Variable noise-covariance Kalman filter based instantaneous state observer for industrial robot

Abstract

To enhance the productivity and quality, high speed and high precision industrial robots are required. In the highspeed motion, the dynamic torque of industrial robot prevents the precise operation. To achieve high speed and high precision operation, the dynamic torque calculation is often used to compensate the dynamic torque. Although, it is difficult to calculate the accurate dynamic torque because the accurate dynamic parameters is difficult to obtain. To estimate the dynamic torque without the dynamic torque calculation, this paper uses the disturbance observer (DOB). To achieve the instantaneous dynamic torque estimation of industrial robot, the authors have proposed an instantaneous state observer (ISOB). However, the load torque estimated by the ISOB becomes noisy because the acceleration sensor has the measurement noise. To overcome this problem, this paper proposes a new method for instantaneous load torque estimation using the ISOB and a variable noise-covariance (VNC) Kalman filter. The effectiveness of VNC Kalman filter based ISOB is confirmed by the numerical simulation and experiments using the single joint of industrial robot arm.