Improving Accuracy of the Alpha-Beta Filter Algorithm Using an ANN-Based Learning Mechanism in Indoor Navigation...

Area of Science:

• Robotics and Artificial Intelligence
• Sensor Fusion and Navigation

Background:

• Indoor localization is crucial for location-based services but faces accuracy challenges with existing inertial measurement unit (IMU) techniques.
• Miniaturized sensors provide rich user context data, enabling smarter applications, yet traditional IMU-based indoor navigation methods exhibit significant accuracy and consistency issues.

Purpose of the Study:

• To present a novel solution for enhancing indoor navigation accuracy using a learning-to-prediction model.
• To improve the prediction accuracy of indoor navigation algorithms by employing an artificial neural network.
• To address the limitations of current IMU-based indoor localization systems.

Main Methods:

• Developed a learning-to-prediction model-based artificial neural network for indoor navigation.
• Utilized next-generation inertial measurement units (IMUs) with accelerometers, gyroscopes, and magnetometers for data acquisition.
• Implemented an artificial neural network-based learning module to optimize alpha-beta filter parameters (alpha and beta) for minimizing sensor reading errors.

Main Results:

• The proposed system effectively tracks object locations in indoor environments where GPS is unavailable.
• The artificial neural network improved the prediction accuracy of sensor readings.
• Experiments demonstrated that the alpha-beta filter integrated with the learning module significantly outperformed the traditional alpha-beta filter in terms of Root Mean Square Error (RMSE).

Conclusions:

• The learning-to-prediction model, powered by an artificial neural network, offers a significant improvement in indoor navigation accuracy.
• Optimizing the alpha-beta filter parameters through a learning module enhances the reliability and precision of indoor localization systems.
• This approach provides a robust solution for accurate indoor navigation, overcoming the limitations of conventional IMU-based techniques.