Vision aided Navigation for Autonomous Underwater Vehicle

As highly sophisticated missions using AUV (Autonomous Underwater Vehicle) expand to include more hazardous situations, there are increasing needs to techniques which provide accurate navigation information. In order to provide continuous and accurate navigation information, integrated navigation systems have been used in which several sensors are used. Especially, among there sensors, vision is a powerful sensing modality and could be used in many tasks. This paper proposes vision/low-cost INS navigation for AUV. Continuously digitized image provides position and attitude using image processing and direct geo-referencing. Image processing methods used for proposed system are noise filter, Harris corner and Ransac algorithm. The integrated Kalman filter of proposed system estimates INS error and the estimated errors are fed back to the SDINS. Proposed system is validated through simulation and AUV test. The simulations and experimental results show that vision/low-cost INS integrated navigation system give performance as much as high cost INS