Sensor Fusion Based on Embedded Measurements for Real-Time Three-DOF Orientation Motion Estimation of a Weight-Compensated Spherical Motor

Abstract

This paper presents a sensor fusion method to estimate the three degrees-of-freedom (3-DOF) orientation of a ball-joint-like permanent-magnet spherical motor (PMSM) and its angular velocity using embedded sensors that simultaneously measure the existing magnetic flux density (MFD) field and the back-electromotive-force (back-emf), which serve as inputs to a Kalman filter (KF) based sensor fusion system for full state estimation of 3-DOF angular displacement and velocity in real-time. Formulated in quaternion representation, the effectiveness and accuracy of the sensor fusion system consisting of an artificial neural network (ANN) that determines the 3-DOF orientation from measured MFD and an emf-velocity model, has been experimentally evaluated on an additive manufactured prototype PMSM- Weight Compensating Regulator (WCR) by comparing the estimated orientation and angular-velocity with that measured by two most used methods, optical laser-beam system, and inertial measurement unit (IMU). The experimental findings demonstrate that the KF-based sensor fusion effectively overcomes the MFD sensor noise and IMU drift problems and is capable of simultaneous measurements of 3-DOF angular displacement and velocity with improved accuracy relative to the popular IMU measurements.