$3K extra for Full Self-Driving Capability could be a bad deal

[Haxster]

Why?

1. While the safety benefits of (the now really expensive) Enhanced Autopilot should be of value in the short term, it's unclear as to when Tesla will deploy the Level 5 full self-driving features that approach their aspirations.

2. It's entirely possible that the current hardware and upgrades will not be sufficient to do the job. You might need to buy a new Tesla with "Enhanced" Self-Driving Capability.

3. If Tesla IS able to provide compelling reasons to have the option, you're out a $1k penalty for the upgrade (unless they change this)... if you still even own the car. If not, you've saved $3K for functionality that you may never use.

It's unknown where Tesla will draw the line between self-driving-like features that they enable on Enhanced Autopilot cars vs Self-Driving Enabled cars. For example, they could decide that auto park works with autopilot, but auto supercharge only works with the self-driving option.

Time will tell.

 

[Chopr147]

If I were to buy now I would probably skip the $3000 "FULL" option. Lease? Definitely skip it.
Elon says by the end of 2017. So, we know it won't happen until at least 2018
Then, legal mumbo-jumbo to legalize it. 1 person dies by abusing the system can set back the regulations months. Not fair but it is what it is

It's entirely possible that the current hardware and upgrades will not be sufficient to do the job. You might need to buy a new Tesla with "Enhanced" Self-Driving Capability.

Who knows. 3 years from now Tesla decides Lidar is the future and changes all hardware to currently built models.
I hope not but I have a feeling radar will be like analog and lidar-digital sometime in the future.

 

[Omer]

If I were to buy now I would probably skip the $3000 "FULL" option. Lease? Definitely skip it.
Elon says by the end of 2017. So, we know it won't happen until at least 2018
Then, legal mumbo-jumbo to legalize it. 1 person dies by abusing the system can set back the regulations months. Not fair but it is what it is

Who knows. 3 years from now Tesla decides Lidar is the future and changes all hardware to currently built models.
I hope not but I have a feeling radar will be like analog and lidar-digital sometime in the future.

Yeah when Elon says radar is better than lidar, he's referring to today's metrics. In a few years time, solid state lidar might be significantly cheaper than radar and far exceed it in specs.

 

[rog]

I have mentioned something similar in another thread (and got some "love" for it ) - for sure the $3000 investment on the "full autonomy" part at this point is purely speculative. It's a bold assumption that Tesla makes on something they think they can implement with software only...

 

[Waiting4M3]

By the time the current AP2 HW is obsolete, you may not even need to worry about owning a Tesla and get stuck with the HW any more, just summon the closest one in your neighborhood, and let Tesla manage maintaining and upgrading the cars in their fleet

 

[tliving]

They may delay/not deliver fully on the promises, but one way to look at it is paying for extra hardware that will likely be put to more use in the future (and won't be able to be upgraded post-delivery). I'd opt for it if I were buying now knowing it will be a long time before the value was realized.

 

[Gizmotoy]

They may delay/not deliver fully on the promises, but one way to look at it is paying for extra hardware that will likely be put to more use in the future (and won't be able to be upgraded post-delivery). I'd opt for it if I were buying now knowing it will be a long time before the value was realized.

Maybe I'm missing your point, but it IS upgradable post delivery.

This is one feature it might be wise to wait on. Even once it works, who knows how long it'll take to get this past regulators.

 

[MarkS22]

Well, he did say "Level 5" which is a defined level of autonomy.

However, it brings up a great point. There's "twice as good as a human driver" but it doesn't end there. In two years, will we see Flir (to detect animals and pedestrians more clearly at the side of the road)? Will the cameras have even better dynamic range or higher resolutions? Will they go to 10 cameras? Will DrivePX3 offer 3 times the computational performance of DrivePX2? Will they add two more radars for better weather driving?

The bottom line is that it's great to have a confirmed suite to get to the minimum requirements of Level 5, but there will always be safety improvements from there.

 

[calisnow]

Well, he did say "Level 5" which is a defined level of autonomy.

However, it brings up a great point. There's "twice as good as a human driver" but it doesn't end there. In two years, will we see Flir (to detect animals and pedestrians more clearly at the side of the road)? Will the cameras have even better dynamic range or higher resolutions? Will they go to 10 cameras? Will DrivePX3 offer 3 times the computational performance of DrivePX2? Will they add two more radars for better weather driving?

The bottom line is that it's great to have a confirmed suite to get to the minimum requirements of Level 5, but there will always be safety improvements from there.

He said Level 5 - he used the phrase? I have been too lazy to listen to the announcement call. Also - how long do you give the current hardware suite @MarkS22? I project 24 months before the next round of sensor and computer upgrades.

 

[MarkS22]

He said Level 5 - he used the phrase? I have been too lazy to listen to the announcement call. Also - how long do you give the current hardware suite @MarkS22? I project 24 months before the next round of sensor and computer upgrades.

InsideEVs said:
"Tesla’s Elon Musk says the added equipment will allow for full level 5 autonomy at some point"

And, yes, it looks like they're settling into a 24 month product cycle. Based on a two samples, but it makes sense.

 

[xborg]

I agree.
By the time they turn on this feature, there will be AP 3.0 which will be more capable.
Paying $3000 now and hoping to save $1000 is not a good bet.

 

[AmpedRealtor]

Having only one radar unit (forward facing) stuck out at me during the announcement. I would think that in order to fully remove the driver from the equation, one would want rear and side facing radars. As Elon is on record saying, radar is more reliable than video cameras. Wouldn't you want your car to know that it was about to be rear-ended or t-boned? There would be some advantages to an autonomous vehicle having this information. First, it could activate safety features like seatbelt pre-tensioners ahead of a side or rear collision. It could also take evasive action to avoid a collision you didn't even know was going to happen. The car should compute the speed and/or rate of deceleration of the car behind you, and when the Model S determines that the vehicle behind you cannot slow in time and will hit you, it can at least warn you a few seconds in advance or get out of the way.

My primary concern with paying for the Self-Drive is that we have no way of knowing, outside of Tesla's promise, that the new hardware suite is sufficient for self-drive. As we have learned from past experience, Tesla doesn't have an aversion to breaking promises. Elon has already broken several promises relating to Autopilot 1.0. His description of how summon would work on private property never came to be. He said that it will eventually be able to read street signs and traffic lights, yet that never came to be. Maybe these things are still coming?

 

[Ohji]

Nvidia is slated to release Xavier in Q4 of 2017. It will have 20 TOPS of performance at 20 watts of power consumption. Drive PX2 is up to 24 DL TOPS at 250 watts. I don't know if the "DL TOPS" (Deep Learning Trillion Operations per second) referenced for Drive PX2 directly correlates to the 20 TOPS mentioned for Xavier. But at 20 watts, one should be able to equip an automobile with multiple Xavier autonomous driving modules, increasing performance and providing redudancy -- all at a fraction of the power necessary with Drive PX2.

Introducing NVIDIA Xavier | NVIDIA Blog

Elon did say that the processor might be upgradeable with AP 2.0, as they made it fairly easy to access...

Even bigger gains should occur as Nvidia/Intel/AMD move to smaller nodes (10nm, 7nm).

 

[tliving]

Maybe I'm missing your point, but it IS upgradable post delivery.

This is one feature it might be wise to wait on. Even once it works, who knows how long it'll take to get this past regulators.

Oh, I missed that point. I thought the 4 cameras to 8 cameras part in this was talking about a hardware difference. But then I looked again and see that it is upgradable post-delivery, so I guess that means they all come with the 8 cameras?

 

[gearchruncher]

My DS today didn't even try and sell me FSD. It was more like "oh, yeah, if you want that it will be $3,000 now or only $4K later". Like he understood it was so out there that hardly anyone is going for it yet.

 

[Dstrohl]

Lets also be clear that while Elon said Level 5, the official description doesn't say anything about L5, and could fit L4 almost as well.

Also, the talk about the drive automous drive from CA to NY next year was not a commitment to deliver the product then, a demo drive with a factory car would meet his goals just as well, leaving actual product delivery to happen long after that.

While I could see the technology working in most cases by next year, I really don't see the legislative approvals happening by then.

 

[calisnow]

...He said that it will eventually be able to read street signs and traffic lights, yet that never came to be. Maybe these things are still coming?

I remember reading that the ability to read stop signs and hundreds of other street signs around the world is already baked into Mobileye's EyeQ3 but that Tesla made a decision not to enable it at the launch of 7.0 for some reason.

 

[MarkS22]

Having only one radar unit (forward facing) stuck out at me during the announcement. I would think that in order to fully remove the driver from the equation, one would want rear and side facing radars. As Elon is on record saying, radar is more reliable than video cameras. Wouldn't you want your car to know that it was about to be rear-ended or t-boned? There would be some advantages to an autonomous vehicle having this information. First, it could activate safety features like seatbelt pre-tensioners ahead of a side or rear collision. It could also take evasive action to avoid a collision you didn't even know was going to happen. The car should compute the speed and/or rate of deceleration of the car behind you, and when the Model S determines that the vehicle behind you cannot slow in time and will hit you, it can at least warn you a few seconds in advance or get out of the way.

I think we'll also be due for more innovative autonomous interiors. The fact is, the Model S interior is designed for driving. It would be considerably safer, for example, to allow the front seats to rotate so they're facing backwards. (Scary as hell for most of us, but still safer.) That's not even taking into account comfort.

Making consumers feel less vulnerable will help with autonomous adoption. With all those sensors and so much room in the frunk (thanks to no ICE), what about massive external front airbags? Something along these lines: External Airbags / Youtube. Apparently they can reduce force by 30%. When you have reliable sensors and known actions of a computer, you can start to deploy some unique safety mechanisms reliably. The instant reaction times of a computer combined with a 30% reduction in force (plus rear facing front seats) could do incredible things for safety.

 

[Joeker88]

Thank you so much for the info. I went ahead and changed my order to just AP 2.0 since the full driving capability won't be ready anytime soon. If it does I'll pay then.

 

[irvineboy]

So what is the difference between this fully autonomous $3000 option vs the enhanced autopilot that costs $4000? I thought the enhanced AP already lets the car drive itself??