Real-time implementation of INS/GPS integration for vehicular navigation applications utilizing artificial intelligence.

摘要

Most of the present vehicular navigation instruments comprise global positioning system (GPS) and inertial navigation system (INS) and are widely used in several civilian and military applications. The integration of both systems results in reliable navigation system of superior accuracy, if compared to a stand alone INS or GPS. Several techniques, predominantly based on Kalman filtering (KF), are presently used for INS/GPS integration. This research aims at: (1) investigating new methodologies based on artificial intelligence (AI) techniques including neural networks and fuzzy logic systems to replace KF; (2) implementing adaptive neuro-fuzzy inference system (ANFIS) and radial basis functions neural networks (RBFNN) in real-time for vehicular navigation applications. In this study, the AI-based module (ANFIS or RBFNN) is designed to work in real-time to provide a continuous position update of the INS outputs using GPS position components. During this update procedure, the AI-based module is trained to mimic the latest vehicle dynamics so that it can provide the vehicle position accurately during GPS outages. (Abstract shortened by UMI.).