Position Estimate of Off-Road Vehicles Using a Low-Cost GPS and IMU

A fusion system of a low cost Garmin GPS 17N with IMU was developed for positioning an off-road vehicle. A Kalman filter was designed to integrate data from both sensors. By tuning the measurement covariance matrix, the different estimated performance was obtained. For the case of reducing bias, a constant covariance matrix was used. Measurement and fusion results showed that, at a vehicle speed of about 1.34 m/s, the mean bias in the East axis of the fusion system was 0.42 m comparing with the GPS mean bias of 1.29 m, and the mean bias in the North axis was reduced to 0.31 m from 1.48 m. The heading angle drift error of IMU was also compensated. For the case of smoothing path in parallel tracking, a variable covariance matrix was adapted. The GPS signals were smoothed out. While the vehicle moved along North-South at a speed of about 1.34 m/s, the standard deviation in the East axis of the fusion system was 0.03 m comparing with the GPS standard deviation of 0.06 m. For both cases, the redundant information was fused and the update frequency was increased to 9 Hz from 1 Hz of the GPS. In particular, the fusion system could still maintain its position while the interval of the GPS signal outage was 20 seconds. A prototype system was installed on a sprayer for vehicle positioning measurement.