This paper presents an innovative approach to enhance the navigation of UAVs in GNSS denied environments. Considering the limited space, power, and size of small UAVs, the proposed approach does not require any sensors on the UAV. The typical repeated dynamic patterns of such UAVs are related to the actuators to offer a useful information for estimating the UAV navigation states. Machine learning (ML) classifier has been employed to detect these repeated dynamic patterns, then according to the detected pattern, an appropriate constraint/update is utilized to enhance the navigation solution through EKF to obtain a better estimate of the UAV states. Different test scenarios where conducted to verify the ability of the proposed approach to aid the INS solution during GNSS signal outages. The solution after fusing the Vheicle Model (VM) is enhaced by 98% compared to low cost stand-alone IMU solution.