Wrong derivatives of constraintDynamics

[LudovicDeMatteis]

Bug description

The acceleration and contact forces derivatives computed by the function computeConstraintDynamicsDerivatives are incorrect on Release v3.0.0. It seems like only the case of 6D contact is affected. Here is a small python test for different robots, replicating the test test-cpp-contact-dynamics-derivatives of Pinocchio. It compares computed derivatives against finite differences.

The issue if already known by the maintainers and is being worked on

• Pinocchio version: 3.0.0

List of broken derivatives (to my knowledge)

• computeConstraintDynamicsDerivatives of 6D constraints