ros2_impedance_controller

Robot impedance controller designed with the ros2_control framework and Pinocchio. Default branch ROS2 distro: Jazzy

• Custom build dependency: kinematic_pose_msgs
• Simulation dependency: ros2_descriptions
• Handy reference generator: impedance_reference_generator

Instructions

Simulation with Robotic Arm

Use launcher default arguments:

ros2 launch ros2_impedance_controller simulation.launch.py

ros2 control set_controller_state impedance_controller active

Simulation with Spot leg

ros2 launch ros2_impedance_controller simulation.launch.py robot:=spot_leg controller:=leg_impedance_controller

ros2 control set_controller_state leg_impedance_controller active

About

The controller implement the classical impedance control law:

$$\color{Black} \mathbf{\tau_{act}} = \mathbf{g}(\mathbf{q}) + \mathbf{J}(\mathbf{q})^{T}\,[\mathbf{\Lambda}(\mathbf{x})\,\ddot{\mathbf{x}}_{d} + \mathbf{\Omega}(\mathbf{x}, \dot{\mathbf{x}})\,\dot{\mathbf{x}}\, -\mathbf{\Lambda}(\mathbf{x})\,\mathbf{\Lambda_d}^{-1}\,(\mathbf{D_d}\,\dot{\mathbf{e}} + \mathbf{K_d}\,\mathbf{e}) + \color{DarkGreen} (\mathbf{\Lambda}(\mathbf{x})\,\mathbf{\Lambda_d}^{-1} - \mathbf{I})\,\mathbf{f_{int}} \color{Black}]$$

The last green term is optional. Without inertia shaping, the force-torque sensor feedback is not required. Then, the law simplifies to:

$$\color{Black} \mathbf{\tau_{act}} = \mathbf{g}(\mathbf{q}) + \mathbf{J}(\mathbf{q})^{T}\,[\mathbf{\Lambda}(\mathbf{x})\,\ddot{\mathbf{x}}_{d} + \mathbf{\Omega}(\mathbf{x}, \dot{\mathbf{x}})\,\dot{\mathbf{x}}\, -\mathbf{D_d}\,\dot{\mathbf{e}} - \mathbf{K_d}\,\mathbf{e}]$$

Please be aware that, without inertia shaping, the desired inertia is the robot current inertia. Refer to Ott, C., 2008, Cartesian Impedance Control of Redundant and Flexible-Joint Robots for further details.