AfKh-OpenIMU Generator: An open-source platform for simulating INS grades and generating datasets for machine learning applications

Mohammed Aftatah, Abdelhak Khalil and Khalil Zebbara

Data and Metadata, 2025, vol. 4, 1150

Abstract: This paper presents AfKh-OpenIMU Generator, an open-source platform developed in Java and deployed via Docker, aimed at supporting machine learning research in inertial navigation. The objective is to address two major challenges: the high cost of inertial navigation systems (INS) and the lack of large, labeled datasets required for training neural networks. Our platform simulates all six inertial sensors, including three accelerometers and three gyroscopes, using configurable error models that incorporate bias, scale factor, and stochastic noise. From user-defined reference trajectories, it generates raw inertial data and supports large-scale data augmentation by varying noise profiles, enabling the creation of diverse datasets without requiring physical hardware. Simulation results demonstrate high fidelity with real-world INS performance. The generated data yielded root mean square error (RMSE) values of 32.98 meters for low-cost INS, 8.99 meters for industrial-grade INS, and 1.07 meters for tactical-grade INS. In addition, the data augmentation mechanism allows dataset expansion by up to 10,000 times, significantly enhancing training robustness and helping to prevent overfitting in deep learning models. Our platform provides a flexible, low-cost, and reproducible solution for generating realistic inertial data. It facilitates the development and evaluation of machine learning algorithms for sensor fusion and secure navigation, making it particularly valuable for research in GPS-denied environments.

Date: 2025
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:dbk:datame:v:4:y:2025:i::p:1150:id:1056294dm20251150

DOI: 10.56294/dm20251150

Access Statistics for this article

More articles in Data and Metadata from AG Editor
Bibliographic data for series maintained by Javier Gonzalez-Argote ().