gradslam/gradslam

PyTorch Auto Differentiation

Closed this issue · 2 comments

Where is the implementation of PyTorch auto differentiation ?

Hi, thanks for sharing the code. I checked the code of gauss_newton_solve in icputils.py but didn't found you using auto differentiation to implement LM solver.

Thank you!

@JeffreyYH I'm confused about the issue.

Do you mean that we don't have an explicitly implemented backward pass for ICP? That is intended, as we build each of our modules with PyTorch ops that are inturn differentiable (and thus we don't need an explicit backward method).

Thank you!