Abstract: Abstract: Measuring steering wheel angle in real-time normally requires accurate installation and reliability in a typical auto-pilot sensor system for agricultural machinery. There are mainly three types of steering wheel angle measurement at present. In the first type, the displacement sensors are widely used to measure the stroke of the steering cylinder, thereby indirectly obtaining the steering angle. In the second type, the resistive or magnetic angle sensors (hall-type) are used to measure the kingpin rotation angle, and then geometric relations are used to calculate the steering angle. In the third type, a gyroscope was selected to measure the angular rate of the steering axle for the angle increment. Nevertheless, all these types need to be updated, when connecting to the navigation controller of a tractor. The first type requires easy installation and high reliability. The second type has high detection accuracy, but also needs easy installation. Since the third type is simple to install, the angle accuracy depends mainly on the gyro's zero drift, random drift and cumulative error. In this study, a measuring system of steering wheel angle was proposed using the combination of GNSS attitude and motor encoder. In this system, there was no need to install sensors on the steering wheels, or connect to the navigation controller. Firstly, the tractor attitude and ground speed were captured with a high precise GNSS terminal (MC4), and then the reference point of the vehicle precisely compensated the ground speed under the lever arm compensation with the roll and heading of a tractor. A dynamic motion model was selected to roughly calculate the wheel steering angle of the vehicle using the measuring system of steering angle. Secondly, a transfer model was established to real-time measure the steering wheel angle rate using the encoder of the electric wheel, thereby obtaining the steering axle angle rate, particularly considering the linear features of full hydraulic steering gear and the noise of hydraulic leakage. Thirdly, a Kalman filter was used to obtain a highly accurate steering wheel angle, where the rough angle was selected as the observed value, whereas, the steering axle angle rate was used to estimate the steering angle. A static hydraulic steering experiment was conducted to identify the transfer model of the steering wheel, where the LX1204's transfer coefficient of hydraulic steering was 0.046 542. The field tests included dynamic steering and path tracking. A fixed motor and a ping-pong tuning wheel were used for driving in the dynamic steering tests with the specific speed of 5, 10, 15 and 20 r/min. The results showed that the standard deviation of output angle in the Kalman filter was less than 1°, compared with the angle sensor in the hall effect. Path tracking experiments included 3 times straight-line steering in mid-/high-speed, and 4 times curve path steering. Specifically, the standard deviation was less than 0.91° in straight line tracking, while the deviation was 2.56° in the curve path tracking, and the largest measuring error was about 5° during the tractor wheel angle beyond ±20°. It inferred that the measuring system presented a better accuracy, when driving in a straight line, compared with the hall effect angle sensor. The navigation error of the auto-pilot system was less than 2.5 cm in the straight-line steering, while the navigation error was about 9.0 cm in the curve path tracking. Consequently, this angle measuring system can be used in the straight-line steering for the field work, or auto headland turn and harrowing in modern agriculture.