Position Estimation of Vehicle Using GPS Data and Internal Sensor Data

This paper deals with an estimation method of vehicle position using differential GPS, a fiber optic gyro and wheel encoders. The estimated position must be accurate enough to be used for vehicle control. To realize this accuracy, the following factors must be considered. (1) The delay and scarcity of GPS data. (2) The drift of the gyro. (3) The inconstancy of the travel distance per encoder pulse due to the differential mechanism and slip of wheels. (4) The effect of the position of the GPS antenna on position estimation. We propose an observation model of GPS, a fiber optic gyro and wheel encoders, which considers these four factors. A position estimation method is developed on the basis of this observation model. A prototype of autonomous mower was produced to evaluate the position estimation method. The relation between the performance of the mower and the GPS accuracy is discussed in terms of the parallelism, the uniformity of the space, and the straightness. The experimental results show that the merit of using GPS data is remarkable for maintaining the parallelism. If GPS is more accurate than lm, it is shown that using GPS improves the uniformity of the spaces. We can have the prospect for mowing all over the ground if the overlap is allowed to be 0.2m and the GPS accuracy is lm.