Pink motions

A library of robot motions performed by inverse kinematics and model predictive control:

Usage

You can try out a scenario from the library directly using pixi:

pixi run scenario --robot ur5 --qpsolver quadprog

Or equivalently uv run examples/play_scenario.py with the same arguments. In Python, this example simply does:

import pink_motions

pink_motions.play_scenario(
    name="ur5",
    dt=0.005,  # timestep in seconds
    qpsolver="quadprog",
)

This will open a MeshCat tab in your web browser playing the scenario for its prescribed duration.

See also

Pink motions are specified using a combination of several open-source libraries:

• MeshCat for (optional) visualization
• Pink for differential inverse kinematics
• Pinocchio for forward kinematics functions: Jacobians, Jlog6, ...
• qpmpc for linear model predictive control, used in biped and humanoid scenarios
• robot_descriptions.py to load robot descriptions

The library is also related to the following projects:

• ik_qpbenchmark: Test set for QP solvers with differential IK problems generated from pink motions.
• LoIK: Constrained differential inverse kinematics solver that was evaluated on pink motions.
• Pinker: Next-generation differential inverse kinematics library, which was benchmarked using pink motions.