In order to improve the path tracking accuracy of intelligent agricultural machinery automatic navigation in the greenhouse, a distance determination method based on dynamic look ahead distance pure pursuit model and its path tracking control method was proposed. The position deviation and heading deviation of the intelligent agricultural machinery in the greenhouse are obtained using the Ultra Wide Band (UWB) module and the electronic gyroscope. In order to improve the adaptive ability of the pure pursuit model, a quantitative analysis of the posture deviation of the agricultural machinery and the fitness function is performed according to the deviation of the agricultural machinery's pose. In order to improve the efficiency of the algorithm, the inertia weight coefficient is improved. The particle swarm optimization (PSO) algorithm is used to determine the optimal look ahead distance in the pure pursuit model in real-time. The speed control function is constructed to control the variable speed of the agricultural machinery under different pose deviations.The prototype test results show that the average deviation is 24.4 cm, the average steady-state deviation is 4.3 cm, the average navigation time is 13.2 s, and the average stable distance is 318.1 cm when the linear path is tracked in the three initial states. The various indicators of path tracking are better than the constant speed and fixed line-of-sight test under the same conditions. The method proposed in this study has a good path tracking effect and can meet the needs of agricultural machinery for automatic navigation operations in the greenhouse.