Skip to content

Conversation

@keeminlee
Copy link
Collaborator

This update introduces full support for calibration-based nonlinear EKS smoothing.
Key changes:

  • --calibration argument now accepts a .toml file to load camera parameters and automatically trigger the nonlinear (calibrated) smoothing path. This replaces the linear PCA emission model with a JAX-compatible multi-view projection function (h_fn) built from camera intrinsics/extrinsics.

  • Added a multi-view fly toy dataset with calibration.toml for testing nonlinear smoothing and triangulation consistency.

  • Implemented JAX-based reimplementations of OpenCV triangulation and projection functions, ensuring differentiable and JIT-compatible observation models for the EKF.

  • Added pytests to validate numerical parity between JAX and OpenCV triangulation/projection outputs.

@themattinthehatt
Copy link
Collaborator

thanks @keeminlee!! I'll take a look at this when I have a chance, might not be until thursday

Copy link
Collaborator

@themattinthehatt themattinthehatt left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

looks good! minor changes requested

Copy link
Collaborator

@themattinthehatt themattinthehatt left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

please add nonlinear tests using fly data to tests/scripts/test_multicam_example.py

@themattinthehatt themattinthehatt merged commit 3395efd into main Nov 20, 2025
1 check passed
@themattinthehatt themattinthehatt deleted the calibration-aware branch November 20, 2025 21:04
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.

3 participants