A GPS-Free Air Vehicle On-Board Positioning Mechanism Applying Feature Matching Technique
Chia-Ta Hsieh, Yushun Wang · 2024
In the contemporary era, Global Positioning System (GPS) technology has reached a high maturity, and its unparalleled convenience has led to the relegation of other technologies to secondary roles in navigation. However, the dependence of both military and civilian sectors on GPS increases the likelihood of GPS becoming a primary target in the event of conflicts. Anticipating potential loss of GPS functionality in future combat scenarios, there is a renewed emphasis on GPS-denied navigation. In this work, the authors propose replacing GPS with visual localization using image-based techniques such as offline map, image feature extraction, and image stitching to calculate location. The proposed image-based localization method is simulated and tested for its ability to handle rotation and scaling in real images. The simulation results indicate that this method performs good adaptability to the rotation and scaling of real images. In addition, considering the longer time required for offline map reading, an image localization optimization method is proposed to improve real-time performance. This method achieves real-time image localization functionality through the stitching of local maps. The simulation results show that after applying positioning optimization methods, the performance has improved 166 times compared to before. It is believed that this method can be practically applied in the development of GPS-denied positioning capabilities for unmanned aerial vehicles (UAV) or fighter jets, reducing operational risks and enhancing national defense security.