Add INERTIAL_FIXED_REPRESENTATION to velocity representations #144
Conversation
- substitute for some quantites the default mixed naming with a more general one. - refactor jacobian_dot velocity repr handling by selecting the appropiate adjoint functions
There was a problem hiding this comment.
Pull request overview
This PR adds support for the INERTIAL_FIXED_REPRESENTATION frame velocity representation to the rigid body dynamics algorithms in the ADAM library. This expands the library's capability to handle a third type of velocity representation alongside the existing body-fixed and mixed representations.
Key changes include:
- Added
INERTIAL_FIXED_REPRESENTATIONto theRepresentationsenum - Implemented transformation logic across all core RBD algorithms (CRBA, Jacobian computations, RNEA, ABA) to properly handle the inertial-fixed frame
- Updated test infrastructure to include the new representation in test configurations
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| src/adam/core/constants.py | Added INERTIAL_FIXED_REPRESENTATION enum value to support the new frame representation type |
| src/adam/core/rbd_algorithms.py | Implemented inertial-fixed representation logic across all major algorithms including appropriate adjoint transformations, Jacobians, and dynamics computations |
| tests/conftest.py | Extended test configuration to include INERTIAL_FIXED_REPRESENTATION in velocity representation test cases and added mapping to iDynTree representation |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Co-authored-by: Copilot <[email protected]>
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| B_X_BI = math.factory.eye(batch_shape + (6,)) | ||
| B_X_I = math.factory.eye(batch_shape + (6,)) | ||
| transformed_acc = math.factory.zeros(batch_shape + (6,)) | ||
| elif self.frame_velocity_representation == Representations.MIXED_REPRESENTATION: | ||
| B_X_BI = math.adjoint_mixed_inverse(base_transform) | ||
| B_X_I = math.adjoint_mixed_inverse(base_transform) | ||
| omega = base_velocity[..., 3:] | ||
| vlin = base_velocity[..., :3] | ||
| skew_omega_times_vlin = math.mxv(math.skew(omega), vlin) | ||
| top3 = -math.mxv(B_X_BI[..., :3, :3], skew_omega_times_vlin) | ||
| top3 = -math.mxv(B_X_I[..., :3, :3], skew_omega_times_vlin) | ||
| bot3 = math.factory.zeros(batch_shape + (3,)) | ||
| transformed_acc = math.concatenate([top3, bot3], axis=-1) |
There was a problem hiding this comment.
Are these changes in the non-inertial fixed code path intentional?
There was a problem hiding this comment.
Yup! Before everything was written in mixed (e.g. B_X_BI), also for body fixed and I wanted to make it more generic.
But now, thinking twice, I could define a generic frame C and specify in a comment to which frame I'm referring to (e.g. C = B[I]).
traversaro
left a comment
traversaro
left a comment
There was a problem hiding this comment.
Just a minor comment.
|
Thanks @traversaro! Merging. |
This pull request adds support for the
INERTIAL_FIXED_REPRESENTATIONframe velocity representation throughout the rigid body dynamics algorithms.Support for INERTIAL_FIXED_REPRESENTATION in core algorithms:
INERTIAL_FIXED_REPRESENTATIONto theRepresentationsenum inconstants.py.block_index,jacobian,relative_jacobian,jacobian_dot,CoM_jacobian,rnea,aba, and related helper logic) inrbd_algorithms.pyto implement logic for the inertial-fixed representation, including appropriate transformations, Jacobians, and derivatives. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]Test infrastructure updates:
INERTIAL_FIXED_REPRESENTATIONin the list of velocity representations tested inconftest.py, ensuring coverage in all relevant robot configurations.📚 Documentation preview 📚: https://adam-docs--144.org.readthedocs.build/en/144/