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

Tesla promoting self-driving video was staged

This site may earn commission on affiliate links.
What an absolute debacle. Much more will come, just wait.

Turns out this actually was the path to Level 4/5 all along, because I have a feeling Tesla will end up owning liability for the DDT in many of these accidents. Take that SAE.
 
Last edited:
Fm3L5qTaEAAQTFa

Fm3Mn2gakAAZwYA


I do think the legal issue comes from Tesla selling FSD when they knew they had staged a deceptive demo. That opens up Tesla to fraud claims. The fact that Elon told them they would just backfill with production code later shows that his plan was to sell vaporware and just "figure it out" later.

If Tesla had not sold FSD in 2016 and been honest that it was a concept demo of a future product, I think Tesla would have been fine.

Seems like he worded it carefully to his staff so they wouldn't be uncomfortable working on and then he re-framed it to the target audience the way he planned to the whole time.

The whole episode sounds an awful lot like a few recent examples I can think of.
 
No. What Tesla did in the demo is not what Waymo does. Waymo uses HD maps as a prior. They don't drive just by following premapped routes. Waymo uses camera vision, lidar and radar to drive in real-time, on the fly. Also, Waymo did not stage a demo to sell something that did not exist yet. Waymo developed autonomous driving and only sold rides once they could actually do the rides with no human driver.
I don't see how that's any different. The static problem is solved by localizing ego onto HD maps, the dynamic problem (cars, pedestrians, etc) is solved with sensors. From Ashok's statement that's sounds awfully similar to what Tesla did on their initial demo. They pre-mapped the route into 3D so the car didn't need a working BEVNet.

Waymo creates demos to sell to investors, not customers, they don't have any customers. Tesla does the same, but at the same time also sells intermediate deliverables to customers. One makes money, the other does not.
 
I don't see how that's any different. The static problem is solved by localizing ego onto HD maps, the dynamic problem (cars, pedestrians, etc) is solved with sensors. From Ashok's statement that's sounds awfully similar to what Tesla did on their initial demo. They pre-mapped the route into 3D so the car didn't need a working BEVNet.

The difference is that Waymo does have a working and complete BEVNet. Waymo does not pre-map to compensate for a lack of BEVNet. Tesla did. The difference is how they use HD maps.

Waymo creates demos to sell to investors, not customers, they don't have any customers. Tesla does the same, but at the same time also sells intermediate deliverables to customers. One makes money, the other does not.

Wrong. Waymo does have customers. Waymo is doing paid driverless rides to the public.
 
It sounds like you're confusing them. Tesla used FSD demos to sell investors and do capital raises before the company became profitable, they gave investors rides in the thing long before there was any FSD beta available to the public.
The first video was released the same day they started selling FSD to customers. October 19th, 2016
 
The video was released the same day they started selling FSD to customers. October 19th, 2016

Like I said, Tesla was selling to investors and customers alike. Customers can pay to get intermediate deliverables on some totally non-committal time schedule, and at the same time Tesla was trying to sell the future potential of FSD to investors in order to raise more capital.
 
The difference is that Waymo does have a working and complete BEVNet. Waymo does not pre-map to compensate for a lack of BEVNet. Tesla did. The difference is how they use HD maps.

I think you're confused. BEVNet isn't some industry-standard term, it's not used outside of Tesla, it's just something the Autopilot Team at Tesla dreamed up to help solve the static driving problem. That model was what enabled them to go from this dodgy demo video to something shipable.

Wrong. Waymo does have customers. Waymo is doing paid driverless rides to the public.

Good one! I guess I'll go install the app and get rid of Uber and Lyft now.
 
I think you're confused. BEVNet isn't some industry-standard term, it's not used outside of Tesla, it's just something the Autopilot Team at Tesla dreamed up to help solve the static driving problem. That model was what enabled them to go from this dodgy demo video to something shipable.

In your own words, Tesla used HD maps to handle the static part because they lacked a complete perception stack. My point is that Waymo did not do that. Waymo does have a full and complete perception stack, and a complete prediction and planning stack too.
 
I’m sure that will be Tesla’s argument in court in the various cases underway right now, and we’ll see how it goes over.

Tesla was already sued over the video itself in 2017, a class action that resulted in a $5.4m payout across owners. You can find YouTube videos of people getting their little cheques.

What’s coming to light right now is part of a damages suit from the family of Walter Huang, whose vehicle crashed while on Autopilot. There’s a separate class action for deceptive marketing of FSD since 2016, and then two other Autopilot fatality cases right now.

Theres also a phantom braking lawsuit, and then a lawsuit from a Texas police department where five members were injured when a Tesla on autopilot slammed into a parked cruiser while they were performing a tradfic stop I think — this is what led to the emergency responder hubbub from the OTA update to the NHTSA investigation.

All of them will use this as part of their cases in slightly different contexts.
 
I think you're confused. BEVNet isn't some industry-standard term, it's not used outside of Tesla, it's just something the Autopilot Team at Tesla dreamed up to help solve the static driving problem. That model was what enabled them to go from this dodgy demo video to something shipable.
Here's a paper demonstrating a Bird's Eye View network from 2017. I don't think it's something that Tesla invented.
Here's Waymo's paper: https://arxiv.org/pdf/2209.05324.pdf
 
At the time of the 2016 video, there was only AP available for actual use, and those were limited to highways. So I assumed the video was showing future functionality of getting AP working on local roads.

In addition, when that video was released, it actually got a lot of scrutiny, mainly because it looked impressive at a glance. Several youtubers reposted the video slowed down to actual speed, which made it really obvious that the car actually didn't have a good run (lots of phantom braking and going very slow in some spots). Others noticed there were some cuts in the footage that looked like it was done to intentionally hide a flaw.

Later when it was revealed that it was staged, and all the really bad stuff was edited out, it did not surprise me in the least.

I can understand some people never seeing some of this resulting analysis and dialog at the time, but for me, I was on the fence about buying the FSD addon, and with some due digiligence, I decided that the functionality shown in the video wasn't actually good enough for me to commit to the purchase. I'm not saying the lawsuit is without merit; I think Tesla/Musk need to stop being over-optimistic about major functionality like this.

But I just wanted to point out that there were observable warning signs even back in 2016 that local road AP (FSD or whatever) was in rough shape.
 
  • Like
Reactions: diplomat33
It’s the promotional video for FSD Hardware which came out when they started selling the first version of FSD Capability in 2016 (see the capture from the website above). FSD has never been an option, only FSD Capability (which is not capable of FSD! The promotional video is just showing that the hardware is capable, not the software)
Actually the hardware was not capable either. I had to pay $1000 to upgrade the hardware in my 2018 so that I could subscribe to FSD. So basically nothing was capable in 2016.
 
I’m one of those owners. Based on Tesla’s lies, I believed full FSD was imminent in 2016
I leased my first Tesla in May 2018. The salesman tried very hard to convince me to add FSD to a 3 year lease, telling me it would be fully functional by the end of the year, since it just needed final testing and regulatory approval, and then the price for FSD would sky rocket. So even the sales people were convinced or lying. I ended up only adding EAP, however the customer next to me believed the sales person and told me I would be sorry. I was not!

I did end up buying out the lease in 2021, then subscribed to FSD, since it was now finally available, a few months after the initial lease expired.
 
This was only a secret to fanboys, who probably still believe FSD is just around the corner, ready to go, just pending government regulations. 3 months maybe, 6 months for sure!

Those who can think rationally, called it out as a fake shortly after it came out.
It was staged??? OMG .. next you'll tell me the pictures on frozen food boxes aren't representative of the contents!
 
I can admit that I was duped by FSD in the beginning. In 2016, I believed the video was proof that Teslas were or would soon be self-driving. It was a big reason why I became interested in Tesla and wanted to buy a Tesla. Back then I was a huge Tesla FSD supporter. I used to post here about how Tesla was going to solve FSD any day now. LOL. It was only after Autonomy Day in 2019 that I started to see that FSD was not going to happen.