Gyroscope offset estimation using panoramic vision-based attitude estimation and Extended Kalman Filter

Inertial sensors suffer from different types of errors that cause drift in the attitude measurement. Gyroscopes are subject to thermal drift that will cause the attitude to drift over time due to the integration of sensor error by the attitude algorithm. The gyroscope drift, named as offset, can be estimated using Extended Kalman Filter (EKF). In strapdown systems, the direction of gravity measured by accelerometers is used to correct the drift. However the tilt measured by the accelerometers is corrupted by the vehicle dynamics when turning or accelerating. Vision systems are not vulnerable to the vibration and can provide a good attitude reference. This paper investigates the attitude and the dynamic gyroscope offset estimation using an quaternion-based EKF using the attitude reference from a panoramic Ultra Violet (UV) vision system and an inertial based Attitude Heading Reference System (AHRS). The results of the proposed methods are validated against an AHRS.