Timing of FSD poll

[Wicket]

I'm curious about general expectations people have for the timing of Tesla FSD tech. The timing might depend on precisely the nature of the system being judged a success but I'm going for approximately Level 4 (or 5) here, available as an expensive option or sub to buyers of the vehicle, in a fairly wide geographic expanse (at least 1 full state), and over 99% of driving scenarios (or 100% of most people's typical use). Corner cases like driving in a blizzard can be ignored, or driving in Mumbai. The car can still have a driving configuration for handling tough cases but it should be almost never used (plausibly a license is still required).

For people driving in the territories the tech is allowed, they should generally agree that 'I never have to drive unless I want to'.

Additional thoughts welcomed.

 

[KJD]

In 2018 Elon Musk said that my Model 3 would turn itself into a robo-taxi and increase in value.

Still waiting.

 

[DarthPierce]

I have been saying to unbelieving friends and family since she was born in 2015 that my daughter wouldn't need a driver's license unless she wanted it as a hobby. So march 2031 or bust!

 

[Tam]

...Additional thoughts...

Elon Musk already answered that question: Autonomous driving would start 2 years ago in 2019 and by 2020 or last year, the streets would be flooded with robotaxis all over.

 

[Wicket]

Elon Musk already answered that question: Autonomous driving would start 2 years ago in 2019 and by 2020 or last year, the streets would be flooded with robotaxis all over.

I'm curious why you think you wrote this comment. Clearly it doesn't answer the question.

 

[Tam]

I'm curious why you think you wrote this comment. Clearly it doesn't answer the question.

The answer has been answered. Because it was answered by Elon Musk in 2019 but it didn't happen in the past, so based on its past performance, the current answer is "Never."

L4 has been achieved by Waymo with LIDAR. Without LIDAR, I doubt Tesla will ever achieve L4.

Pure Vision still has to solve basic safety issues such as collisions, curves of deaths...

 

[Wicket]

The answer has been answered. Because it was answered by Elon Musk in 2019 but it didn't happen in the past, so based on its past performance, the current answer is "Never."

L4 has been achieved by Waymo with LIDAR. Without LIDAR, I doubt Tesla will ever achieve L4.

Pure Vision still has to solve basic safety issues such as collisions, curves of deaths...

I assume because Waymo has solved l4 they are now expanding as fast as possible to capture the trillions of dollars that market is worth.

And why do you think that vision is unable to eventually catch up to this solved problem that waymo is now pulling down cash and market cap for solving already?

 

[DriveMe]

Here is my thinking:

• We’ll need a year or so of wide FSD Beta to collect enough training data from the field
• We’ll need Dojo to come on-line to help solve most edge cases and to speed up the training cycle
• We may or may not need HW 4.0 in the cars (TBD)
• We’ll need a year or more to convince the regulators

I’d say 2024, the very earliest for a wide Level 4 release. Perhaps Level 3 in late 2022 - early 2023. That would be my guess.

 

[Tam]

I assume because Waymo has solved l4 they are now expanding as fast as possible to capture the trillions of dollars that market is worth.

Achieving L4 is one thing but achieving commercialization is another.

Waymo used to buy Velodyne LIDAR alone that cost about $75,000 at it looks like a Kentucky Fried Chicken bucket.

While waiting for the cost to come down, it's logical that although an average can't afford the price, commercial fleets can.

To review, L4 "can drive the vehicle under limited conditions and will not operate unless all required conditions are met" which is different from L5 that can drive in "all conditions".

To take advantage of the "limited conditions" of L4, Waymo has expanded its business by partnering with UPS, Daimler Trucks, J.B. Hunt, Ryder, and Stellantis. So instead of being geofenced in a small area such as Chandler, AZ, it's geofenced along the preset routes, depots across the western states of the US.

And why do you think that vision is unable to eventually catch up to this solved problem that waymo is now pulling down cash and market cap for solving already?

People have been touting Tesla's advantages of the vast amount of data collections from owners, AI, Dojo, Pure Vision... but the fact still remains that it has not solved the problem of basic safety issues. Collisions still happen from a very slow speed of summoning in a private garage to a smart summon in a parking lot. Collisions and fatalities still happen at a high speed as well. Look at the latest Youtube on the latest pure vision FSD beta: The system still makes frequent mistakes.