A Novel Hybrid NN-ABPE-Based Calibration Method for Improving Accuracy of Lateration Positioning System

Area of Science:

• Robotics and Automation
• Signal Processing
• Machine Learning

Background:

• Lateration positioning systems rely on distance measurements to known beacon locations for target object estimation.
• Accuracy is often compromised by factors like obstacles, signal reflections, and hardware imperfections.

Purpose of the Study:

• To propose a novel hybrid calibration method to improve the accuracy of lateration positioning systems.
• To address the challenges posed by real-world disturbances affecting measurement precision.

Main Methods:

• A two-step hybrid approach combining Apparent Beacon Position Estimation (ABPE) and Neural Networks (NN) was developed.
• Extensive testing of 16 NN architectures, 10 learning algorithms, and 12 activation functions was conducted in MATLAB.
• The method was validated in indoor scenarios with and without obstacles, and the impact of training sample size was analyzed.

Main Results:

• The hybrid NN-ABPE method demonstrated significant improvements in positioning accuracy.
• High accuracy was achieved even with a reduced number of calibration reference points.
• The influence of different uniform sampling patterns on system accuracy was quantified.

Conclusions:

• The proposed hybrid NN-ABPE method offers a robust solution for enhancing lateration positioning accuracy.
• This technique is effective in overcoming common disturbances and achieving precise localization in challenging environments.
• The method's efficiency in scenarios requiring fast system setup and minimal calibration data was confirmed.