Path planning is the premise of greenhouse robot operation, and an optimal continuous barrier free path is planned with great significance. An algorithm based on the combination of improved A* algorithm and dynamic window approach was proposed to solve real-time path planning of mobile robot in greenhouse. The core was based on the search algorithm of traditional A* algorithm. Aiming at the problem of too many inflexion points, the key point selection strategy was improved. The dynamic window method was integrated to construct a global optimal path evaluation function. Local obstacle avoidance was achieved through ultrasonic sensors to achieve real-time optimal path planning. The simulation results showed that compared with the traditional A*, Dijkstra, RRT algorithms, the improved A* algorithm had a smoother path and higher efficiency, which was conducive to the motion control of the robot in the greenhouse, which showed the effectiveness of the algorithm. Considering the size of the robot, the grid map of the real environment was expanded to ensure the safety of the path. The experimental results showed that the fusion algorithm can satisfy the smoothness of path and effectively avoid obstacles. The mobile robot can achieve autonomous navigation, and the tracking error was kept within 0.22m, and the positioning error was no more than 0.28m, which met the actual needs. The research result had an important reference value for the application of greenhouse mobile robot navigation.