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

FSD fails to detect children in the road

This site may earn commission on affiliate links.
There’s a difference between a one off article about a particular subject and a series of stories release on multiple platforms on a regular basis about multiple subjects. If you can’t see the difference, well… as they say, ignorance is bliss.
I see you want to prolong the irrelevant.

What about all the conservative outlets who were dead set against Tesla until now? What do you have to say?

Please, don’t be freakin tribal. Pollutes your sense of reasoning.
 
These tests prove nothing,
It just proves that what Tesla says is true. Tesla says that FSD Beta will hit children. There’s nothing else I can think of that would be worse, which is exactly what Tesla says will happen.

It just doesn’t seem that complicated. It’s a limited capability system that will hit stationary (and moving!) objects in many situations; you just have to find those situations.
 
Last edited:
  • Like
Reactions: DrGriz
The problem with the ad was: Its speed was set at 40MPH. Tesla's system did slow down to 22MPH and then gave controls to the driver at the moment of collisions.

NCAP night time was set at 60Km/H or 37.5MPH, but the big difference for that night was: The mannequin was traveling in the same direction as the car. Tesla has more problems with stationary (first responder NHTSA investigation) and cross-path obstacles (2016 Autopilot fatal accident). Still, its chance of stopping for obstacles traveling in its direction is vastly increased.

NCAP kid mannequin was tested at 30Km/hr or 19MPH. The slower the speed, the better chance for an AEB system to work. The NCAP speed was lower than the ad's speed of 22MPH when the system handed the control over to humans.

The ad must adhere to the above factors to prove that Tesla works.
I was under the impression that AEB only slowed the vehicle down something like 20-25 mph, but the OM doesn't seem to indicate as such. What it does mention is if AEB is occurring there is an icon displayed (along with an alert and sound) that indicates it is in progress.
 
Tam,

I have to note here that the 2016 accident was on Mobileye hardware and possibly even the software stack.

Tesla has since done their own hardware and software.

That's the whole point: Tesla collision issue is not solved no matter which technology Tesla got it from: MobilEye, non-MobilEye, radar, radarless, immatured technology in 2016, or 6 years after that...

MobilEye broke off with Tesla in 2016 because it wants no part in "...pushing the envelope in terms of safety."
 
Last edited:
  • Informative
Reactions: _Redshift_
Accidentally confirmed the results of the original test?

Whole Mars Catalog
@WholeMarsBlog

In my initial testing it’s been difficult to get the child sized dummy to read as a child. I guess it needs to be more fully dressed
saw many kids in the wild today and system detected them just fine however
haven’t tried driving into it yet, just visualization first



JFC this is sickening to read. Every sentence is just so wrong both from the flawed perception of the Tesla, to the warped testing methodology of WMC
 
I was under the impression that AEB only slowed the vehicle down something like 20-25 mph, but the OM doesn't seem to indicate as such. What it does mention is if AEB is occurring there is an icon displayed (along with an alert and sound) that indicates it is in progress.

True for AEB alone. However, I suspect that most tests have combined AEB and Autopilot or EAP/FSD/FSD beta.

Usually, we want Autopilot or EAP/FSD/FSD beta to brake to a stop. If that fails, we want AEB to kick in as a last-ditch effort even though the Owner's Manual says it only slows down and not to a complete stop.
 
That's the whole point: Tesla collision issue is not solved no matter which technology Tesla got it from, MobilEye, non-MobilEye, radar, radarless, immatured technology in 2016, or 6 years after that...

MobilEye broke off with Tesla in 2016 because it wants no part in "...pushing the envelope in terms of safety."
MobilEye seems to have changed their mind, their new "SuperVision" system sounds like it's basically the same as FSD beta. I guess the difference between 2016 and now is the better driver monitoring.
 
MobilEye seems to have changed their mind, their new "SuperVision" system sounds like it's basically the same as FSD beta. I guess the difference between 2016 and now is the better driver monitoring.
That's the misconception when LIDAR is mentioned, and lots of people think radar is dropped when it's an addition, not a deletion (deletion does happen in Tesla, though).

Same with SuperVision. We assume that it's a deletion of radar when it is not as of last month:



"Combined with the ability to operate at up to 130 kph (81 mph) on any road with clear lane markings – while also taking road curvature into account, and offering the latest in camera-radar fusion capabilities – this update has provided Zeekr 001 owners, overnight, with one of the most advanced highway-assist feature-sets available on the market today... and it's only the beginning."
 
That's the misconception when LIDAR is mentioned, and lots of people think radar is dropped when it's an addition, not a deletion (deletion does happen in Tesla, though).

Same with SuperVision. We assume that it's a deletion of radar when it is not as of last month:



"Combined with the ability to operate at up to 130 kph (81 mph) on any road with clear lane markings – while also taking road curvature into account, and offering the latest in camera-radar fusion capabilities – this update has provided Zeekr 001 owners, overnight, with one of the most advanced highway-assist feature-sets available on the market today... and it's only the beginning."
I was only commenting on MobilEye releasing what is basically a prototype self-driving car (the driver is only there to ensure safety). Nothing to do with sensor suite.
 
  • Like
Reactions: AlanSubie4Life
Whole Mars Catalog
@WholeMarsBlog

In my initial testing it’s been difficult to get the child sized dummy to read as a child. I guess it needs to be more fully dressed
saw many kids in the wild today and system detected them just fine however
haven’t tried driving into it yet, just visualization first



JFC this is sickening to read. Every sentence is just so wrong both from the flawed perception of the Tesla, to the warped testing methodology of WMC
You'll love the followup:


Whole Mars Catalog
@WholeMarsBlog
well it did register it as a person actually but the car was going too fast to stop fully
 
Accidentally confirmed the results of the original test?

Not sure what there is to be confirmed. The system will both see and not see child-sized dummies. Just like with actual children! The probabilities change based on a host of poorly understood variables, and no one can predict exactly what will maximize or minimize likelihood of detection. You just have to try out educated guesses until you get what you want!

Whos sole purpose is to pass a test with a known question.
That’s what is so great about that turn, as you have mentioned. Focusing on accomplishing one task can be very clarifying! It’ll be interesting if they CAN’T do it in the end (I expect they will be able to, though possibly not up to my specified standards at first), but it’s not surprising at all that it was nowhere close to being able to do it initially. It takes time to make the system pass the specified test. I mean, everyone knows how limited FSD Beta is. It’s nearly passable (not to be confused with passable) as a lane-line-following and car-following assist on City Streets, but that is about it.
 
  • Helpful
Reactions: Dan D.
You'll love the followup:


Whole Mars Catalog
@WholeMarsBlog
well it did register it as a person actually but the car was going too fast to stop fully
They can conduct their FSD tests with their own car all they want on private roads with non-living objects. I'm fine with that. Leave the public space (streets) and living beings out of it. I fully expect test results to be cherry-picked and possibly falsified on both sides at this point, any conclusions from random testers are anecdotal at best.

I'll believe a test by the NTSB, NHTSA or IIHS & equivalent from overseas. All the media and blogger testing is too haphazard, except as entertainment.
 
Problem is...when your CEO is CONSTANTLY bragging about all of the magical things that FSD can and will do, to include the car literally driving itself from California to NYC four years ago, to include the autonomous driving tech being safer than humans...etc

then one can kinda understand when its revealed that the same car has no capability to stop itself from running over and killing a child (or two) as it magically drives itself from NYC to California(with no human input), the public would be a bit....upset.

Perhaps if the CEO simply would sometimes....shut the EFF UP. Tesla wouldn't find itself on the defense so much
 
  • Like
Reactions: E90alex