible because they don't believe the car can accurately perceive the world with cameras alone. But often times FSD is correctly perceiving a situation, and then failing to act appropriately due to driving/decision making policy.

Here's a concrete example from FSD 10.2 where the perception stack worked perfectly well, but the driving policy failed. Here's the timestamped YouTube video (don't want to clog up the thread with the embed). The car turns right at a stop sign into a small intersection between the two parallel roads, clearly sees the red lights as they're displayed on the visualization, and proceeds onto the second road through the red light. Very clearly not a failure of the car to perceive the red light, but probably a failure of driving policy. FSD probably thought it was still in the middle of an intersection after turning right at the stop sign, and proceeded through the red because it's been programmed to not linger in the middle of intersections.
 
...there is no guarantee that the maximum [ability of FSD will] be able to drive at an acceptable level for regular consumers and regulators.

This "no guarantee" argument, often used by bears, is misleading. There is no guarantee that the sun will rise in the east tomorrow. The relevant question is: What is the probability that it won't?

To believe FSD won't learn to drive, you have to believe that the pattern recognition required for driving is so difficult that neural net computers can't possibly master it despite Tesla's exponentially increasing data collection and exponentially increasing computing power (Dojo, Hardware 4, etc.). How probable is that?

If you spend some time learning what pattern recognition has already been mastered by neural net computers (games, reading xrays, translating languages, writing poetry, driving Elon home without interventions, etc.), you might see the trend, and conclude that the probability of it failing to continue is very low. AI is just getting started at upending civilization, and superhuman FSD seems just a matter of time.
 
Turn the question around: Does anyone on this Thread think FSD 10.2 is ready to do handle the driving in their own home city. Meaning the chance of a driver intervention is near zero? My response for myself is "no".

Lol, you have it EXACTLY backwards: Elon was clear years ago that "feature complete" means that all FSD functionality is present and has a NON-ZERO chance of driving without intervention.

Tesla will use this standard for its Auditors when they book the revenue backlog from FSD payments (currently held in a asset account, but not booked as revenue).

Its a $1B pool of money on the Balance Sheet that will move to the Income Statement. And no, I don't think European Regulators affect the status of "Feature Complete".

To phrase this issue operationally, which feature of "Navigate on City Streets" do you think may still be missing in FSD beta 10.2 that would NOT meet the definition of non-zero chance of working?

Hint: Human drivers aren't perfect either, that's not the standard for "Feature Complete": (2019Q3 Conf.Call)

Elon R. Musk -- Founder, Chief Executive Officer & Director
Yeah, feature-complete, I mean, it's -- the car is able to drive from one's house to work, most likely without interventions. So it will still be supervised, but it will be able to drive -- it will fill in the gap from low-speed autonomy -- low speed autonomy with Summon. You've got high-speed autonomy on the highway, and intermediate speed autonomy, which really just means traffic lights and stop signs.​
So feature-complete means it's most likely able to do that without intervention, without human intervention, but it would still be supervised. And I've gone through this timeline before several times, but it is often misconstrued that there's three major levels to autonomy. There's the car being able to be autonomous, but requiring supervision and intervention at times. That's feature complete. Then there's -- and it doesn't mean like every scenario, everywhere on earth, including ever corner case, it just means most of the time.​
And then, there's another level which is that we think it's -- that from a Tesla standpoint, we think the car is safe enough to be driven without supervision. Then the third level would be that regulators are also convinced that the car can be driven autonomously without supervision. Those are three different levels.​