Operation of ground vehicles in remote unstructured terrains is a challenging task. While information on terrain parameters such as elevation, slope, and soil content and its use in tasks such as path planning and vehicle simulation is immensely useful, it requires detailed analyses that are computationally expensive and time-consuming. At the same time, higher-level decisions made without a full understanding of the terrain properties and vehicle capabilities are bound to be suboptimal, possibly even leading to mission failure. In this paper, we approach the problem from a decision-making standpoint to address both the aspects described above. The approach divides a given terrain into smaller cells and progressively finds the path to the goal. At each step, the operator’s utility between competing immediate terrain cells is compared so that the vehicle chooses the best next step considering detailed vehicle simulation. To account for the increased computational effort, a pool-based distributed computing architecture is employed to speed-up pathfinding and provide resilience towards failure of processors and communication links. The pool architecture also admits heterogeneous, geographically distributed processors and storage locations. Concurrently, the multiattribute utility formulation takes into account the tradeoff between different attributes and the uncertainty in the vehicle performance. The proposed method can be used for medium and long-range navigation in unstructured environments for efficient rough path planning or re-planning based on changing mission requirements and dynamically arriving terrain data. The method is demonstrated on data acquired from USGS sources.