Objective  To propose a whole area coverage strategy of agricultural robot in complex farmland environment, and reasonably plan the working traversal path of agricultural robot.

Method  The complex farmland working environment model was defined according to the actual production environment of agricultural robot, and the concepts of first-level partition and second-level partition were established. The idea of genetic algorithm mutation operation was introduced to establish a high-quality feasible solution generation method of simulated annealing algorithm based on greedy mechanism. Based on the establishment of the concept of solution set diversity, an improved method of simulated annealing algorithm based on adaptive heating was designed to solve the problem of the optimal traversal sequence between partitions. The A* algorithm was combined with the eight-neighbor search method to plan the cross-regional connection path of agricultural robot. By this way, the scheme designed in this paper could achieve that the robot covered the whole working area.

Result  The simulation results showed that, compared with the traditional genetic algorithm and simulated annealing algorithm, the path length planned by the improved simulated annealing algorithm was reduced by 14.7% and 10.1% respectively, and the number of iterations during convergence was reduced by 9.8% and 59.1% respectively. The repeating rate of the traversal path of the agricultural robot in the simulation test of whole area coverage was 14.86%. The path repetition rate in the field traversal test of the high ground-clearance spraying robot was 15.83%.

Conclusion  The research results can provide a research idea for the full traversal coverage of agricultural robot in complex farmland environment.