-
Notifications
You must be signed in to change notification settings - Fork 121
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Lie Labs API example #476
Lie Labs API example #476
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Some comments about the behaviors of overriding.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I like the clean interface, can't wait to use it!
Some features I would love to see:
- Conversions to/from quaternions & matrices, even though they might be specific to SE3 and maybe out of scope, but that's a really really common usecase in robotics.
- I find the Sophus API to be more intuitive in terms of accessing the matrix data, and accessing the individual rotation and translation components. Maybe we can take some inspiration from that?
53f8c0c
to
f2849fa
Compare
Made a new version of the API based on the discussion on operator overload, and also the suggestions for things like |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Again, just my subjective opinion, but the new version looks great to me!
9d0dd10
to
9ac3ab3
Compare
The api looks great!
I think it would be useful if indexing was implemented.
Maybe it would be nice to keep the type consistent for typehint etc. |
9ac3ab3
to
613004b
Compare
d500ebd
to
0e23e8e
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM with all the feedback now addressed in the other PR.
Thanks @luisenp and @fantaosha for developing everything on Lie groups alpha and thanks to everyone else (on the PR and offline) for the valuable feedback!
0e23e8e
to
3130d06
Compare
Seeking feedback
This PR has a small example illustrating the current features of our new Lie group tensor library for PyTorch. We are looking for feedback on:
Current features in
alpha
LieTensor
subclass oftorch.Tensor
serving as entry point.torch
, but w/o a custom backward pass; we rely on torch autograd for this).LieTensor
parameters usingtorch
optimizers.compose
(*
) andtransform_from
(@
).Features planned for
beta
vmap
support.