This paper proposes a recursive receding horizon path planning algorithm for unmanned vehicles in nonuniform environments. In the proposed algorithm, the map is described by grids in which nodes are defined on corners of grids. The planning algorithm considers the map as four areas, namely, implementation, observation, explored, and unknown. The Implementation area is a subset of the Observation area, whereas the Explored area is the union of all the previous Observation areas. The path is planned with a receding horizon planning strategy to generate waypoints and in-between map updates. When a new map update occurs, the path is replanned within the current Observation area if necessary. If no such path exists, the search is extended to the Explored area. Paths can be planned by recursively searching available nodes inside the Explored area that can be connected to available nodes on the boundary of the Explored area. A robot platform is employed to conduct a series of experiments in a laboratory environment to verify the proposed path planning algorithm.