Aerospace Instrument-Making Annotation << Back Methodology for Improving the Accuracy of Data Processing in on-board UAS Systems Antamoshkin O.A., Gulyutin N.N., Ermienko N.A., Kretinin V.V., Trukhanov E.V. This paper presents a comprehensive methodology for improving the accuracy of multimodal data processing in onboard computing systems of unmanned aerial systems (UAS). The methodology includes noise suppression, geometric image correction, the use of machine learning algorithms, and computational optimization through model quantization and hardware acceleration. The application of modern neural network architectures, such as CNN and PointNet, ensured high classifi cation accuracy. Experimental results obtained in both simulated and real-world environments demonstrated an accuracy increase of up to 94 %, along with a reduction in processing time and energy consumption, making the proposed approaches promising for practical use in monitoring and control tasks. Keywords: Unmanned aircraft systems, multimodal data, data processing, Kalman filter, neural network quantization, computational optimization. DOI: 10.25791/aviakosmos.10.2025.1511 Pp. 11-20.