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

Tesla autopilot HW3

This site may earn commission on affiliate links.
Obligatory Swindon Magic Roundabout reference:
now-87bc897c-4eb4-4d2c-938d-38485e299587-1210-680.jpg
WTF. I wouldn’t even know where to start without studying the layout ahead of time.
 
Concerning no NoA in the UK, whether there is regulation holding it back;
In Germany atm, Model 3 has NoA, S/X do not. Can't be regulation.
Plus, as long as it's a L2 system, the only applicable international rule to base national regulation on is the Vienna convention which does not regulate software updates and level 2 assistance systems.
I highly doubt there is regulation regarding L2 systems, and please, if anyone claims the opposite, link to the regulation in text form.
 
Looks like the Model 3 MCU/AP computer in about the same place as the in S & X

Nice video. Yes I agree the location is similar to Model S/X APE though the latter have it higher up and horizontal above the glove compartment instead of behind it (their MCU is behind their screen though).

At 2:02 you can see the liquid cooling system that makes the Model 3 swap probably one step harder than Model S/X. In addition to it being integrated with the MCU-equivalent which probably needs to be swapped with the APE to keep the task at a reasonable length, which may then require additional software pairing and install.
 
  • Like
Reactions: croman
Obligatory Swindon Magic Roundabout reference:
now-87bc897c-4eb4-4d2c-938d-38485e299587-1210-680.jpg

I’ve seen this many times before but for the first time I think I get what they were trying to accomplish with it. I get the idea for using smaller roundabouts to feed a larger one to avoid queues or traffic lights. I also get the idea of using outer ring lanes to facilitate transfers from adjacent roads to the next... I’m just not quite convinced this exact configuration was the best way to go about it. :)

Certainly an interesting task for a self-driving car.
 
  • Like
Reactions: ironwaffle
If the power supply is on board the FSD unit, yes, if the redundant power supply is off the FSD board it's a valid question.
The power supply is on board. The revelent portion of the PCB was highlighted during the talk. One supply is near the top left, the other near the bottom left.
Yes, I’m asking because I do think the power is off-board. I don’t see how it could be onboard in such a small package, especially two of them.
The entire board only uses 75 watts, a 50 watt buck converter is trivial to design in minimal space. www.ti.com has a tool that will give you multiple options for such a thing.

Of course, the power source (12v battery) is off board with redundant feeds to the two on board converters.
 
Duh, of course there needed to be a typo in title that I totally missed and cannot edit apparently?
Just report the post and tell the moderators what you are requesting they do for you. If you do that infrequently enough, you'll find most moderators are often polite enough to do so for you. In this way, if you continue to force yourself to post correctly in the first place (proofread slowly many times -- this takes time), you will minimize your need to have moderators correct your mistakes to a level that they can probably handle. I used to think the time to proofread would take away from my ability to communicate effectively due to the high cost of proofreading. I've found that the expediency of not proofreading is not worth it if it gets close to the edit time limit. If I get close to the expiry of the edit time period (15 minutes away), I just delete the message and put it in a local editor and proofread for a few more hours then post. For important or long posts, it's not even worth posting in the first place until fully proofread. You could even make an off-site repository of a draft and post a link to the draft on TMC, saying you'll post the final version on TMC once you're done.
 
Last edited:
  • Funny
Reactions: lunitiks and bedoig
Of course, the power source (12v battery) is off board with redundant feeds to the two on board converters.

That's what I'm asking about. Do all Tesla models built from Oct 2016 to March 2019 include a redundant power source feed, or does that redundancy need to be added at the time of the FSD computer retrofit to satisfy the "fully-redundant system" requirements?
 
That's what I'm asking about. Do all Tesla models built from Oct 2016 to March 2019 include a redundant power source feed, or does that redundancy need to be added at the time of the FSD computer retrofit to satisfy the "fully-redundant system" requirements?

I believe in the presentation, it was said they already have redundant power and network feeds. I do not think it would be very feasible to retrofit that.
 
  • Informative
Reactions: croman and GSP
I think the kangaroo issue it that don't touch the ground when moving. So any system that relies on visible pavement or vertical positioning for depth estimation will think a mid air roo is further away than it really is.

radar will know the distance of the roo, doesn't matter if it's airborne or not so long as it's in the forward arc of the radar.
 
Yah, however if the roo is moving across the road (tangential to the car's motion), it will not show up well due to being at the the same speed relative to the car as the stationary objects. (Stopped vehicle problem)

It'd still show an azimuth shift on the radar return, and be recognizable that way.

However, at the Autonomous event Tesla talked about training the networks to correctly identify airborne cars as well as pedestrians, so I don't think an airborne kangaroo will be that much more difficult to train the cameras for.