Integrated GPS/DR Vehicle Navigation System Based on Sequential and Square Root Kalman Filters

Abstract

Global Positioning System (GPS) has become part of many applications in life. In mountainous terrains and around buildings, GPS reception is compromised. In dense urban canyons, signals bounce off the buildings creating multipath reception and provide erroneous measurements. To overcome GPS bandwidth and signal fading problems, Navigation solutions are built on GPS measurements fused with inertial sensors to provide dead reckoning (DR) based position solution. Solution for land vehicle Navigation System using GPS, inertial sensor and odometer is presented. The sensors fusion is performed based on conventional, sequential (SKF) and square root Kalman (SRKF) filters. SRKF based on Cheolesky factorization for covariance matrix P. Simulations are performed on real data, with precisely known covariance’s to simulate mathematical stability, performance and processing time required by each method on a high end microprocessor. The results demonstrate integrated system using SRKF has better performance in stability and estimation accuracy than conventional and sequential filter.