This repoistory contains platform- and Middleware-independent code for a Joint Force Controller. It maintaines an internal joint state and trajectory sample time. As this class is controlling one joint only, a higher-level controller needs to manage the controller ensemble consisting of one controller instance per joint.
update_joint_states()
Determines new joint state based on current force data
on_transition()
Is called once the higher-level controller decides to transition from trajectory to force control. Stores reference data from the time of transition.
calculate()
Performs force control calculations. Resulting new desired position and velocity can be accessed via get_p_des() and get_v_des().
finish_iteration()
Stores data for next iteration.
reset_parameters()
Resets all controller parameters to their default values. A joint time can be given which will be stored as a trajectory sampling time offset.
The high-level needs to call the methods in correct order. force_ is a pointer to the force value and the high-level needs to make sure that it points to the newest force data before performing force control calculations.
These are the steps for the high-level:
- Update forces
- Call
update_joint_states() - Decide whether to transition into force control. If so, call
on_transition() - Calculate new desired state using
calculate(). Retreive position and velocity via getters - At the end of each update loop, call
finish_iteration()
The high-level itself can decide at which point the force control is terminated (e.g. some target force is reached) or if force control continues indefinitely. Before new goals, reset_parameters() should always be called to start with clean controller states.
Example usage can be found in the ta11_controller repository, where ta11_controller_impl.h is most interesting.