Autonomous navigation of small UAVs is typically based on the integration of inertial navigation systems (INS) together with global navigation satellite systems (GNSS). However, GNSS signals can face various forms of interference affecting their continuous availability. For small UAVs, employing low-cost inertial measurement units (IMU), such GNSS outages lead to a rapid deterioration in the positioning accuracy, becoming beyond practical use within a minute, with potentially catastrophic outcomes in the case of beyond visual line of sight (BVLOS) operation. To address this issue, recent research has explored the integration of a vehicle dynamic model (VDM) as the primary process model in the navigation system of small UAVs, showcasing significant improvement in navigation performance during GNSS outages without requiring the implementation of additional sensors. While prior studies mainly focus on conventional UAV configurations, this thesis specifically targets delta-wing UAVs. Such platforms, appealing for a variety of applications due to their enhanced aerodynamic efficiency, present challenges such as highly coupled flight dynamics due to their underactuation. This research presents a comprehensive procedure for dynamic model characterization of delta-wing UAVs, utilizing a hybrid approach which combines open-air wind-tunnel experiments and the processing of real flight data using a filter error method. This approach allows for the characterization of i) platform aerodynamics, ii) propulsion system dynamics and iii) platform inertial properties, all crucial elements required for the implementation of a VDM navigation system. Candidate aerodynamic models are selected via step-wise regression, and numerical values for the model parameters are determined using two distinct methodologies. The performance of both parameter estimation approaches is then evaluated in a VDM-based framework through independent test flights, demonstrating a significant enhancement in positioning accuracy during GNSS outage when integrated with consumer-grade MEMS inertial sensors. It is also demonstrated that the benefits of such a VDM framework extend beyond improved positioning accuracy. Synthetic wind and angle of attack estimates derived from the VDM framework are validated along with the advantages of airspeed measurement redundancy. Furthermore, the general validity of the determined dynamic model is verified via a second delta-wing test platform, demonstrating its portability and the significant improvement in autonomous navigation that it produces. These investigations reinforce and expand the literature in the domain of model-based navigation, proposing novel approaches for the rapid and efficient characterization of new platforms.