Reduce unnecessary temporary disengagements #584
Conversation
This status was defined as a bump error, but it isn't actually due to bumps. It's also triggered by driver overrides that are somewhat higher than the "steer_override" threshold. By removing this temporary disengagement, the car is much more predictable on bumpy roads AND in turns that require assistance from the driver.
Gernby
changed the title
|
@Gernby , interesting. So with state |
|
Actually, I'm not certain whether the EPS continues responding to torque requests during state 4, but it doesn't cause any errors (on my 2018 Accord), and OP's behavior is more desirable. |
|
The reason why I say OP's behavior is more favorable is because it doesn't reset PID. This is a huge benefit in a curve. Here is a video showing a curve that would normally require me to rapidly grab the steering wheel before hitting the curb due to a bump. |
|
confirmed this change is good on Accord. Not sure about other models. |
|
If it helps, this change was merged into one of @kegman 's branches on 3/27/19 (not his default branch), which I believe has been used by many Honda's since then. https://github.com/kegman/openpilot/tree/kegman-plusGernbySteering-0.5.10 |
|
oh yes, that helps! |
|
verified. Nice change. |
sidebar shouldnt pop up when tapping df button.
This status was defined as a bump error, but it isn't actually due to bumps. It's also triggered by driver overrides that are somewhat higher than the "steer_override" threshold. By removing this temporary disengagement, the car is much more predictable on bumpy roads AND in turns that require assistance from the driver.
This status was defined as a bump error, but it isn't actually due to bumps. It's also triggered by driver overrides that are somewhat higher than the "steer_override" threshold. By removing this temporary disengagement, the car is much more predictable on bumpy roads AND in turns that require assistance from the driver.