In recent years, ground autonomous mobile robots are widely used in public safety. Although the global path planning problem has been extensively studied in the past decades, there are still many problems in path planning for uneven terrain. In this paper, we first propose Matrix 3D A* algorithm, which improves the traditional A* algorithm by introducing parallel computing and matrix operations, and can significantly improve the solution speed of the path planning problem. Then, we introduced elevation information and the maximum feasible angle of the robot based on the Matrix 3D A* algorithm to achieve the solution of the path planning problem for uneven terrain. In addition, we improve the future estimation distance calculation method based on Chebyshev distance and Euclidean distance, and improve the heuristic weights, and the above improvements can offer three solutions: an accurate optimal path with the least movement cost, a greedy solution path in a relatively short time, and a path that is sub optimal in its result but more efficient in its solution process. Finally, we verified the soundness of the proposed algorithms by performing experiments on real environments.