Kalman filter based fine force sensation with periodic component extraction

This paper proposes a force sensing method with periodic component extraction based on Kalman filter algorithm. In a ball-screw system exhibiting high friction, a dither is inserted to the desired reference signal to eliminate the effect of friction. The force estimation by a disturbance observer is integrated with the force estimation by Kalman filter to construct the force sensation with effective oscillatory component extraction. The disturbance observer estimates the oscillatory torque caused by dither signal. The Kalman filter uses this oscillatory torque as a measurable input to estimate the periodical torque, the first derivative and second derivative of periodical torque. The torque with oscillatory component extraction is generated from estimations achieved by Kalman filter. Moreover, because the force sensation is based on Kalman filter, the effect of noise in force estimation is suppressed effectively. Therefore, it is possible to increase the bandwidth of the force sensing. The effectiveness of the proposed method is verified by numerical simulation results based on experimental data of force control of a ball-screw system.