MCU 3

[Tes La Ferrari]

Anyone think this is on the way soon?

or do you think it will form part of the new interior that is supposed to be very soon to be revealed (in Q2 I think)

 

[Naonak]

MCU3 and AP3 likely go hand-in-hand is my guess.

AP3 is what will deliver on the promises of AP2. Although, AP2/2.5 never did deliver on the promises of AP1, so it's anyones guess at this point.

 

[ucmndd]

MCU3 and AP3 likely go hand-in-hand is my guess.

Why? They've had basically zero interdependency to date.

MCU 2 is barely a year old. I don't think anything changes until the Model S refresh, if/whenever that happens.

 

[Derek Kessler]

Someday, but probably not for at least another year or two. The current Intel chipset is more than enough to handle everything the MCU needs to do and then some.

 

[Naonak]

Why? They've had basically zero interdependency to date.

MCU 2 is barely a year old. I don't think anything changes until the Model S refresh, if/whenever that happens.

Because it's costly to maintain multiple discrete systems in a vehicle. If MCU3 can handle all the in-car duties as well as AP, they will do that from a cost standpoint. As said above, MCU2 is more than adequate for the moment, so we won't see an MCU3 any time in the near-term, just like we won't see AP3 any time in the near term.

I suspect AP3 and MCU3 will be part of the same package. The "upgrade" that AP2.5 people are getting won't be a true AP3, as I'm sure some hardware is going to be missing (SSLIDAR, among others) that will comprise the AP3 package. The current AP2.5 hardware is not adequate to handle autonomous driving in anything but pristine conditions. As we all know, most driving is not under pristine conditions. It will require handling virtually all edge cases through some sort of learning algorithm and sensors that aren't available on any Tesla at the current time. *IF* we see AP3 in the near-term, it will not be capable of autonomous driving, which is what everyone wants FSD to be, and it won't be that. We may have something called FSD, but it's not what people are envisioning. Not on current hardware.

Just a quick example... how do you differentiate a wheel bending/flat tire inducing pothole from a mark in the road or a shadow visually? Sometimes you can't. You need hardware that is not visually based for that. The car *WILL* have to handle that.

 

[darxsys]

Because it's costly to maintain multiple discrete systems in a vehicle. If MCU3 can handle all the in-car duties as well as AP, they will do that from a cost standpoint. As said above, MCU2 is more than adequate for the moment, so we won't see an MCU3 any time in the near-term, just like we won't see AP3 any time in the near term.

I suspect AP3 and MCU3 will be part of the same package. The "upgrade" that AP2.5 people are getting won't be a true AP3, as I'm sure some hardware is going to be missing (SSLIDAR, among others) that will comprise the AP3 package. The current AP2.5 hardware is not adequate to handle autonomous driving in anything but pristine conditions. As we all know, most driving is not under pristine conditions. It will require handling virtually all edge cases through some sort of learning algorithm and sensors that aren't available on any Tesla at the current time. *IF* we see AP3 in the near-term, it will not be capable of autonomous driving, which is what everyone wants FSD to be, and it won't be that. We may have something called FSD, but it's not what people are envisioning. Not on current hardware.

Just a quick example... how do you differentiate a wheel bending/flat tire inducing pothole from a mark in the road or a shadow visually? Sometimes you can't. You need hardware that is not visually based for that. The car *WILL* have to handle that.

Agreed. FSD as a name is totally misleading. Proper FSD is 10 years away at least, no matter what Elon says. Listen to actual machine learning/self-driving experts.

Plus this: Tesla has a self-driving strategy other companies abandoned years ago

I feel bad for everyone falling for these statements about full self driving by end of year and other stuff that came out of Elon's mouth regarding this topic.

 

[AeroDepartment]

Agreed. FSD as a name is totally misleading. Proper FSD is 10 years away at least, no matter what Elon says. Listen to actual machine learning/self-driving experts.

Plus this: Tesla has a self-driving strategy other companies abandoned years ago

I feel bad for everyone falling for these statements about full self driving by end of year and other stuff that came out of Elon's mouth regarding this topic.

FSD is definitely possible with existing sensor suite (8 camera, 12 ultrasonic, front radar) the software and the computer on board would definitely need to be HW3. There really isn’t any “expert” on the subject of self driving and battery technology. Like seriously?! There are people calling themselves “experts” when they’re not even an engineer or scientist who worked in that particular field. How can one be an expert in a field that is so very new? There are business analysts trying to dabble in the tech sector and they have some nerve to mess with fields completely unrelated to their own. Why can’t you all just leave us engineers and scientists alone and let us do our job and not disrespect our work. It’s like going to a doctor and telling them that you’re a doctor or “expert” now just because you know how to use Mayo Clinic and google.