Releases: stack-of-tasks/pinocchio
Pinocchio 1.3.0
This new release introduces analytical derivatives in the corpus of Pinocchio.
This feature is still under development but can already be used both in C++ and Python.
This new release also fixes a bunch of bugs related to Eigen and Boost.
Pinocchio 1.2.9
This is mostly a maintenance release:
- Fix bug in. lower bounds in Model class
- Update documentation structure (additional work is needed)
- Improve the compatibility with Python 3.x
Pinocchio 1.2.8
This is mostly a maintenance release:
- Fix some bugs in JointModel{Translation,Spherical} for ABA algorithm
- Fix a duplication issue in RobotWrapper
- Improve compatibility with recent version of Boost >= 1.67.0
- Romeo is now loaded from the official romeo_description repository
Pinocchio 1.2.7
This is mostly a maintenance release:
- Fixes and computation improvements for Lie group operations.
- Adding pickle for spatial classes.
- Allow loading of URDF tree directly from an XML stream.
Release 1.2.6
This is mostly a maintenance release with various fixes to comply with Boost variadic macro on recent OS.
It also adds new convention with a LOCAL and a WORLD frame to express Jacobian quantities.
Release 1.2.5
This a maintenance Release. We added some algo to compute the time variation of the Jacobians together with the variation with respect to time of the centroidal momemtum matrix.
HPP-FCL works now with Eigen for linear algebra.
Release 1.2.4
This is mostly a maintenance release, with some fix with respect new urdfdom
versions, it handles Eigen support with hpp-fcl
.
Release 1.2.3
This release fixes some issues with respect to 1.2.1.
API modifications
Interpolate, Differentiate, Integrate are now algorithmic struct which can be efficiently overloaded.
Bindings
Add FCL object bindings
Release 1.2.1
Summary
This release is a minor patch of the previous release 1.2.0.
This release is directly accessible as a Debian package. Please see https://github.com/stack-of-tasks/pinocchio/wiki/installation for further details.
New
- The Python bindings are aligned and free of unnecessary allocations
- Add documentation option which allows to not install the documentation
- Introduces container::aligned_vector to automatically create an std::vector with specific aligned allocator
API modifications
- Remove JointDense.
- Remove JointGeneric
Fixes
- Solves the parsing of geometries in URDF module
- Fixes alignment issues on 32 bits architecture
Release 1.2.0
Summary
The main modifications concern the update of the code to comply with the Humanoid Path Planner (HPP).
This release is directly accessible as a Debian package. Please see https://github.com/stack-of-tasks/pinocchio/wiki/installation for further details.
New Features
- Add Joint{Model,Data} classes based on Joint{Model,Data}Base and Joint{Model,Data}Variant. Those classes call directly the visitors and make Variant accessible throw methods
- Add partial Joint{Model,Data}Composite. They allow a stack of joints without adding any Inertia.
- Increase Frame class. Frames can be of several types (BODY, JOINT, SENSOR, etc) and reflect the robot tree as it appears in the URDF conventions. Frames have two attributes: parent which the direct Joint parent in the Joint tree, previousFrame which correspond to the parent Frame in the tree of Frames).
- Adding Python parser which is able to read models written in Python.
- Add algo checker to check the validity of a model.
- Improve documentation.
API modifications
- Model has some methods deprecated. The default name has been removed.
- The Geometry classes have been updated and several methods have been set to deprecated. They now use Frames as parent instead of Joint directly: a Geometry is now supported by a BODY.
- Add active collision pair flags in GeomData which define the active collision pairs.
- Unify naming conventions (nframes, njoints, etc).
Fixes
- The UDRF parser can now deal with more complex topologies. It properly handles the stack of geometries for each BODY.
- Improve packaging mainly around the Python part.