In order to improve the performance of autonomous navigation system of direct seeding machine in paddy field and other harsh environments, an optimal control method was proposed. It can adaptively adjust PD control parameters based on fuzzy logic. According to the kinematic model of chain system, the state space model was established. Then, according to the optimal control theory, the PD controller was designed. The range of parameters of Kp was from 0.8 to 2.1 and Kd was from 1.0 to 3.5, which were determined by the field experiment. Finally, the fuzzy adaptive optimal control method was put forward. Simulation result showed that the fuzzy adaptive optimal control method can improve the stability and rapidity compared with the PD controller. The experiment results showed that the mean absolute lateral deviation of the method was less than 0.021m on the cement pavement. And the mean absolute lateral deviation of the method in paddy field was less than 0.040m. The method can effectively improve the stability and rapidity of the autonomous navigation control system compared with PD controller and pure pursuit controller.