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

Elon: FSD Beta tweets

This site may earn commission on affiliate links.
The problem with all of this is it's dependent on a ton of interconnected technology and falls apart as soon as any of the links fall apart. (just as an example, I was driving to work today and had no connectivity in suburban Minneapolis.) Not to mention the issues faced when you get to less-traveled areas, new obstacles, etc.
But so does the Internet, and GPS, and FM radio for that matter. I don't think anyone is saying these must be 100% available, but certainly these are sources of data the car can draw upon when they are available.
 
Comparing the reliability of Internet and GPS seems like a poor idea, I guess unless you drive somewhere without sky visibility. ;)
Actually, there are plenty of tunnels where GPS is unavailable. I agree that it makes sense to use them as auxiliary sources but relying on them is a recipe for failure.

Prime example - they are completely reconstructing a street near me, changing it from a 2 lane road with a dirt shoulder to a road with a center turn lane and 8 foot shoulders. They are also straightening some of the curves. If the car relied on map data or GPS data it would end up on the wrong side of the road or in the ditch.
 
  • Like
Reactions: Sigma4Life
Comparing the reliability of Internet and GPS seems like a poor idea, I guess unless you drive somewhere without sky visibility. ;)
I live in a neighborhood that has a parking deck covering it (almost 1.5 miles in circumference and also 30 Tesla Superchargers so it is Tesla heavy) and my parking space is in a nested area. Depending on which entrance/exit I use it can be over a mile of no GPS driving with lots of turns and ramps. So I have cell (internet) but no GPS.
 
You aren't being very clever. We have been discussing this topic to death over the last 5 years.

Do the cars need to be able to drive when the HD Maps are not accurate / out-of-date ? Yes. If they can drive when they don't have the benefit of accurate HD maps, why do they need HD Maps to start with ?

Everyone in the industry understands (explicitely stated by GM, MobilEye among others) that HD Maps are not scalable.

I'm not thinking about high definition maps, but low definition ones which have semantic information from human drivers (where do people actually drive? Like humans understand natural language signs that the robots don't). The information humans use to drive most of the time by watching others and their experience with the routes they drive on. I agree that the system should be able to operate with no map information but would perform better, more naturally and more confidently when there are data available, which should be most of the miles travelled eventually.
 
Elon Musk suggesting FSD Beta increasing to 1 million users by the end of this year probably means there will be a lot of new FSD purchases/subscriptions and probably dropping/significantly-lowering Safety Score requirements for some regions. I believe cumulative Tesla vehicle production will be close to 4 million by end of this year including worldwide vehicles and older ones without the necessary hardware. Maybe there will be a lot more FSD Beta countries added by end of this year too as I would think current FSD is mostly US buyers.

Is this before or after one million robotaxis before the end of the year 2020?
 
I think they will miss big on this number
After finding the original video / quote from All-In Podcast, he seems to correct himself by suggesting the numbers are quite approximate:
  • "like 100 thousand people in the beta"
  • "expanding that to probably millions of people -- or a million, I don't know -- on that order by the end of the year"
I suppose being generous, he probably rounded up from ~60k -> 100k current beta, and I guess 0.5M could be considered "on that order" going from 5 "hundred thousand" 10⁵ to a half "million" 10⁶?

Even more broadly, this is then basically a rephrasing of his previous comments of wide beta later this year without requiring safety score, so he was just approximating the number of beta-requesters by the end of this year to be closer to 1 million than 100k.

 
Comparing the reliability of Internet and GPS seems like a poor idea, I guess unless you drive somewhere without sky visibility. ;)
Like in a city? :)

You would be surprised how often the car falls back on dead reckoning while in built-up (downtown) areas. My point was that a system that uses live traffic information to predict slow-downs is a potential "helper" for the car .. useful when available, but the car should not have to rely on it to drive safely (like a GPS signal). That's why the Tesla approach is so interesting, since it assumes the car will need to find its way even without maps or any advance knowledge of the roads it will be driving. But that doesnt mean it cannot/should not take advantage of extra data when available.
 
  • Like
Reactions: sleepydoc
  • "like 100 thousand people in the beta"
  • "expanding that to probably millions of people -- or a million, I don't know -- on that order by the end of the year"
That makes sense.

Expanding the beta to anyone who buys FSD is well within what they can do this year - just that they will be taking a much bigger risk. Though one could argue, there isn't a 10x risk by going from 100k to 1M - but qualitatively if Beta goes from 100k to 1M users, people expect it to be better.

We have to just hope that - just like Summon - enough people will find it unusable and don't use it until FSD actually gets better.
 
Comparing the reliability of Internet and GPS seems like a poor idea, I guess unless you drive somewhere without sky visibility. ;)
From your location .... looks like you are not in a large metro. I can assure you in downtown, if you come up from an underground parking lot, the car takes a while to find GPS. But cell signal is available even before you reach ground level.
 
"We have to just hope that - just like Summon - enough people will find it unusable and don't use it until FSD actually gets better."

Lmao. The lengths people go to in defending crappy code and design. I'm secretly hoping "FSD" gets banned so they go back to working on the 95% use case, Autopilot. It's been embarrassing since Openpilot surpassed controls quality all the way back in 2019.
 
  • Funny
Reactions: momo3605
Lmao. The lengths people go to in defending crappy code and design. I'm secretly hoping "FSD" gets banned so they go back to working on the 95% use case, Autopilot. It's been embarrassing since Openpilot surpassed controls quality all the way back in 2019.
Can you provide any unbiased quantitative studies showing Openpilot exceeding FSD beta performance?
 
  • Like
Reactions: JupiterMan
Looking forward to hearing about how they broke through the local maxima. Maybe robotaxi prototype will be shown?
Since it is a recruiting event, I'd expect more of: here are all the cool tools we have, difficult problems we need to solve and how we could use your help. Looking forward to hearing about FSD 4 hardware.

I wonder if this is an omen that Andrej isn't coming back.
 
Last edited:
Not an fsd beta tweet but looks like this will likely be 85mph:
I wonder if this is due to actual improvement in Tesla Vision, or they just had too many complaints from Elon? 85mph seems like it'll clear all of the forced disengagements I've had, since that means it won't actually disengage until about 87mph or so.

EDIT: longer minimum following distance - I guess that's how they managed it. Can't complain about that. I want a longer maximum following distance too! EDIT 2: Never mind, this is the following distance 2 rather than 1 or whatever, no change. I still want longer maximum following distance. Current setting is ridiculously tight; pretty sure it is less than 3 seconds!