Aerospace Technic and Technology

The subject matter of the article is the processes of functioning and intelligent error correction of strapdown inertial navigation systems (SINS) for transporting unmanned aerial vehicles (UAVs) under conditions of limited accessibility or complete absence of GNSS signals. The goal is to perform a comprehensive systems analysis of instrumental error sources in MEMS sensors and to provide scientific justification for the effectiveness of applying deep machine learning methods combined with dynamic certification technology based on a Stewart platform to enhance the accuracy and reliability of autonomous navigation. The tasks to be solved are: to systematize the main types of errors in micro-electromechanical systems (MEMS) within SINS; to perform a comparative analysis of traditional (Kalman, Madgwick filters) and modern intelligent signal processing methods; to justify the feasibility of transitioning from static calibration procedures to "board-in-the-loop" dynamic certification using precision equipment; and to conduct mathematical modeling of the impact of uncorrected navigation errors on the movement safety of transport UAVs within layered airspace. The methods used are: analytical review of scientific sources; inertial navigation theory; synthesis of convolutional (CNN) and recurrent (LSTM) neural network architectures; methods for mathematical modeling of flight kinematics in the Python environment; and methods for dynamic sensor certification using a hexapod (Stewart platform). The following results were obtained: a detailed analysis of factors destabilizing MEMS-SINS accuracy was conducted, identifying gyroscope zero-drift, thermal dependence of scale factors, and the impact of vibrational noise from the propulsion system. It was shown that linear filtration models do not achieve sufficient accuracy under high flight dynamics and the complex noise characteristics of low-priced segment sensors. A new concept of dynamic calibration in the "board-in-the-loop" state on the UAV was proposed, enabling consideration of structural elasticity and the specific installation features of a particular vehicle. The technology's hardware foundation is a precision Stewart platform that recreates complex spatial movements with six degrees of freedom and high accuracy, simulating real turbulence and maneuvering using a multidimensional shaping filter. Based on the obtained data, an individual parametric profile (an "error passport") was formed, integrated with adapted neural network correction methods. To assess the criticality of navigation errors, the concept of a "safety corridor" was introduced and tested as an integral indicator of system reliability. Conclusions. The scientific novelty of the results obtained is as follows: 1) was established that the integration of dynamic certification technology based on a hexapod and LSTM recurrent networks allows maintaining the UAV within the safety corridor for up to 10–12 minutes of autonomous flight, which is 3–4 times higher than the performance of traditional static calibration methods; 2) was determined that the most promising direction for onboard software development is the use of CNNs for primary noise filtering and LSTMs for time-drift compensation; 3) the practical significance of the research lies in the possibility of ensuring stable autonomous navigation for transport UAVs in Electronic Warfare (EW) environments and during precision logistics operations without GNSS support; 4) the use of Python as an integration environment allowed for the creation of a seamless information loop between the hardware Stewart platform and intelligent data processing algorithms. 

Mwenegoha, H., Moore, T., Pinchin, J., Jabbal, M. Model-Based Autonomous Navigation with Moment of Inertia Estimation for Unmanned Aerial Vehicles. Sensors, 2019, vol. 19, iss 11, article no. 2467, pp. 1-22. DOI: https://doi.org/10.3390/s19112467.

Kukla, O. Pohibki avtonomnoyi navigaciyi u nevelikih bezpilotnih litalnih aparatah ta metodi yih podolannya [Autonomous navigation errors in small unmanned aerial vehicles and methods for overcoming them]. Viprobuvannya ta sertifikaciya - Testing and certification, 2026, no. 1 (11), pp. 99-107. DOI: https://doi.org/10.37701/ts.11.2026.11. (In Ukrainian).

Wang, D., Dong, Y., Wu, J., Southey, R. Constrained MEMS-Based GNSS/INS Tightly Coupled System With Robust Kalman Filter for Accurate Land Vehicular Navigation. IEEE Transactions on Instrumentation and Measurement, 2020, vol. 69, iss. 7, pp. 5138–5148. DOI: https://doi.org/10.1109/TIM.2019.2955798.

Shen, C., Zhang, Y., Liu, J., Cao, H., Tang, J., Li, X., Shi, J. Dual-optimization for a MEMS-INS/GPS system during GPS outages based on the cubature Kalman filter and neural networks. Mechanical Systems and Signal Processing, 2019, vol. 133, pp. 1-15. DOI: https://doi.org/10.1016/j.ymssp.2019.07.003.

Shahrawy, A., Shawky, M. A., Soliman, A. M., Khan, W. U., Almogren, A., Abdellatif, A. G., Shah, S. T. Breaking through GNSS outage: advanced stochastic model for MEMS IMU in navigation. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2025, vol. 18, pp. 1-15. DOI: https://doi.org/10.1109/JSTARS.2025.3581379

Abouelsoud, A. A., Namerikawa, T., Fath El Bab, A. M. R., Maged, S. A. Stewart Platform Manipulator: State Estimation Using Inertia Sensors and Unscented Kalman Filter. Proceedings of the 2016 3rd International Conference on Information Science and Control Engineering (ICISCE), Beijing, China, 2016, pp. 1136-1140. DOI: https://doi.org/10.1109/ICISCE.2016.244.

Titterton, D. H., Weston, J. L. Strapdown Inertial Navigation Technology. 2nd ed. London, The Institution of Engineering and Technology, 2004. 558 p. DOI: https://doi.org/10.1049/PBRA017E.

El-Sheimy, N., Hou, H., Niu, X. Analysis and Modeling of Inertial Sensors Using Allan Variance. IEEE Transactions on Instrumentation and Measurement, 2008, vol. 57, iss. 1, pp. 140-149. DOI: https://doi.org/10.1109/TIM.2007.908635.

Saboor, A., Islam, M. K., Soh, W. Comparison of Madgwick and Kalman Filter Based Orientation Estimation Algorithms for Low-Cost IMUs. International Journal of Navigation and Observation, 2020, vol. 2020, article ID 3945378, pp. 1-12. DOI: https://doi.org/10.1155/2020/3945378.

Brossard, M., Bonnabel, S., Barrau, A. AI-IMU Dead-Reckoning. IEEE Transactions on Intelligent Vehicles, 2020, vol. 5, iss. 4, pp. 585-595. DOI: https://doi.org/10.1109/TIV.2020.2980758.

P. Virtanen, R. Gommers, T. E. Oliphant, M. Haberland, T. Reddy, D. Cournapeau, et al. SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python. Nature Methods, 2020, vol. 17, pp. 261-272. DOI: https://doi.org/10.1038/s41592-019-0686-2.

Abadi, M., Barham, P., Chen, J., Chen, Z., Davis, A., Dean J. TensorFlow: Large-Scale Machine Learning on Heterogeneous Systems. arXiv preprint, 2016. 19 p. DOI: https://doi.org/10.48550/arXiv.1603.04467.