Abstract:
This paper introduces an adaptive disturbance estimation and compensation approach for delta parallel robots using three methods. The first method is based on the adaptive Kalman filter (AKF), the second method uses the Low pass filtered robot dynamic model (LFDM) while the third method is acceleration measurement based (AMB) method which utilizes the measured moving platform acceleration directly into the robot dynamical model. The considered disturbance is joint friction, uncertainty and unmodeled dynamics, their effects are represented as lumped disturbance torque vector. The estimation
performance is evaluated using the mean square error (MSE) as a performance measure. To control the robot, the nonlinear robot model is linearized using feedback linearization through the estimated disturbance which is adaptively scaled using an adaptive tuning gain to overcome the limitations of the transient response of the estimated disturbance. The tuning is governed by a simple developed sliding surface depending on the error between the desired and actual joint angles. The tuned disturbance is added directly to the classical proportional–derivative (PD) controller output control signal for disturbance
compensation and trajectory tracking. Based on the results, a comparison among the three methods is studied. The comparison shows that the AKF method is the most accurate that tracks the desired trajectory in the presence of disturbance and noise. The other methods are not recommended
