Sensor fusion and integration using an adapted Kalman filter approach for modern navigation systems

Modern navigation systems require the integration of different sensors that are either employed as primary navigation method (e.g. dead reckoning in car navigation) to provide position information at regular time intervals (e.g. satellite based positioning for providing a start position and regular absolute position updates) as well as additional backup sensors. In most common systems, however, sensors are mainly employed in a stand-alone mode and no integrated position determination is performed. In our study, a new sensor fusion model based on an adapted Kalman filter has been developed to obtain an optimal estimate from the measurements of all available sensors. The concept for the integration and combined position determination has been employed for the combination of observations of GPS, wireless or mobile phone location services (MPLS) and dead reckoning (DR) sensors employed in vehicle navigation systems. In the following, it has been used for pedestrian navigation and guidance. The results of simulation studies are presented in the paper. The concept can also be extended to vehicle navigation in dense high-rise urban environments, where positioning problems due to blockage of GPS signals could be solved either by DR prediction or updating with MPLS for higher positioning accuracy is expected to be achieved using the 3G network, and the combination of the two.