FSD rewrite will go out on Oct 20 to limited beta

[FloridaJohn]

I agree with Turnem.

I think EAP will get FSD beta in January. And wide public release at the end of Q1 or Q2 at the latest. Though they may disable some features in the public/wide release at first.

That will definitely suck for those people getting three months of free FSD at the beginning of the year.

 

[EVNow]

L4 is defined as fully autonomous in a limited ODD, irrespective of the size of the ODD.

This is exactly the problem with L4.

ODD can be made smaller and smaller - till it almost vanishes.

Using this kind of theoretical definition makes SAE irrelevant and useless.

Tesla can also claim L4 - some 100 ft road in a desert with no traffic.

 

[mikes_fsd]

This is exactly the problem with L4.

ODD can be made smaller and smaller - till it almost vanishes.

Using this kind of theoretical definitions make SAE irrelevant and useless.

L4 is defined as fully autonomous in a limited ODD

Even Waymo CEO thinks the SAE levels are complete BS.

 

[diplomat33]

This is exactly the problem with L4.

ODD can be made smaller and smaller - till it almost vanishes.

Using this kind of theoretical definitions make SAE irrelevant and useless.

I concede that L4 is an incomplete definition since requires additional information in order to be complete. But how do you propose we fix this problem? As the SAE explains, there are many possible types of ODD. Coming up with a complete nomenclature that fits all possible ODDs is not really possible.

Even Waymo CEO thinks the SAE levels are complete BS.
View attachment 622584

He said he does not love them. That is not quite the same thing as saying that they are "complete BS".

 

[mikes_fsd]

He said he does not love them. That is not quite the same thing as saying that they are "complete BS".

Uhm, his example is just 2 steps ADAS and FSD, that completely trashes the 5 SAE level approach altogether...

 

[Snuffysasa]

Even Waymo CEO thinks the SAE levels are complete BS.
View attachment 622584

I agree with John

 

[Snuffysasa]

Uhm, his example is just 2 steps ADAS and FSD, that completely trashes the 5 SAE level approach altogether...

There really should just be 2 categories, ADAS and Self Driving (autonomous).

If you want to break down those categories into more, like L1 and L2... but not really necessary.

and if you want to break down Self driving into ODDs or L3... but again not really necessary

 

[mikes_fsd]

I concede that L4 is an incomplete definition since requires additional information in order to be complete. But how do you propose we fix this problem? As the SAE explains, there are many possible types of ODD. Coming up with a complete nomenclature that fits all possible ODDs is not really possible.

Better ODD would be "all paved roads" + "in all non 'Weather Warning' conditions" as issued by Active Alerts

 

[mikes_fsd]

There really should just be 2 categories, ADAS and Self Driving (autonomous).

If you want to break down those categories into more, like L1 and L2...

I don't...

 

[EVNow]

As the SAE explains, there are many possible types of ODD. Coming up with a complete nomenclature that fits all possible ODDs is not really possible.

That is why they should apply practical standards. That is how they create - for eg., USB standards. They specify in what conditions certain devices etc have to work, at what speed etc. You can't leave critical operating parameters like ODD to manufacturer discretion.

For eg., SAE (or its US affiliate) can say, FSD Level X is when
- It works in top 100 counties, by population
- In all weather conditions, except when certain enumerated emergencies occur

Not just these, various levels should be such that they are *real* practical steps to get to full autonomy. There appears to be little difference between L3/L4 - but a huge spasm between L2 & L3 (if we don't consider trivial ODD). Infact, now that we have several companies trying FSD at various stages of development - they could make a 10 step gradation that goes from simple CruiseControl to FSD in all conditions, everywhere 90% of people can safely drive.

 

[diplomat33]

Uhm, his example is just 2 steps ADAS and FSD, that completely trashes the 5 SAE level approach altogether...

There really should just be 2 categories, ADAS and Self Driving (autonomous).

That makes the problem even worse IMO because it says nothing about the ODD. How do you distinguish between self-driving that only works on highways, self-driving that only works on city streets, self-driving that only works in Los Angeles, self-driving that only works on rural roads, self-driving that only works during the day, self-driving that only works during sunny weather, etc...? At least the SAE levels make a distinction between "limited ODD" and "unlimited ODD".

I get that Krafcik feels that the only true metric should be whether a system can handle all the driving without a human or not but I feel like you do need to specific the ODD. The ODD is also important. If you feel that L4 is bad because it does not specify the ODD then essentially combining L3, L4 and L5 into the same category is even worse.

For example, let's say Tesla does do autonomous driving that works on all roads in the US and Waymo has autonomous driving that is only available in certain geofenced areas. Both would fall under the same category of "autonomous driving" when they are clearly not the same.

Better ODD would be "all paved roads" + "in all non 'Weather Warning' conditions" as issued by Active Alerts

That will miss a lot of possible ODDs. What about self-driving that only works on I95 or only works on city streets or is geofenced to just one particular city or neighborhood or only works during the day etc...?

 

[Snuffysasa]

This is exactly the problem with L4.

Wrong. This is not a problem with L4 def, not a problem with SAE levels at all.

This is a problem with you (and many many others) using SAE levels for purposes they are not meant to be used for.

Yes L4 could mean a single closed off 1km stretch road at 5mph on Fridays at 2-3pm.... And there is nothing wrong with that.

You are trying to use SAE levels to define/measure the capability of a system or how technically advanced a system is, but that is inappropriate and not the purpose of SAE levels.

 

[Snuffysasa]

That makes the problem even worse IMO because it says nothing about the ODD. How do you distinguish between self-driving that only works on highways, self-driving that only works on city streets, self-driving that only works in Los Angeles, self-driving that only works on rural roads, self-driving that only works during the day, self-driving that only works during sunny weather, etc...? At least the SAE levels make a distinction between "limited ODD" and "unlimited ODD".

You don't use SAE levels for that. Use some other terms, maybe these terms do not exist, and maybe you see a need to create them. But you should not try to modify SAE levels for this purpose.

 

[diplomat33]

You don't use SAE levels for that. Use some other terms, maybe these terms do not exist, and maybe you see a need to create them. But you should not try to modify SAE levels for this purpose.

I am not suggesting that you modify the SAE levels. The whole idea of the SAE levels is that they categorize FSD based on who is responsible for driving with a distinction between a limited ODD and an unlimited ODD but purposely avoid getting into specific ODDs. Hence why L4 requires additional information about the ODD. I am fine with the SAE levels as they are.

I am pointing out the fallacy in saying L4 is bad because it covers too many ODDs while suggesting a better classification would be only 1 FSD "level" that would encompass even more ODDs.

 

[diplomat33]

That is why they should apply practical standards. That is how they create - for eg., USB standards. They specify in what conditions certain devices etc have to work, at what speed etc. You can't leave critical operating parameters like ODD to manufacturer discretion.

For eg., SAE (or its US affiliate) can say, FSD Level X is when
- It works in top 100 counties, by population
- In all weather conditions, except when certain enumerated emergencies occur

I get that people want "levels" that are more descriptive of how good the FSD is. They want to know if the FSD can work where they live. But as I mentioned before, the problem is that there are too many parameters for ODD.

SAE, p14 defines ODD as "Operating conditions under which a given driving automation system or feature thereof is specifically designed to function, including, but not limited to, environmental, geographical, and time-of-day restrictions, and/or the requisite presence or absence of certain traffic or roadway characteristics. "

IMO, it would not be practical to try to create standards that cover all possible parameters.

They could make a 10 step gradation that goes from simple CruiseControl to FSD in all conditions, everywhere 90% of people can safely drive.

That is an interesting idea. But I think coming up with 10 steps that cover everything and does not miss anything would be tricky.

 

[Silicon Desert]

A friend posted this photo on FB, it's in San Francisco. I'd dearly love to see the FSD beta do a u-turn from the second from the left lane.

Yep, this is the intersection of Mission with Van Ness in the northbound direction.
Many humans get confused themselves, so I wouldn’t be surprised if FSD messes it up.

I always thought that sign was a test to see if I was a democrat or a republican

 

[bobby g]

I always thought that sign was a test to see if I was a democrat or a republican

Surprised any option let's you go straight

 

[Silicon Desert]

Surprised any option let's you go straight

ha yes, that would be an "independent"

 

[run-the-joules]

A friend posted this photo on FB, it's in San Francisco. I'd dearly love to see the FSD beta do a u-turn from the second from the left lane.

Yet another reason to never go to that wretched hive of human feces and property crime. The only thing it had going for it were the restaurants and with COVID the good ones are mostly closed.

 

[Lindamon]

It's interesting that passenger aircraft have been able to do almost everything - take off, fly and land, since the 1960s. Certainly auto-landing was perfected and thought to be safer than a human over 50 years ago.

Yet we still like pilots to do it manually. Many accidents are down to pilot error, but we still seem to prefer them to automation.

Personally I hate being a passenger in a car, I'd much rather be driving. It's not a trust issue, I just get car sick if I'm trying to look at my phone too much and driving gives me something to do.

You have no idea what you are talking about, and I’m not particularly interested in correcting your misconceptions. Good day to you.