Model based GPS/INS integration for high accuracy land vehicle applications

In this paper, we consider the problem of trajectory reconstruction for the mobile mapping system Davide. Davide is equipped with a custom GPS/INS system and a number of dead-reckoning sensors, such as odometers, inclinometers, etc. The application falls in the framework of state reconstruction with multi-sensor architectures, for which GPS/INS integration is a paradigmatic problem. In this paper, to maximize the accuracy of trajectory reconstruction, we propose to use a combination of two strategies, namely: a) to use a customized dynamical model of the vehicle, that includes nonholonomic constraints and load transfer in roll and pitch, and b) to use advanced nonlinear filtering techniques such as Montecarlo filters. The performance of proposed algorithm is illustrated by means of both simulations and real vehicle data.