Seeker gyro calibration via model-based fusion of visual and inertial data

Vision-guided autonomous platforms require inertial stabilization of the imaging sensor. This is typically achieved by using a gimbaled system with inertial rate sensors, such as gyros. Using low-cost gyros requires estimation of their error parameters, such as bias and scale-factor. This paper presents a motion model-based method for robust estimation of these parameters via fusing the inertial measurements with the imaging sensor´s data. Using the camera motion model renders the estimator more robust to image noise and feature point extraction and tracking errors. Using a 2-D motion field of the feature points on the image sequence, a novel constraint is derived, which couples this field with the velocity and rotation of the camera. Introducing this constraint as an implicit measurement model, together with the motion model of the camera, into the implicit extended Kalman filter, yields a robust recursive estimator of the gyro error parameters, that is independent of the scene´s structure.