Neural Network-Based Ambiguity Resolution for Precise Attitude Estimation with GNSS Sensors

Abstract

Accurate navigation and control of aerial vehicles necessitate precise estimations of both position and attitude. Measuring an aircraft's rotation entails comparing vectors in distinct reference frames, typically involving inertial and body axes. Traditionally, a GNSS sensor-based matrix, comprising a minimum of three sensors, is employed for this purpose, leveraging carrier phase measurements. Nevertheless, challenges like multipath interference, frequency lock loss, cycle slips, and significant clock drifts can impede the accurate resolution of integer ambiguities. To tackle these issues, a neural network-based approach has been developed to enhance the management of extensive data and bolster the reliability of carrier phase ambiguity resolution. By harnessing carrier phase differences and pseudorange information, diverse neural network configurations can be trained to resolve ambiguities and estimate the precise attitude of the GNSS sensor matrix. This solution can be deployed either in isolation or in conjunction with other attitude sensors, such as gyroscopic data, to enhance overall accuracy