ML-Based Inertial Navigation System Diagnosis for Underwater Vehicle

This paper investigates ML methods for diagnosing failures in an integrated Inertial Navigation System (INS) of the underwater vehicle ODIN UAV. Using dynamic equations of motion and considering measurement noise, a dataset was generated during normal operation and during various failures. Data preprocessing, exploratory analysis and feature engineering were conducted. To solve the problem of INS technical state classification, multiple machine learning models were evaluated and optimized using training data and techniques to reduce the number of dimensions. A comparative analysis of the obtained results was carried out. The paper contributes insights into effective ML-based diagnostics for enhancing navigation system reliability in autonomous underwater robots.