A novel curve fitting based discrete velocity estimator for high performance motion control

Abstract

Robotics applications typically need high accuracy and repeatability obtained through precise measurement and control of the robot states which is not available in many cases due to the absence of the necessary sensor motivating the use of estimators. A typical scenario is represented by applications where position sensors (encoders) are present in the robot and velocity measurements are needed by the control scheme. Velocity estimators based on curve fitting (CF) are proposed and evaluated in this paper. A novel CF-based velocity estimator is developed for fitting a smooth curve by computing its polynomial parameters from a given set of position data points. The advantages of this estimator are essentially four: ease of implementation, high operational bandwidth with small phase lag and amplitude attenuation, high computational speed and effectiveness when used with low resolution position sensors. Furthermore, the novel method is model free. The efficacy of the estimator is confirmed by extensive simulation studies while experimental validation evidence is also provided on a single degree of freedom system demonstrating the accuracy and effectiveness of the proposed methods.