In order to realize the remote dispatching management of multi-machine cooperative navigation operation in complex farmland environment, the research of global path planning based on improved A-star algorithm and Bezier curve was carried out. Multi-machine cooperative operation path planning in farmland operation environment was introduced, which was divided into global path planning and local dynamic obstacle avoidance; The improved A-star algorithm was used to optimize the global path and corner optimization, and the global path was smoothed based on Bezier curve; According to the randomly generated obstacle environment map and Zhuozhou Experimental Farm environment map, the multi-machine cooperative global path planning algorithm was simulated by using Matlab platform. The simulation results showed that by adjusting the value of weight w(n) in the improved A-star algorithm, the search efficiency was significantly improved, the running time of the optimized algorithm was 0.832s in the simulation experiment based on Zhuozhou Experimental Farm. Through corner optimization, the number of turns was effectively reduced when the path length was similar. Similarly, after using Bezier curve to smooth the path, the peak at the corner was well optimized to ensure the smooth progress of agricultural machinery in the actual farmland operation, which preliminarily met the real-time and smooth requirements, and provided a basis for further solving the multi-machine cooperative path planning in the field environment.