This package provides a Cartesian pose controller that converts desired end-effector poses and twists in task space to joint position commands for a robotic manipulator.
It is a chainable controller that works with the task space reference generator to obtain references.
The control law is illustrated in the figure below:
Note that the task space references should be expressed in the root_link_frame (see Controller Configuration), which defines the coordinate frame used as the reference for all pose and twist commands sent to the controller. The root_link_frame parameter must match the root link described in the robot description, this is because the kinematics interface assumes this frame as the reference frame for its calculations.
If the desired twist is not provided (and the controller is configured correctly), the controller computes the desired twist from the desired pose.
To guarantee controller stability, users must ensure that the gain matrix K is positive definite.
The matrix K is constructed as a diagonal matrix from the user-provided gain vector k_matrix_gains (see Controller Configuration). These gains determine the controller's convergence speed.
Control parameters for the cartesian pose controller are defined in the file cartesian_pose_controller_parameters.yaml.
Please refer to cartesian_pose_controller_parameters.md for the documentation.
To test this controller in simulation, please refer to the documentation of ur10_reference_generators_demo.