Abstract: Off-road path planning plays a strategically important role in special mission scenarios such as emergency rescue. However, existing algorithmic models often suffer from low accuracy in passability modeling and poor computational efficiency in complex terrains. To improve planning accuracy and efficiency, this study introduces the Soil Moisture–Strength Prediction model (SMSP II) to estimate the Rated Cone Index (RCI) of soil, and combines it with the Vehicle Cone Index (VCI) to construct a traversability grid map that integrates multiple constraints and soil passability indicators. On this basis, an improved NSGA-II algorithm is proposed, incorporating a divide-and-conquer strategy and heuristic search, and is further integrated with the Dijkstra algorithm to build a multi-level hybrid path optimization framework. Experimental results demonstrate that the proposed method improves computational efficiency by approximately 45%, enhances path passability by 2%, and reduces path length by about 2.1%, while maintaining path feasibility. The findings verify the superior performance of the proposed technical framework for off-road path planning in complex geological environments.