Enhanced Inertial Navigation System Accuracy Through Statically Calibrated MEMS Accelerometer Drift Compensation via Deep Learning Feature Extraction

**Abstract:** This paper introduces a novel approach to improve the accuracy of Inertial Navigation Systems (INS) by utilizing deep learning for feature extraction and static calibration of MEMS accelerometer drift. Traditional static calibration methods often struggle to capture the complex, non-linear drift behavior of MEMS accelerometers. This research proposes a system employing a convolutional neural network (CNN) to extract salient features from time-series accelerometer data acquired during a stationary calibration period. These features are then utilized by a regression model to predict and compensate for residual drift error, resulting in improved INS accuracy. The proposed method demonstrates a substantial reduction in INS error propagation compared to conventional calibration techniques, offering a practical and computationally efficient solution for enhanced navigation performance.