Enhancing Localization Accuracy and Reducing Processing Time in Indoor Positioning Systems: A Comparative Analysis of...

Area of Science:

• Artificial Intelligence
• Robotics
• Sensor Fusion

Background:

• Indoor positioning systems (IPS) are crucial for mobile tracking and navigation.
• Existing IPS often face challenges with accuracy and processing speed.
• AI models offer potential for enhanced IPS performance.

Purpose of the Study:

• To comparatively evaluate AI models for indoor positioning.
• To assess localization accuracy and processing time of different AI algorithms.
• To identify optimal AI models for mobile tracking applications.

Main Methods:

• Comparative analysis of Artificial Neural Networks (ANNs), Long Short-Term Memory (LSTM), Recurrent Neural Networks (RNNs), and Kalman filters.
• Utilized Received Signal Strength Indicator (RSSI) and 9-axis ICM-20948 sensor data.
• Implemented data cleaning and feature selection techniques for error reduction.

Main Results:

• The Recurrent Neural Network (RNN) model demonstrated the best performance.
• Achieved a localization error of 0.247 m with a processing delay of 0.077 s.
• Evaluated performance within a 12 m × 9.5 m area using four anchors.

Conclusions:

• Recurrent Neural Networks (RNNs) are highly effective for indoor positioning.
• Model selection based on test and validation data is critical for effective mobile tracking.
• AI advancements significantly improve localization accuracy and efficiency in IPS.