"FSD is a few months away" for Cybertruck, says Elon Musk (5/9/24)

[CyberGus]

Can I volunteer to be a Beta tester

Sponsored

 

[Giskard]

Couldn't CTs at least get Autosteer portion of Autopilot, for freeways?? That’s basic Mobileye stuff on my 2016 MS, FFS.

Just finished SLC-Tahoe-SLC, prepping for UT-VT-UT in late June (which I do every summer), and actually prefer AP over FSD for long freeway drives. I may choose the Rivian for this summers trip if no AP and we have a NACS adapter by then.

Basic AP/NOA would be fine for CT until they’re ready to roll out FSD beta (in my ‘18 MX) needed a lot of supervision with lane changing and other odd behaviors on my 2023 XC trip, does that weird increasing follow-distance creep behavior that is more aggravating than relaxing (even on Mad Max). And Mad Max setting makes it change lanes more often, requiring extra brain cycles and more stress than just handling all driving myself, when all I really want from that setting is to stay in the respectful nether reaches of a Semi draft.

Since you mention rear-steer as an programming obstacle (good point), the engineers should simply disable Rear-steer for Basic AP on the freeway, and just port the fleet AP to the truck (recalibrated for CT camera locations). That would satisfy the majority of desired use-case for now, I would think? I wouldn’t use FSD in a metro anyway…only reason I’ve tried it periodically is to gauge progress of the FSD team. Besides, with edge-to-edge ML now, we should be teaching the truck how to drive anyway!

My guess is it isn't significantly easier to retrain AP to account for rear steering than it would be for FSD. I suspect we'll get them both around the same time. Locking the rear steering out for highway AP is an interesting idea, though. Like you said I imagine most would be happy to make that compromise in the short term. I quite like FSD, but highway AP is a game changer for longer trips.

Just to clarify, too, I do not work for Tesla nor do I have any kind of insider knowledge about what's going on. I just follow posters on X (and to a lesser extent YouTube) that talk about stuff like this. They're usually pretty accurate, but not always spot on. I also work in the software industry so I have a feel for what it takes to delivery features like this - you can't just throw warm bodies at it.

 

[Jhodgesatmb]

Just need AP.

FSD is a waste of money

Finally someone shared their true position on this Now I can just ignore anything you say on the topic of FSD.

 

[heems]

The steering being the different mechanism makes sense to me. Any steering will need new model / testing. So even AP / lane keep will not happen cause it needs to steer. Forget auto park, lane change and FSD. Although I assume once steering is abstracted the other pieces that need steering will fall into place quickly.

 

[Sjohnson20]

Autopilot would be nice. I don’t trust fsd especially with all the crazy drivers trying to take cybertruck pics.

 

[Stuck4ger]

Just need AP.

FSD is a waste of money

That's your opinion. I'm curious as to how much time or road miles you've spent with FSD engaged. I've done thousands, and it's definitely a huge driver relief mode and safer overall. Yes, I've stopped it from doing things that I thought were about to go very bad, I've stopped it from changing lanes because it didn't know what lane it really needed to be in (and I've been wrong about that several times). It's definitely way safer than manually driving shortly after dusk when deer are next to invisible alongside the road. Overall, I would use FSD almost all the time as I really like how it allows me to have my head on a swivel.

 

[Stuck4ger]

Autopilot would be nice. I don’t trust fsd especially with all the crazy drivers trying to take cybertruck pics.

I've had FSD slow down as well as speed up to avoid a lane infringement. When it speeds up it is often a surprise because unless you're spending all your time with the rear view and side mirrors, it's tough to realize the true threat of the traffic behind you (their distance and closure rate). FSD is continuously monitoring such.

 

[Darthamerica]

I think a lot of people would be happy with getting basic autopilot at this point.

CT is using different hardware from what the OG AP was built for. It wouldn’t be trivial to bring that functionality to CT. I think while it might be possible, it would likely cause further delays for FSD on CT.

 

[flamaest]

A few is more than a couple. Sounds like we are looking at end of year, if we factor in Elon time.

 

[webspeedracer]

yes so if rear-steer is the bottleneck, FSD engineering could disable rear steering on freeway driving and give us simple AP for long roadtrips. Wes Morrill’s CT engineering team has already given us a “disable rear-steer” for Baja mode, so the control mechanism is already in the software!

Seems like a low-overhead means of placating all of us FS-CT owners for the interim, until they can fully-bake an FSD that includes rear-steer for all speeds and all driving contexts.

The steering being the different mechanism makes sense to me. Any steering will need new model / testing. So even AP / lane keep will not happen cause it needs to steer. Forget auto park, lane change and FSD. Although I assume once steering is abstracted the other pieces that need steering will fall into place quickly.

 

[webspeedracer]

CT is using different hardware from what the OG AP was built for. It wouldn’t be trivial to bring that functionality to CT. I think while it might be possible, it would likely cause further delays for FSD on CT.

I would bet that Robotaxis FSD hardware will be same (or iterative version of same) as Cybertruck hardware, both for economies of scale and for programming simplicity (minus rear-steer).

So I would expect that porting FSD (Supervised) to Cybertruck will bear fruit in their most-important endeavor, FSD (Unsupervised) in Robotaxi.

 

[Stuck4ger]

yes so if rear-steer is the bottleneck, FSD engineering could disable rear steering on freeway driving and give us simple AP for long roadtrips. Wes Morrill’s CT engineering team has already given us a “disable rear-steer” for Baja mode, so the control mechanism is already in the software!

Seems like a low-overhead means of placating all of us FS-CT owners for the interim, until they can fully-bake an FSD that includes rear-steer for all speeds and all driving contexts.

If something would be that easy then Tesla would have done it. Usually us armchair designers don't truly understand all the constraints. I recently had somebody (a smart but non EV person) tell me that if the battery died you should just tow the Tesla and use regenerative braking to charge the battery. That's a surefire method of seriously damaging, if not destroying, your car.

 

[GatorCyber]

No FSD for Cybertruck for the first year makes no sense. On the last earnings call Elon and another dude confirmed that licensing FSD to other OEMs was basically plug and play. Sell them the license and hardware and flip a switch, and voila, FSD.
And no way 4-wheel steer matters. I guess Elon lied on the earnings call.
I don't think Tesla should be delaying FSD on Cybertruck any longer and should be honest as to why.

 

[Darthamerica]

I would bet that Robotaxis FSD hardware will be same (or iterative version of same) as Cybertruck hardware, both for economies of scale and for programming simplicity (minus rear-steer).

So I would expect that porting FSD (Supervised) to Cybertruck will bear fruit in their most-important endeavor, FSD (Unsupervised) in Robotaxi.

I have no idea what Robotaxi HW Will be based on. It could be all SW based for use on existing vehicles or it could be something new. But I agree that it’s likely CT and whatever Robotaxi is will share technology.

There are fewer than 10K CT out there now. If I was running Tesla I wouldn’t divert any resources from Robotaxi to CT at such low numbers. It would seem better to let CT FSD development parallel or piggyback on Robotaxi.

 

[rizvend]

$99 a month back until FSD is released might be nice or free supercharging. Til then. ? yeah dreaming on.

Sponsored