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

Tesla Full Self-Driving Beta V9 Will ‘Blow Your Mind,’ Says Musk

This site may earn commission on affiliate links.
As of right now, it seems that the extra $10K I paid for FSD will provide automatic lane changes for faster speed, exiting passing lane and taking exit ramps. This is available only on freeways with a destination in navigation. I am not aware of anything else it will do beyond the standard cruise control and Autosteer. Hopefully, additional functions are to come. Better performance on curves would be a huge plus.
 
That doesn’t even make sense at a logical level.

Vision > Vision + Radar

If, in theory, you could get near perfect with vision, then it only makes sense that adding another sensor into the mix will make you even more better. It only works if you constrain the problem to force trades, but even those constraints will be somewhat arbitrary.

I think his intent is that vision can be good enough that you don’t need radar, but that’s certainly not what he’s saying.
 
  • Informative
Reactions: LightngMcQueen
As of right now, it seems that the extra $10K I paid for FSD will provide automatic lane changes for faster speed, exiting passing lane and taking exit ramps. This is available only on freeways with a destination in navigation. I am not aware of anything else it will do beyond the standard cruise control and Autosteer. Hopefully, additional functions are to come. Better performance on curves would be a huge plus.
I think it gets you automatic parking too and the ability to use it on local roads.

I change cars every 2.5-3.5 years so I didn’t even consider FSD - no point in paying so much for something that may not even be fully complete during my course of ownership.
 
  • Like
Reactions: jimmym1981
That doesn’t even make sense at a logical level.

Vision > Vision + Radar

If, in theory, you could get near perfect with vision, then it only makes sense that adding another sensor into the mix will make you even more better. It only works if you constrain the problem to force trades, but even those constraints will be somewhat arbitrary.

I think his intent is that vision can be good enough that you don’t need radar, but that’s certainly not what he’s saying.
Mathematically, what if radar is negative? then Vision > Vision + Radar!

Back into programming terms, merging 2 sensors and deciding which one wins can add complexity and error. So I understand where he's coming from.
 
Mathematically, what if radar is negative? then Vision > Vision + Radar!

Back into programming terms, merging 2 sensors and deciding which one wins can add complexity and error. So I understand where he's coming from.
Exactly.

People can't get out of the linear thinking.
Well if I have 2 of something than it is better than 1 of something.
That is rarely a given.

But when you add multiple + disparate sensors and trying to unify the data, you have to make some decisions. (just some random thoughts)
  1. You cannot get them to agree 100% - which one do you take as "right" and in which situations? How do you write logic to make that decision for ALL situations?
    • eg LIDAR cannot read traffic light color but knows it is there so you have to rely on camera vision to give you the answer.
    • other examples are not that simple.
  2. You can chose to solve vision with lidar and vision with camera separately
    • that is a LOT of time and money.
    • do you unify them or run them in parallel? and where / how do you unify them?
    • there is no guarantee that the two separate stacks can easily tied together, might require a medium (a NN stack that consumes the output of two vision stacks and puts out one final result)
    • now you have 2 vision stacks to maintain, train, fix, etc
  3. You can do "sensor fusion" and then feed the data into a your main NN stack
    • this is what Tesla seems to be walking away from right now...
People tend to only look at the positive of each component, but each and every component has its drawback.
Complexity itself is a big tradeoff!
 
Exactly.

People can't get out of the linear thinking.
Well if I have 2 of something than it is better than 1 of something.
That is rarely a given.

But when you add multiple + disparate sensors and trying to unify the data, you have to make some decisions. (just some random thoughts)
  1. You cannot get them to agree 100% - which one do you take as "right" and in which situations? How do you write logic to make that decision for ALL situations?
    • eg LIDAR cannot read traffic light color but knows it is there so you have to rely on camera vision to give you the answer.
    • other examples are not that simple.
  2. You can chose to solve vision with lidar and vision with camera separately
    • that is a LOT of time and money.
    • do you unify them or run them in parallel? and where / how do you unify them?
    • there is no guarantee that the two separate stacks can easily tied together, might require a medium (a NN stack that consumes the output of two vision stacks and puts out one final result)
    • now you have 2 vision stacks to maintain, train, fix, etc
  3. You can do "sensor fusion" and then feed the data into a your main NN stack
    • this is what Tesla seems to be walking away from right now...
People tend to only look at the positive of each component, but each and every component has its drawback.
Complexity itself is a big tradeoff!
Great summary. Most of us "armchair specialists" (including myself) are rather high on top of the left side of the Dunning-Kruger scale and fail to see the complexity of it all.
 
  • Like
Reactions: mikes_fsd
I'd expect to be blown away and also underwhelmed. Blown away that a car can do any sort of self-driving, but underwhelmed at how well it does compared to a human driver. Same thing with Advanced Summon. If you told me 10 years ago I'd have a car that can drive around in a parking lot, I'd be blown away. Doesn't mean it's actually practical.
Except - in shopping plaza parking lots - Smart Summon is practical and useful.
Just set a pin with "Go to target" and relax.
Rain, snow, sun, it all has worked surprisingly well since I've tried it again.
 
That doesn’t even make sense at a logical level.

Vision > Vision + Radar

If, in theory, you could get near perfect with vision, then it only makes sense that adding another sensor into the mix will make you even more better. It only works if you constrain the problem to force trades, but even those constraints will be somewhat arbitrary.

I think his intent is that vision can be good enough that you don’t need radar, but that’s certainly not what he’s saying.
Who knows less? Do two people that know nothing know even less than one person that knows nothing? Click and Clack discuss this at the link below.

Car Talk - Skip to about the 37 minute mark
 
Except - in shopping plaza parking lots - Smart Summon is practical and useful.
Just set a pin with "Go to target" and relax.
Rain, snow, sun, it all has worked surprisingly well since I've tried it again.

Found the one person that actually uses Smart Summon.

What you really meant to say is that it's practically useless, right?
 
Found the one person that actually uses Smart Summon.

What you really meant to say is that it's practically useless, right?
no, practical AND useful
  • practical - of or concerned with the actual doing or use of something rather than with theory and ideas.
    "there are two obvious practical applications of the research"
  • useful - able to be used for a practical purpose or in several ways.
 
  • Like
Reactions: Silicon Desert
Except - in shopping plaza parking lots - Smart Summon is practical and useful.
Just set a pin with "Go to target" and relax.
Rain, snow, sun, it all has worked surprisingly well since I've tried it again.
Haven't seen a single video on youtube where someone described it as practical. So I'd like to see a video of it being practical. Every video I've seen says it is more practical to walk to car, unless there is bad weather.
 
  • Like
Reactions: boonedocks
Haven't seen a single video on youtube where someone described it as practical. So I'd like to see a video of it being practical. Every video I've seen says it is more practical to walk to car, unless there is bad weather.
I don't post videos on Youtube, but as I mentioned above it was more than practical to me not long ago and it was not because of weather. Having said that. I only used it when I needed it for a reason. I prefer to walk anywhere whenever I can. I guess we are getting off the topic, yet you reminded me..... my wife will drive around in circles at the gym parking lot to find a parking spot near the front door and then go in and run 10 miles on a treadmill. :rolleyes: If she were driving the Tesla she would probably have it come to her every time.