commaai
VW PQ: Update HCA Status to use Mode 2 (Status 7)
This PR aims to switch VW PQ to use HCA Status 7 (HCA7) instead of HCA Status 5 (HCA5), which would increase the available lateral torque usable by openpilot. HCA7 lacks the recentering force that constantly fights openpilot for lateral torque in HCA5. The lack of this recentering force increases the torque that openpilot can command. This change wasn’t previously merged due to issues controlling HCA7, but with the new MLSIM models, architecture updates, and OpenDBC refactor, many recent users (including myself) are running HCA7 and report it to be more stable and controllable & an increase in torque compared to HCA5. This behavior should be present in stock openpilot as it gets rid of almost any torque warning or lack of torque on the highway & increases available torque for lower speed driving. Validation
- Route (Golf MK6):
4a824aa67cbfcd85/00000012--6672ca6b12 - Route (NMS Passat):
b3f9167062e09c61/0000008e--622dc1eab4more routes will be given if needed.
Just to set expectations, I'm down with merging this if the data look good, but it's going to take a while to evaluate. I would have expected this to make the lateral accel torque controller behave oddly.
I see that some test routes have been provided. I'd like to see before/after with the same cars and same route driven.
Just to set expectations, I'm down with merging this if the data look good, but it's going to take a while to evaluate. I would have expected this to make the lateral accel torque controller behave oddly.
I see that some test routes have been provided. I'd like to see before/after with the same cars and same route driven.
just to be clear, routes with the same vehicles running HCA5, and then HCA7 correct? I will publish the routes as soon as I get them
Just to set expectations, I'm down with merging this if the data look good, but it's going to take a while to evaluate. I would have expected this to make the lateral accel torque controller behave oddly.
I see that some test routes have been provided. I'd like to see before/after with the same cars and same route driven.
MK6 HCA5 route: 4a824aa67cbfcd85/0000001c--e45fa436f1
Just to set expectations, I'm down with merging this if the data look good, but it's going to take a while to evaluate. I would have expected this to make the lateral accel torque controller behave oddly. I see that some test routes have been provided. I'd like to see before/after with the same cars and same route driven.
NMS Passat HCA5 Route: b3f9167062e09c61/000000a6--c5dcb34484/8 (segments 7-11)