Visual navigation of rape direct seeder unit can improve the intelligent and reduce labor intensity,so a visual navigation controller which was made up of fuzzy controller and PD controller with dead zone was proposed.The fuzzy controller was used to track path,and PD controller with dead zone control the front wheel steering.Through the two wheel ackerman steering model and camera imaging model,the law of image detection path changing with the rape seeder unit pose changes was got.Through the law,fuzzy controller rules can be designed.Image target path in the image was determined when the rape direct seeder distance the target ditch in the right place.And,the angle error and intercept error between image real-time detection path and image target path were the input of fuzzy controller.The output of fuzzy controller was frontwheel target steering angle.PD controller with dead zone was used to control front wheel angle.The field test results show that:the maximum lateral deviation was less than 6 cm when the speed of rape seeder was 0.5 m/s and 0.8 m/s,and the maximum lateral deviation was less than 10 cm when the speed is 1.0 m/s.