Sensor-minimal motion control of variable stiffness actuators based on the passivity principle and velocity estimation

This paper investigates a sensor-minimal approach to motion control in variable stiffness actuators. This control approach aims at reducing the need of sensors for tracking a desired trajectory. By this, problems caused by collocation and additional noise in measurement signals could get reduced along with economic benefits due to lower system complexity. Control laws are designed using a passivity-based approach which requires only one sensor, i.e, for measuring actuator position. Due to the demand of information about actuator velocities in the control law, those are calculated using a velocity compensator, numerical calculation, and linear filtering. The parameters of the controller and the velocity compensator are selected based on the global asymptotic stability requirements. To compare the three methods, those are compared regarding the example of the variable torsion stiffness actuator (VTS). Using the controller with the velocity compensator shows good results in simulations and experiments. Numerical calculation achieves slightly worse results while linear filtering does not show feasible results.