Development of a low-cost GPS/INS integrated system for tractor automatic navigation

The use of low-cost single GPS receivers and inertial sensors for auto-guidance applications has been limited by their reduced accuracy and signal drift over time compared to real-time kinematic (RTK) differential GPS units and fiber-optic gyroscope (FOG) sensors. In this study, a prototype low-cost GPS/INS integrated system consisting of a triangle-shaped array of three Garmin 19x GPS receivers and an Xsens inertial measurement unit (IMU) to improve the accuracy of position and heading angle measured with a single GPS receiver was developed. A triangular algorithm that uses data collected from the three single GPSs mounted on the angular points of a triangular frame was designed. A sensor fusion algorithm based on the Kalman filter combining the GPS and IMU data was developed by integrating position data and heading angles of a triangular array of GPS receivers. The optimized values of two noise covariance matrixes (Q and R) for the Kalman filtering were determined using the Central Composite Design (CCD) method. As compared to the use of a single Garmin GPS receiver, use of the developed GPS/INS system showed improved accuracy performance in terms of both position and heading angle, with reductions in root mean square errors (RMSEs) from 2.7 m to 0.64 m for position and from 8.9º to 2.1º for heading angle. The accuracy improvements show new potential for agricultural auto-guidance applications.