Improving GPS/INS Integration through Neural Networks

TL;DR

This paper presents an improved GPS/INS integration method that uses selective filtering before neural network processing to reduce computation time while maintaining accuracy in positioning data.

Contribution

It introduces a novel filtering approach prior to neural network processing to enhance speed without sacrificing accuracy in GPS/INS data integration.

Findings

Processing time is significantly reduced.

Positioning accuracy remains unchanged.

Filtering improves efficiency of neural network processing.

Abstract

The Global Positioning Systems (GPS) and Inertial Navigation System (INS) technology have attracted a considerable importance recently because of its large number of solutions serving both military as well as civilian applications. This paper aims to develop a more efficient and especially a faster method for processing the GPS signal in case of INS signal loss without losing the accuracy of the data. The conventional or usual method consists of processing data through a neural network and obtaining accurate positioning output data. The new or improved method adds selective filtering at the low-band frequency, the mid-band frequency and the high band frquency, before processing the GPS data through the neural network, so that the processing time is decreased significantly while the accuracy remains the same.