Sensor fusion technique has been commonly used for improving the navigation of autonomous agricultural vehicles by means of combining complimentary sensors mounted on such vehicles for the position and attitude angle measurements. In this research, sensor fusion via an Extended Kalman Filter (EKF) was used to integrate the attitude angle estimates from the Digital Elevation Models (DEMs) and Terrain Compensation Module (TCM) sensor to improve the roll and pitch angle measurements of a self propelled sprayer. The fusion algorithm was also developed to improve the three-dimensional positioning of the sprayer, in particular the elevation measurements of a GPS receiver mounted on the sprayer. Vehicle attitude and field elevation were measured at two speeds, 5.6 km/h and 9.6 km/h, using a set of onboard sensors including a real-time kinematic-differential GPS receiver (RTK-DGPS), a TCM sensor and an Inertial Measurement Unit (IMU). A second order auto-regressive (AR) model was developed to model the TCM roll and GPS-based pitch errors. The derived error states were incorporated into the EKF algorithm and the measurement noise covariance was estimated from the AR model, which limited the fine tuning of noise covariance to the process noise covariance only.

The EKF estimations were compared with the IMU measurements to validate the performance of the developed fusion algorithm. For the slow speed test data, the mean and standard deviation of the errors of roll (Mean: -0.2244º, Std. Dev.:1.471º) and pitch (Mean: 0.0597º, Std. Dev.: 0.6621º) from the EKF estimates were reduced considerably compared to that of the errors of roll (Mean: 0.2157º, Std. Dev.: 2.4610º) and pitch (Mean: 0.0473º, Std. Dev.: 1.3230º) from DEM. Medium speed test data also showed considerable improvement in the attitude angles estimated using the developed EKF algorithm. The fusion algorithm for improving the elevation measurement of the GPS also showed promising results. Thus, the fusion algorithm was effective in improving attitude and the navigational accuracy of the self-propelled agricultural sprayer, which in turn will also facilitate the automatic control of the implements that interact with the soil surface on undulated topographic surfaces.