Navigating unstructured environments poses challenges for mobile robots. Because such environments are partially observable, ground mobile robots must consider both global traversability and local obstacles. In this paper, we proposed a subgoal-based hierarchical motion planning framework for mobile robots to navigate unstructured environments. This framework comprises a global path planner, a subgoal planner, and a local path planner. This framework enables the mobile robot to effectively avoid local obstacles that cannot be represented on the global map. The hierarchical framework plans subgoals in local coordinates to avoid obstacles when following the planned global path. The local planner utilizes subgoals to generate local paths. To evaluate the proposed framework, we constructed a simulated unstructured terrain environment with diverse obstacles. The hierarchical framework exhibited better and safer quantitative performance than a conventional global path-based method, and the mobile robot successfully navigated unstructured environments by using the hierarchical framework while avoiding obstacles in local coordinates in real time.