First Jerry rig everything, now mark rober 🧐

[freyguy]

TLDR:
Jerry rig everything and Mark Rober have both criticized Tesla's technology, with Rober downplaying its camera vision compared to LiDAR, leading to disappointment over their behavior and possible intentional oversight in testing.


In case you missed it Jerry rig everything put the trailer hitch through hell of the cybertruck trying to really throw dirt on the Tesla name. He’s a relatively big channel, but now mark rober who has way many followers 65m and decides to try and make Tesla camera vision less superior than LiDAR. I can’t believe there is so much going on with these platforms downplaying the amount of tech and engineering that Tesla has developed over the years. I’m actually ashamed of both channels for this behavior, whether it be for clicks or views it’s complete bs. Additionally during his testing he was only using autopilot and not FSD. I do t wanna say this was intentional but he’s a nasa engineer so something like that would never be an oversight.

Sponsored

 

[Travis64]

Should of kept lidar

 

[freyguy]

Should of kept lidar

they never had it as far as I’m aware.

Before Tesla fully shifted to Tesla Vision with just cameras, their cars used a mix of sensors. They had radar in the front for distance and speed detection, ultrasonic sensors around the body for close-range stuff like parking, and of course, the cameras. The big change came when they ditched radar in 2021 for new models, betting everything on the camera system.

 

[TexasRaider]

Vision alone can only go so far. Aka in ideal conditions. The reason why us humans slow down or don’t drive in certain conditions, ie blizzards or sand storms.

I can’t wrap my brain around why a manufacturer would choose only one sensing technology instead of a network of different types for different weather/road conditions.

Fact: I’m not an engineer.

 

[CyberGus]

Even humans can be fooled by optical illusions.

 

[Limes2]

they never had it as far as I’m aware.

Before Tesla fully shifted to Tesla Vision with just cameras, their cars used a mix of sensors. They had radar in the front for distance and speed detection, ultrasonic sensors around the body for close-range stuff like parking, and of course, the cameras. The big change came when they ditched radar in 2021 for new models, betting everything on the camera system.

And Tesla added radar back in specific to the HW4-equipped Model S and Model X, starting in early 2023. This is the custom design HD radar that they wanted to experiment with and see if it helped FSD.

The new Model 3, Y and CT do not have this radar. Maybe it didn't provide any benefits? Not sure but it is interesting to me that Tesla is willing to try new stuff and put it into production to gather real world data.

 

[Outdoors]

Many have and will pile on. It is human nature to join the crowd. Elon does have a pretty good track record of success. Often that success comes when the mindset of many are drifting from one idea to another. Elon is focused even when he appears not so.

Till then let people be the herd.

Buffalo Jump

 

[REM]

I unsubscribed soon as I made my way through this hit piece.

It's funny how these are coming with more frequency after DOGE started dusting off old logbooks in high places.

 

[REM]

Should of kept lidar

Why do you think this? Technical explanations included.

 

[REM]

Vision alone can only go so far. Aka in ideal conditions. The reason why us humans slow down or don’t drive in certain conditions, ie blizzards or sand storms.

I can’t wrap my brain around why a manufacturer would choose only one sensing technology instead of a network of different types for different weather/road conditions.

Fact: I’m not an engineer.

People seem to think slapping on more sensors = more accuracy or better data; but that's not how it works.

Lidar introduces a tremendous amount of noise in a dynamic system. Tesla is training an AI model, and it must mimic our way of navigating the world so that we understand it, and it understand us.

 

[cbr1000]

Rober popped up in my recommended videos yesterday. I didn't watch it... I could see he is placing a massive mirror on the road. As long as someone isn't moving a giant wall in the middle of the road and painting a photorealistic scene, or moving a 50 ft mirror on the road I feel pretty safe.

Everything is testable to extreme for click bait. I wish YouTube would cut back paying these people so much - it encourages videos for the purpose of making money only versus the sharing of knowledge, or good entertainment.

 

[jameskk]

Vision alone can only go so far. Aka in ideal conditions. The reason why us humans slow down or don’t drive in certain conditions, ie blizzards or sand storms.

I can’t wrap my brain around why a manufacturer would choose only one sensing technology instead of a network of different types for different weather/road conditions.

Fact: I’m not an engineer.

because of the noise it creates in the data stream. Tesla vision, using raw photo reads, can simulate lidar with depth perception exceeding humans. Tesla regularly validates this data using cars with actual lidar. this is what makes Tesla Vision so magical. They spent years developing and getting this right, and it is getting better with ai.

 

[Derektsla]

People seem to think slapping on more sensors = more actually or better data but that's not how it works.

Lidar introduces a tremendous amount of noise in a dynamic system. Tesla is training an AI model, and it must mimic our way of navigating the world so that we understand it.

that’s correct. As far as I understand, there are a few reasons the Lidar went away.

1. Cost, LiDAR has gotten cheaper, but it’s still costly compared to cameras.

2. LiDAR provides a specific sort of data, but I believe it made it harder to train the FSD models. If the LiDAR says there is an obstacle, but vision says there is not, which one is right. (Spoiler, it’s NOT always LiDAR). Vision provides more context I believe. It can see turn signals, brake lights, type of object, etc. This data can be used to anticipate behavior that LiDAR can’t. I believe LiDAR was intended to validate what vision was seeing, but maybe it wasn’t enough to justify the cost?

3. I don’t think the lidar performed as well in bad conditions as people thought. I remember reading that the noise introduced by harsh snow or rain made it harder to make decisions. (This is my speculation, I have seen it help in foggy conditions, but not sure about rain, etc).

I am an engineer, but I am NOT an expert on this stuff.

 

[Sjohnson20]

I seem to remember the radar was being blamed for the phantom braking. It was detecting objects like overpasses and thinking they were obstacles. So they went with the cameras. Lidar was too expensive at the time. Maybe they will have to end up adding some kind of lidar or other sensor to make it unsupervised.

 

[Cybertruck2024]

I am not sure if the vehicle needs LIDAR, but every camera needs to be self cleaning. They are doing the cybercab testing in Austin for a reason, the weather allows it. In places it snows regularly, you would have to shut down the fleet for 4+ months a year.

I personally use FSD about 70% of the time in the summer and 10% of the time in the winter. Cameras are occluded, speed is limited. You genuinely would need to pull over every 30 seconds in the winter to keep the cameras clean, FSD is unusable in bad weather conditions.

Sponsored