HW2.5 capabilities

[xspace]

Hi everybody. How are those of you capturing event trigger images doing it? Are you monitoring the wifi signals, then parsing them out? Are they not encrypted?

I searched through this thread, as well as on the internet, but couldn’t find the answer.

Thank you!

 

[ABC2D]

Hi everybody. How are those of you capturing event trigger images doing it? Are you monitoring the wifi signals, then parsing them out? Are they not encrypted?

I searched through this thread, as well as on the internet, but couldn’t find the answer.

Thank you!

The car is rooted

 

[verygreen]

they used to send cleartext snapshots to tesla-hermes-snapshot.s3.amazonaws.com (port 80), no idea if they still do.

HW2 Anonymous Snapshot

 

[xspace]

The car is rooted

Cool. So, does he ssh into it, and read locally stored compressed files? Or does he duplicate transmissions, sending them to another server?

Thank you.

 

[xspace]

they used to send cleartext snapshots to tesla-hermes-snapshot.s3.amazonaws.com (port 80), no idea if they still do.

HW2 Anonymous Snapshot

Excellent, thank you. I’ll read through this!

 

[xspace]

Excellent, thank you. I’ll read through this!

I just went to the url:80, and the following was the response:

AccessDeniedAccess DeniedC643992D36F8F997QGMiuK4MIh7

Either pilot error on my part, or they’re now preventing access?

 

[brkaus]

I just went to the url:80, and the following was the response:

AccessDeniedAccess DeniedC643992D36F8F997QGMiuK4MIh7

Either pilot error on my part, or they’re now preventing access?

I believe the way you see the file is to
Intercept it at your home firewall as it is getting sent up. Need some sort of transparent proxy at the firewall.

 

[xspace]

I believe the way you see the file is to
Intercept it at your home firewall as it is getting sent up. Need some sort of transparent proxy at the firewall.

OK, thank you!

 

[3Victoria]

Is it possible that the snap-shot is taken if there is disagreement between the vision system and the map?

 

[Padelford]

Ok, so now that I know the snapshots for my car stopped because the time for them was limited to just 23 hours, I rolled the time back and drove around some more.

Only managed to trigger a few slope up/down. The confusing part is oftentimes slope up would have a slope down in it too so I am not sure how NN decided if it's a slope up or down it's facing (because it's both).

From some experience, I’ve found that AP 2.0 doesn’t handle ascending or descending roadway well, especially if the roadway turns right or left. I’m glad someone else sees this as well.

 

[brkaus]

From some experience, I’ve found that AP 2.0 doesn’t handle ascending or descending roadway well, especially if the roadway turns right or left. I’m glad someone else sees this as well.

Definitely! Rolling hills with turns. Not even big hills. I was glad that verygreen found triggers for hills. Hopefully some day it will work well.

 

[daktari]

Is it possible that the snap-shot is taken if there is disagreement between the vision system and the map?

Could also be that sensitivity is way to low, meaning false negatives.

Also glad they have a slope up/down trigger, I think my car is tricked to believe even the slightest slope is a bend. Low specificity probably because of lack of 3D perception.

 

[BigD0g]

Well if you think about it on a upward hill with a left or right turn, the car has no clue, since it can’t find the lane lines with the cameras facing the sky, they don’t pivot down like our eyes. This won’t get solved until they are leveraging the HD maps I suspect.

 

[ChrML]

Well if you think about it on a upward hill with a left or right turn, the car has no clue, since it can’t find the lane lines with the cameras facing the sky, they don’t pivot down like our eyes. This won’t get solved until they are leveraging the HD maps I suspect.

Even if you know where the road is, you can't drive without seeing where you go. EAP might, but FSD can't.

HD maps might help by giving info that increases safety causing the car to drive more safely than a driver would by using the what he sees. But it should not decrease safety by driving into an area that can't be verified by the car's sensors.

 

[xspace]

Question about AP2.0. Does anybody else find that in both TACC and AutoSteering, the car will maintain full speed as it approaches a stopped car ahead of it, until pretty hard braking is required? In this game of chicken, I nearly always hit the brakes before the Tesla does. This is bad for regen as well, as it requires significant brakes.

 

[brkaus]

Well if you think about it on a upward hill with a left or right turn, the car has no clue, since it can’t find the lane lines with the cameras facing the sky, they don’t pivot down like our eyes. This won’t get solved until they are leveraging the HD maps I suspect.

I see it on shallow enough slopes (or upward slopes) and turns where I assumed the road road is still visible. ESP where there is some side to side tilt change too. I think it’s hard to understand in monocular vision.

 

[brkaus]

Question about AP2.0. Does anybody else find that in both TACC and AutoSteering, the car will maintain full speed as it approaches a stopped car ahead of it, until pretty hard braking is required? In this game of chicken, I nearly always hit the brakes before the Tesla does. This is bad for regen as well, as it requires significant brakes.

Yes. All the time if the cars are fully stopped. Don't count on it stopping every time either, I've read of cases where it didn't stop. Someone posted on a Texas Tesla email list just this week about an accident. From the sounds of it, totaled the car.

 

[BigD0g]

Yah, that's the untracked car problem, the car doesn't "see" the car sometimes if it's not moving and didn't see it beforehand moving. Never ever trust the car to stop unless, your behind a moving car and even then be very wary around stop / go traffic especially local traffic.

 

[DOCAL]

I see it on shallow enough slopes (or upward slopes) and turns where I assumed the road road is still visible. ESP where there is some side to side tilt change too. I think it’s hard to understand in monocular vision.

I've noticed it on AP1 on some very shallow slopes. I had always assumed that it was treating either the left or right line as a primary reference, and so the apparent curve was throwing it off.

 

[BeesKnees]

I've noticed it on AP1 on some very shallow slopes. I had always assumed that it was treating either the left or right line as a primary reference, and so the apparent curve was throwing it off.

My conclusion as well. I see the IC flash a sharp turn in the road. Exactly what you’d expect if it lost one edge, or prioritized one, on crest or dip. At least what we’d expect if there is no depth data.

I’ve been watching for edge drop out but one wants to be driving at these moments : ) so I don’t know.