Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Honda Pilot tuning / lane centering fixes #408

Merged
merged 1 commit into from Oct 31, 2018
Merged

Honda Pilot tuning / lane centering fixes #408

merged 1 commit into from Oct 31, 2018

Conversation

ghost
Copy link

@ghost ghost commented Oct 25, 2018

Tuned kP and kI and Tire Stiffness Factor settings.
Fixes lane centering issues in the following situations:

  1. Hugging of left lane line during fast left curving roads
  2. Hugging of right lane line during fast right curving roads
  3. Left bias in fast lane on crowned roads (sloping down to the left)
  4. Right bias in slow lane on crowned roads (sloping down to the right)
    Works well with lane_width = 2.85 in pathplanner.py

Tuned kP and kI and Tire Stiffness Factor settings.
Fixes lane centering issues in the following situations:
1.  Hugging of left lane line during fast left curving roads
2.  Hugging of right lane line during fast right curving roads
3.  Left bias in fast lane on crowned roads (sloping down to the left)
4.  Right bias in slow lane on crowned roads (sloping down to the right)
Works better when lane_width = 2.85 (vs 3.7 default) in pathplanner.py
@vikovo
Copy link

vikovo commented Oct 25, 2018

I’m driving a Honda Pilot 2019 ex-l. Noted noticeable improvement on fast lane curves especially against the barricade. Car remained centered and steady. Did not require manual steering.

@rbiasini
Copy link
Contributor

looks good @kegman . Kp seem quite high though. No steering wheel wobble?

@vikovo
Copy link

vikovo commented Oct 26, 2018

I didn’t note any wobble

@jfrux
Copy link

jfrux commented Oct 26, 2018

Going to test this...

@jfrux
Copy link

jfrux commented Oct 31, 2018

Just tested this on my morning commute.
It seemed okay, I didn't notice anything significant or profound in this.
A lot of these minor changes are hard to really gauge their benefit.

I definitely don't see any negative side-effects thus far.

Let me test on my commute home as that is always a bit more vigorous and I'm generally paying more attention to its behavior.

@ghost
Copy link
Author

ghost commented Oct 31, 2018

The changes are subtle, especially if you don't have heavily crowned roads where the slope is steep on the fast / slow lanes. It "sticks" to the center more strongly, especially around fast curves where you are hitting the steering limit. If you are driving on straight flat, center lanes you probably won't notice the difference.

@rbiasini
Copy link
Contributor

ok, there is enough consensus. Will merge.

@rbiasini rbiasini merged commit fdbf1c5 into commaai:devel Oct 31, 2018
@jfrux
Copy link

jfrux commented Oct 31, 2018

Brake noise issue is good, this one I'm not sure its good for 2017.
My pilot is getting lots of steering warnings and shake... weather isn't the best so it could be a fluke...
I'll leave this one up to Riccardo...

@ghost
Copy link
Author

ghost commented Nov 1, 2018

Did the shake happen this morning?

@rbiasini
Copy link
Contributor

rbiasini commented Nov 1, 2018

well, it's just tuning, so it's your call guys. I just need to make sure it does not introduce weird behaviors for anybody.
👎 I'll revert, 👍 I'll keep
@jfrux , @kegman , who else?

@ghost ghost deleted the patch-1 branch November 2, 2018 02:25
@jfrux
Copy link

jfrux commented Nov 2, 2018

I'm not sure I can get behind this change at this time...
I don't see any improvements, I do see some places where it is actually worse in the most simplest of straights and basic turns I get way more alerts and random steering shakes on expressways.

I'm not saying we don't need some adjustments but at this time I don't know if this one is it...

Although it may improve in certain situations like backroads or state routes, I think I'd rather have it back the way it was and just wait it out for HD Maps.

Or maybe we need some sort of config file system built soon that accepts a YAML file so people can make their tweaks without forking and modifying the code.

@ghost
Copy link
Author

ghost commented Nov 2, 2018

Perhaps you're right it only improves for 2018/2019 model. Or perhaps my particular car has some variances that most cars don't. Im fine with a reversion to the way it was before... Its gotta work well for the majority, not just for me.

@rbiasini
Copy link
Contributor

rbiasini commented Nov 4, 2018

allright. I'll revert. Proper solution seem to differentiate between 2017 and 2018+ models

rbiasini added a commit that referenced this pull request Nov 13, 2018
rbiasini added a commit that referenced this pull request Nov 13, 2018
klaus385 pushed a commit to klaus385/openpilot that referenced this pull request Dec 31, 2018
Tuned kP and kI and Tire Stiffness Factor settings.
Fixes lane centering issues in the following situations:
1.  Hugging of left lane line during fast left curving roads
2.  Hugging of right lane line during fast right curving roads
3.  Left bias in fast lane on crowned roads (sloping down to the left)
4.  Right bias in slow lane on crowned roads (sloping down to the right)
Works better when lane_width = 2.85 (vs 3.7 default) in pathplanner.py
klaus385 pushed a commit to klaus385/openpilot that referenced this pull request Dec 31, 2018
icmma pushed a commit to priuscom/openpilot that referenced this pull request Jul 17, 2019
Fix old_car not fingerprinting sometimes
dragonpilot pushed a commit to dragonpilot-community/dragonpilot that referenced this pull request Apr 7, 2020
Tuned kP and kI and Tire Stiffness Factor settings.
Fixes lane centering issues in the following situations:
1.  Hugging of left lane line during fast left curving roads
2.  Hugging of right lane line during fast right curving roads
3.  Left bias in fast lane on crowned roads (sloping down to the left)
4.  Right bias in slow lane on crowned roads (sloping down to the right)
Works better when lane_width = 2.85 (vs 3.7 default) in pathplanner.py
dragonpilot pushed a commit to dragonpilot-community/dragonpilot that referenced this pull request Apr 7, 2020
github-actions bot pushed a commit to LexusRXopenpilotUG/openpilot that referenced this pull request May 4, 2021
* send future accel (0.45 seconds) to car as feedforward, not changing velocity

* add comment for future me

* add comment for future me

* Revert "Disable long derivative if eager accel is enabled (same thing)"

This reverts commit dc785ce.

* Disable long deriv

* Revert "Experimental feature: eager accel (commaai#382)"

This reverts commit 58ee449.

* rename back

* offset but keep gas so car can release brakes easier

* just offset

* Update version

* Update README.md

* Update README.md

* Update README.md

* don't change tuning

* add CP for accel delay

* clip

* always get that confused

* notes

* Update date

* Add param to tune

* Add param to tune
SebastienLubrano pushed a commit to SebastienLubrano/openpilot that referenced this pull request Jul 8, 2021
* send future accel (0.45 seconds) to car as feedforward, not changing velocity

* add comment for future me

* add comment for future me

* Revert "Disable long derivative if eager accel is enabled (same thing)"

This reverts commit dc785ce.

* Disable long deriv

* Revert "Experimental feature: eager accel (commaai#382)"

This reverts commit 58ee449.

* rename back

* offset but keep gas so car can release brakes easier

* just offset

* Update version

* Update README.md

* Update README.md

* Update README.md

* don't change tuning

* add CP for accel delay

* clip

* always get that confused

* notes

* Update date

* Add param to tune

* Add param to tune
SebastienLubrano pushed a commit to SebastienLubrano/openpilot that referenced this pull request Jul 8, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants