ford ALP: restore StarPilot reference guards (narrow lane, laneline std, NaN params) - #179
Merged
Merged
Conversation
…rams) and cap applied correction at 0.0015
The three reference guards (narrow-lane floor, laneline std, NaN params) stay. The 0.0015 post-gain ceiling moves to a follow-up PR so the ceiling value can be chosen from on-road testing; tests revert to the uncapped 0.004 ceiling.
narrow-lane floor, laneline std, NaN params, and a pin on the fallback-bias semantics (constant push by design). All five fail without the guards.
ghbarker
changed the title
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Three small hardenings to the angle-mode lane centering trim, each restoring behavior from the StarPilot reference the port was based on:
np.interpsaturates below its first breakpoint, so the width-tolerance table scored a 1.5 m "lane" (double-stripe repaint, gore-point paint) at full trust 0.81 — with confident probs that's scale 1.0 and full-authority centering between stripes. StarPilot rejected width < 2.6 m; this restores that floor as a smooth edge (2.4–2.8 m) to match the blend design. No change from 2.8 m up.The fourth item from the original version of this PR (a post-gain ceiling on the applied correction) is split out to #181 per review, so the ceiling value can be chosen from on-road testing without blocking these guards.
Verification
Unit tests only — not road-tested (car is down). Five new tests: one per guard (narrow lane, high std, NaN offset, NaN gain) plus one that pins the current fallback-bias semantics as-is; all five fail on bp-dev-ALP without this change. In the existing suite, the fake model fixture regains laneLineStds and a few gain values were adjusted; 25/25 pass.