An Examination of the Relative Merits of Various Sensors for Vehicle Navigation

Eric Abbott and Prof. J. David Powell

Department of Aeronautics and Astronautics
Stanford University

ABSTRACT

The research described in this paper is aimed at providing a quantitative measure and a qualitative understanding of the contribution that each sensor in a land-vehicle navigation system makes to the system's overall performance. This includes a quantitative evaluation of the relative merits of various sensors typically used in existing land-vehicle navigation systems.

The results presented in this paper are based on data obtained from analytical studies and simulations. An analytical error model was derived for each sensor that was examined. These error models were subsequently used in a Kalman filter whose purpose was to generate an estimate of the vehicle's position. The Kalman filter was used in several Monte Carlo simulation studies of various system architectures. In addition, experimental data has been obtained from a prototype navigation system that has been installed in a test vehicle. These experimental data have been used to verify the analytical error models for several sensors.

Results show that the accurate calibration of a navigation system's heading and heading rate sensors to be of prime importance. It has been found that a rate gyro's scale factor exhibits poor observability characteristics, and the use of differential GPS fixes do not substantially improve the estimate of this quantity. In addition, it has been found that a rate gyro's bias error contributes more to positioning error than other rate gyro errors. Finally, we have found that, in the absence of absolute position measurements, the presence of a compass in the navigation system plays a significant role in reducing positioning errors.

a PostScript Version of this Entire Paper is NOT Available

Return to Publications List of the Stanford University GPS Lab