Application of interval and fuzzy techniques to integrated navigation systems

The paper deals with the development of a new algorithm to be used by an INS (Integrated Navigation System) for carrying out accurate position estimation for different types of surface vehicles, including cars and ships. The proposed algorithm combines a neuro-fuzzy Kalman filter with a map matching method, in order to improve the effective real-time system performance when a GPS (Global Positioning System) is fused together with low-cost hardware sensors, such as an odometer and a piezoelectric gyroscope. The possibility of improving the overall numerical reliability of the estimation algorithm by means of an interval arithmetic implementation of the Kalman filter, is briefly outlined. Some experimental results are presented, indicating that rather good performance can be achieved by using the proposed system for estimating the position of a car inside a city route under normal traffic conditions