Abstract: Aiming at the problems of low coverage rate, high repetition rate and weak universality of traversal path planning when lawn mower robots operate in large areas, an improved traversal path planning algorithm combining A^* algorithm and DFS algorithm is proposed. First, according to the known global information of the environment, the target area is divided into multiple sub-areas without obstacles by the cattle farming decomposition method. Then, an undirected graph is constructed according to the adjacency relationship of the sub-regions, and the traversal order of the sub-regions is planned using the DFS algorithm. Finally, the improved A^* algorithm is used to transfer the path across regions and traverse the interior of each sub-region in a reciprocating manner. The simulation experiment results show that the coverage rate of the traversal algorithm reaches 100%, and the traversal repetition rate is 0. The length of the cross-regional transfer path and the number of turns planned by the improved A^* algorithm are reduced by 3.26% and 62.5% respectively compared with the A^* algorithm. The traversal algorithm proposed in this paper has the characteristics of high coverage, low repetition rate, and strong universality. Improved A^* algorithm improves the A^* algorithm by optimizing path smoothness and increasing the anti-collision safety distance to make the planned path smoother, safer and shorter path length. The research results are designed to provide theoretical reference for ergodic path planning of lawn mower robot.