Shane Smiskol
Shane Smiskol
I'm finding so many critical issues with new connect that didn't even seem to be known issues at the time. We need something like opendbc's [test.sh](https://github.com/commaai/opendbc/blob/master/test.sh) that tells you if...
Right now, without vehicle model's curvature factor that assumes we get less curvature than expected at higher speeds, we get 60% less lateral accel than we could at 80 m/s....
Why is this not a function? There's a few places where we check the same thing that could be in one function. Same with buses https://github.com/commaai/opendbc/blob/1a59488c69b5f751af31db17684f90ae3fc067b9/opendbc/safety/safety/safety_hyundai_canfd.h#L162
Can you tell what these two are for? Do they include longitudinal actuation messages or not? ``` static const CanMsg HYUNDAI_CANFD_LFA_STEERING_LONG_TX_MSGS[] = { ... #define HYUNDAI_CANFD_LFA_STEERING_CAMERA_SCC_TX_MSGS(longitudinal) \ ... ``` Hint...
- [x] There are faults we don't handle with FSD long and openpilot lateral. We can 1) detect when FSD is turned on and warn user to switch to TACC,...
Need to find the average across the fleet first @fredyshox can you try this?  
It is heavily rate limited, so if the user engages after a turn where it's wound up, even if Driver+ isn't enabled, the EPS may temp fault because of that...
Confirmed this is what this is:  Tried sending a single frame of the steer bit at 0, but the EPAS didn't re-enable after and caused the ACM to disengage...
Now that we dynamically allow a max curvature from a static max lateral acceleration and speed for Q4, we should be able to yet again bump the curvature rate limits...
We were requesting the stopping state, but the slow ramp down caused the car to undershoot our braking request (not a regression from Toyota tune) https://github.com/user-attachments/assets/248b4d06-4974-43ef-81b8-f7395fb3f814 