Pinocchio 3.3: New Inertia Parametrization, Constraints Parsing and Numpy 2 Compatibility

The Pinocchio 3.3 update improves robotics modeling with the introduction of pseudo inertia and Log-Cholesky parametrization, enabling physically consistent inertial parameter handling. These tools provide new ways to parameterize inertia matrices, perform conversions, and compute Jacobian matrices for dynamic parameters, enhancing both precision and flexibility in robot dynamics modeling.

Additionnly, parsing of equality/connect tags for closed-loop chains in MJCF models has been added, along with compatibility updates for NumPy 2. We finaly improve model merging with the ability to set custom root joint names.

In addition to these new capabilities, this release resolves several key bugs and compatibility issues, including improved Windows builds, MJCF parsing corrections, and stabilized Python bindings. These enhancements make Pinocchio 3.3 a more reliable and powerful tool for robotics developers and researchers.

Links

Pinocchio 3.3 Release Note