Feature: Compute Euclidean Distance Transforms of Binary Obstacle Maps

[mattking-smith]

Purpose

Compute the Euclidean Distance Transform (EDT) of Binary Obstacle maps in 2/3D in C++. This alleviates the need for external libraries like MATLAB and Python to compute EDT of binary images with functions like: bwdist and scipy.ndimage.distance_transform_edt.

Details

• The C++ algorithm implements the Felzenszwalb & Huttenlocher Algorithm to determine EDT of binary obstacles
• Introduced namespace dt (distance transform) under gpmp2
• Wrote unit tests with analytic cases for single and multidimensional signed distance fields
• Validated EDT implementation against existing tests

[mattking-smith]

This PR is associated with issue #27

[mattking-smith]

@dellaert I am not sure what is up with the CI failing to build GTSAM here on the server side. Is this happening on some new update to GTSAM?

However, these failures are not due to the change set introduced in this PR.

[dellaert]

Hmmm, right. Interestingly enough, these failures are probably the result from the Gaussian process factor PR, but all checks for that passed.

Could you ask an AI to update the workflows to match the flags that we have in GTSAM? Make sure it's c++17 etc? I think this repo is pretty old with respect to develop. I'd do it, but I'm a bit short on time and quota :-).

[mattking-smith]

Hmmm, right. Interestingly enough, these failures are probably the result from the Gaussian process factor PR, but all checks for that passed.

Could you ask an AI to update the workflows to match the flags that we have in GTSAM? Make sure it's c++17 etc? I think this repo is pretty old with respect to develop. I'd do it, but I'm a bit short on time and quota :-).

I'll look into it when I get the chance