Engineers Responding to FSD Reports

[PianoAl]

I've been watching FSD Youtubes, seeing the drivers tap an icon when they experience a problem.

I'm curious about how that works on the Tesla side. With 100,000 FSD beta testers out there, they'll be getting a lot of problem reports, and I'll bet that for many of those, the engineers will have a hard time deciphering what the problem is. Sure, sometimes it will be obvious, but I'll bet that often the engineers are scratching their heads over why the driver reported a problem.

For example, at 10:52 the woman in this video pushes the snapshot button. See if you, watching the video, starting at 10:30, with no sound, can tell what went wrong:

 

[Supcom]

I've been watching FSD Youtubes, seeing the drivers tap an icon when they experience a problem.

I'm curious about how that works on the Tesla side. With 100,000 FSD beta testers out there, they'll be getting a lot of problem reports, and I'll bet that for many of those, the engineers will have a hard time deciphering what the problem is. Sure, sometimes it will be obvious, but I'll bet that often the engineers are scratching their heads over why the driver reported a problem.

For example, at 10:52 the woman in this video pushes the snapshot button. See if you, watching the video, starting at 10:30, with no sound, can tell what went wrong:

The car sends telemetry data along with the video snapshot, so engineers have much more information at their disposal than you do watching a video with no sound.

However, your point that Tesla gets a lot of reports is, of course, correct. I suspect that any disconnect, possibly except for up stalk, results in a report. Obviously, the FSD team cannot do a human review of all those reports. But, they likely have a database that stores relevant information so that the team can extract reports matching some issue that they are working.

So, don't expect the FSD team to react to fix your specific issue. But, whatever issues you have anre almost certainly experienced by thousands of other testers. And the more testers sending reports on an issue, the more likely it is that Tesla will address it.

 

[nvx1977]

For example, at 10:52 the woman in this video pushes the snapshot button. See if you, watching the video, starting at 10:30, with no sound, can tell what went wrong:

The obvious fault was that the car stopped way early from the stop sign and then takes its sweet time creeping up to the sign to stop again, a known issue since forever. Where I drive, someone is already leaning on the horn behind me.

 

[Watts_Up]

Obviously, the FSD team cannot do a human review of all those reports. But, they likely have a database that stores relevant information so that the team can extract reports matching some issue that they are working.

In fact, last June 2022, Tesla laid off about 200 Data Annotation Specialists.

Tesla's layoffs hit Autopilot team as AI develops

Tesla is increasingly turning to machines rather than humans as it attempts to develop autonomous vehicles.

www.wral.com

Tesla has developed an automated way to do some of this labeling work​

in recent years, which allows the automaker to streamline its labor force.​

 

[nvx1977]

In fact, last June 2022, Tesla laid off about 200 Data Annotation Specialists.

Tesla's layoffs hit Autopilot team as AI develops

Tesla is increasingly turning to machines rather than humans as it attempts to develop autonomous vehicles.

www.wral.com

Tesla has developed an automated way to do some of this labeling work​

in recent years, which allows the automaker to streamline its labor force.​

For perspective, Elon said before the layoffs they had 1800! people on that team. Makes sense that as auto-labeling becomes better, they won't need so many human labelers.