-
Notifications
You must be signed in to change notification settings - Fork 38
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
Add PI-ADHERENT framework to ML #889
Add PI-ADHERENT framework to ML #889
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.
First of all well done 🎉 the code is a good shape :) I added some comments in the PR. Here are some general comments
- I would avoid adding ONNX here. You can approach the problem as it is done for MANN. The ONNX model is downloaded from Hugging Face during project compilation
- We may call the classes PIMANN instead of VelMANN (something we can discuss F2F)
- there are some leftovers in the comments
bindings/python/ML/include/BipedalLocomotion/bindings/ML/velMANN.h
Outdated
Show resolved
Hide resolved
bindings/python/ML/include/BipedalLocomotion/bindings/ML/velMANN.h
Outdated
Show resolved
Hide resolved
bindings/python/ML/include/BipedalLocomotion/bindings/ML/velMANN.h
Outdated
Show resolved
Hide resolved
src/ML/include/BipedalLocomotion/ML/velMANNAutoregressiveInputBuilder.h
Outdated
Show resolved
Hide resolved
dbbd9d2
to
8bf5fe2
Compare
402ba22
to
7f2fe11
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.
Some minor change I can directly fix (they are just formatting issues)
devices/YarpRobotLoggerDevice/app/robots/ergoCubGazeboV1_1/yarp-robot-logger.xml
Outdated
Show resolved
Hide resolved
The PR will be automatically merged when the CI ends |
Thank you for the effort @evelyd 🎉 |
These changes comprise all of the implementation of the velocity-based features, use of PI-ADHERENT trained models, and trajectory generation capability for the aforementioned model.
I also made a small fix to the Python binding of the
change_fixed_frame
function, because the binding doesn't work otherwise (see #310).