Abstract: Automatic navigation technology for agricultural vehicle can improve position accuracy of the vehicle effectively and realize large-scale production of cropland. An electric bicycle was adopted as experimental platform and system performance with two positioning methods of automatic navigation respectively was analyzed, the one was GPS positioning method and the other was the method based on a fusion of two data source (RTD DGPS and digital compass) by Kalman filter. RTK DGPS and RTD-DGPS were equipped on the vehicle to collect position data, and the data of RTK DGPS was used as standard trajectory. Synthesized weighting values were calculated, which were 0.006 and 0.002 for the GPS positioning method and the fusion method by Kalman filter respectively. The results showed that the fusion method by Kalman filter had a higher accuracy and a better stability than GPS positioning method.