The traditional multi-machine collaborative path planning method exists several problems which ignore the autonomy of the slave and the turning path may overlap on the ground. A multi-machine cooperative navigation path planning method oriented to master-slave cooperation mode was proposed. Firstly, the safety state detection model of agricultural machinery was established based on the directional bounding box algorithm and the separation axis theorem. By comparing the relationship between the sum of the projection radius of each agricultural machinery envelope section on the separation axis and the projection length from the geometric center of the section, the collision between agricultural machinery can be detected. Then, the host path planning model based on the full coverage algorithm was established, and the optimal operation direction angle was solved with the minimum turning number as the optimization objective; according to the size relationship between the minimum turning radius and operation width of the main engine, two kinds of U-shaped and T-shaped turning models were established. Finally, the path planning model of slave was established, and the linear operation path of slave machine was planned according to the relative distance between master and slave in cooperative operation; The cooperative turning modes of master and slave were divided into three types: double U-type, double-T-type and UT type. The cooperative turning strategy was proposed, according to the turning state of the master and slave, the waiting time of the slave was determined to avoid the collision risk caused by overlapping turning paths. Taking the wheat harvest scene under a convex polygon plot as the experimental sample plot, the longitude and latitude coordinates of the plot vertex were extracted by using LocaSpace Viewer, and the rectangular coordinates of the plot were obtained through coordinate transformation. The simulation experiment was carried out by using Matlab. The experimental results showed that the proposed method can plan multi-machine collaborative operation path with high land coverage, short operation time and low power consumption; the master and slave can turn successively according to the designed cooperative turning strategy to avoid collision, when the turning path overlaped; the minimum, maximum and average time of path planning algorithm were 0.453s, 1.563s and 0.951s, respectively. The proposed method can avoid collision risk and can provide an effective global operation path for wheat and silage harvesting.