Determination of navigation parameters of dynamic objects based on adaptive sensor integration in intelligent transport systems
Purpose or research . The aim of the study is to develop and analyze the architecture of a local navigation subsystem for unmanned aerial and ground transport platforms based on adaptive integration of heterogeneous sensor data. The research focuses on improving the accuracy of estimating navigation parameters under conditions of degraded or unavailable satellite navigation signals. Methods . The methodological basis includes system analysis, probabilistic state-estimation methods, dynamic and observation models, and adaptive filtering approaches. To combine data from inertial measurement units, satellite navigation systems, and coordinate sources of an intelligent transport infrastructure, the extended Kalman filter and particle filter are employed, with automatic switching based on the statistical analysis of measurement residuals. Results . An architecture of the positioning subsystem is formulated, incorporating data synchronization, preprocessing, and an adaptive algorithmic core. It is demonstrated that switching to the particle filter when normalized residuals exceed confidence thresholds makes it possible to compensate for GNSS outliers and temporary signal interruptions. Simulation results show a reduction of the root-mean-square positioning error from 4.2 m to 2.1 m and a limitation of maximum deviations to 2.4 m. Various operating modes of the subsystem under different combinations of available navigation sources are analyzed. Conclusion . The proposed method of adaptive sensor integration enables reliable estimation of navigation parameters of dynamic objects and can be used within control systems of unmanned aerial platforms and intelligent transport complexes. The architecture and algorithmic solutions provide the required accuracy and fault tolerance when operating in challenging urban environments.