A Reduced-Order Multisensor-Based Force Observer

This article proposes a reduced-order multisensor-based force observer (RMFOB) for accurately estimating the force exerted on a load. The RMFOB excels in high-frequency noise attenuation and low-frequency time-varying measurement offset compensation by using the combination of force sensor measurements, motor encoder measurements, and motor input signals, properly fused in a Kalman filter (KF) setting, in which the dynamics of the sensors and the system is taken into account. As part of the observer design, an estimator for the force measurement offsets is also derived, so that they can be visualized and quantified. Moreover, novel tools for analyzing the KF-based observer are introduced. To evaluate its performance, the RMFOB is then implemented in both simulations and experiments. Results show that the observer can produce the accurate force estimate by compensating for the time-varying measurement offsets and attenuating the high-frequency noises. Results are presented for various experimental conditions to thoroughly assess the capabilities of the proposed observer.