Real-Time Mosaic-Aided Aerial Navigation: II. Sensor Fusion
AIAA Guidance, Navigation, and Control Conference(2012)
摘要
We present a method for fusion of computer-vision mosaic-image-based motion estimation with a standard navigation system, yielding mosaic-aided navigation that does not rely on any a-priori information. The mosaic-based motion estimation uses inter-relations among images captured in real time during ∞ight. This motion estimation is transformed into residual translation and rotation measurements, which are fed into a Kalman Filter, fusing the inertial measurement and the mosaic-based motion estimation. The proposed method can arrest and reduce the secular growth of inertial navigation errors, and correct measurements of the on-board inertial sensors. Moreover, we show that mosaic-aided navigation outperforms traditional vision-based aiding methods in challenging scenarios, such as ∞ight over low-texture scenes captured by a camera with a narrow flled-of-view. To validate the proposed algorithms, we carried out a comprehensive performance evaluation, including statistical simulation runs and experiments based on real imagery.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络