Concerning autonomous navigation, autonomous vehicles or robots must be able to sense the world, create a map, and use it to localize itself and plan actions. Several approaches were proposed to achieve these goals and, until now, the most successful one was based on offline map construction, followed by online localization and navigation using the created map. The main advantage of this approach is the freedom to use time-consuming techniques to enhance map quality (e.g. loop closure correction, etc.), without  the risk of losing on-the-fly localization due to the lack of resources. In general, Simultaneous Localization and Mapping (SLAM) techniques are used to build the map. Proposed methods that rely on deterministic models design (deliberative) can became complex models aiming to represent the real world, especially to dynamic and semi-structured environments presenting an uncertainty degree, which is the case of agricultural areas. This study aimed to use DSM (Digital Surface Model) and Vegetation Index Maps of an unstructured environment combined with an A* algorithm to perform a path planning of an agricultural mobile robot. The areas used for data acquisition are located at University of São Paulo, Brazil. Aerial images, using an Unmanned Aircraft System (UAS), were acquired at areas of pasture and primary crops (2.1 to 5.7 ha) inside the Campus. Three GSD (Ground Sampling Distance) were used aiming to analyze the performance of the A* algorithm. Georeferencing was performed in an indirect manner, using Ground Control Points (GCP) for adjacent points position estimative. Georeferenced checkpoints were manually inserted in the model and the A* algorithm performed a path planning. The seam-line are computed using two methods: i) Color difference (image-based method) using the Index Maps, where the seam-lines are determined by the field variability; and ii) Height gradient (ground-based method), that uses the DSM data. In these methods the pixel color is related to the t difference between adjacent points. The seam-lines are determined by the largest height gradient.