MFVS/MIMU integrated 6-DOF autonomous navigation in known environments with extremely simple landmarks

This study presents the navigation system of an inboard spacecraft (IS) of the Tiangong space station. This system comprises a multiple field-of-view vision system (MFVS) and a micro inertial measurement unit (MIMU). The MFVS utilizes three mutually orthogonal cameras and takes sparse and identical vision beacons as navigation references. A practical model for the MFVS is proposed based on a pseudo-measurement, which refers to the beacon coordinates in the camera frames. On this basis, three sub-algorithms for image processing and measurement equation forming are designed. Accelerated feature extracting and matching, dilution of precision analysis-based vision beacon selection strategy, and MIMU-aided image smearing elimination are analyzed. An extended Kalman filter is utilized to couple the MFVS and MIMU tightly. The 6-degree-of-freedom pose of the system and the vision beacon coordinates are estimated. The proposed algorithm is verified through a navigation experiment. Results inside a simulated cabin area of 1.5 m3 are obtained. The attitude and position accuracy in each axis reach within 0.04° and 0.41 mm (1σ), respectively. The proposed algorithm meets the requirement of the IS navigation and has potential application in other similar conditions.