Validation of an AI-Assisted Terrain Aided Navigation Algorithm Using Real-World Flight Test Instrumentation Data

This study introduces an enhanced artificial intelligence (AI)-assisted Terrain Aided Navigation (TAN) for a sophisticated jet trainer, building upon our prior research \cite{dtsdasc} by incorporating real-flight test validation. The proposed TAN integrates a high-performance terrain server, a digital elevation model, and an efficient line-of-sight algorithm to facilitate terrain-aided navigation. The system utilizes an advanced search algorithm in conjunction with two filter designs, including adaptive filters that dynamically optimize navigation precision and operational efficiency. A significant development is the AI model's capacity to independently alternate between the resource-intensive search algorithm and a set of filters, thereby maintaining navigational accuracy while facilitating in-flight execution without supplementary hardware requirements. Comprehensive Monte Carlo calculations, validated by flight test instrumentation (FTI) data, indicate that the proposed TAN consistently facilitates low-altitude navigation across diverse operational settings. The incorporation of actual flight data not only substantiates the system's efficacy but also offers novel perspectives on practical implementation obstacles and improvements. These findings signify an advancement in autonomous terrain-aided navigation, connecting simulation with actual flight performance.