made mt1 std dev a bit better - #238
Conversation
|
/gemini review |
There was a problem hiding this comment.
Code Review
This pull request enhances the vision subsystem's pose estimation by introducing an angular baseline calculation for MT1 solves, replacing the previous harmonic distance sum model. It also adds several constants for handling odometry resets during defensive play and implements a shared reset cooldown. A review comment identifies a telemetry issue where the "angularBaselineRad" key in SmartDashboard is shared across all cameras, leading to data overwrites in NetworkTables.
There was a problem hiding this comment.
Code Review
This pull request enhances the vision subsystem by introducing an angular baseline calculation to improve standard deviation estimation for MT1 solves, replacing the previous harmonic sum approach. It also adds defense-specific reset thresholds and a shared cooldown for translation resets to prevent thrashing. The review feedback focuses on improving the robustness of the new computeAngularBaseline method by addressing a potential null pointer exception and ensuring that diagnostic logging to the dashboard is camera-specific to prevent data overwriting in multi-camera configurations.
|
/gemini review |
There was a problem hiding this comment.
Code Review
This pull request enhances the vision subsystem by introducing a more robust standard deviation model for MegaTag1 (MT1) based on angular baseline and adding a "defense reset" mechanism. The defense reset allows for hard-snapping odometry to high-confidence vision readings when significant disagreement is detected under defense contact. A shared cooldown was implemented to prevent reset thrashing. Feedback focuses on potential telemetry overwriting, the accuracy of the angular baseline calculation relative to camera offsets, and the priority of MegaTag2 resets over MT1 within the cooldown window.
|
NOTE: don't merge until worlds branch is merged |
No description provided.