In recent years, the quality of high resolution acquisition platforms for geomatic applications has decidedly increased, especially in the performance of the sensors devoted to image acquisition.

The small size of these new sensors combined with the increase in resolution allows them to be easily mounted onto Unmanned Aerial Vehicles (UAVs); in particular, calibrated, very-high-resolution digital cameras for photogrammetric purposes such as digital terrain model (DTM) and digital surface model (DSM) extraction, orthophotos, and map realization have been used. In this case, the UAV performance allows a high quality product to be obtained, considering the pixel size and the accuracy of the DTM/DSM which could be obtained with automatic procedures.

Several navigation sensors (GPS/GNSS and IMU-MEMS) are embedded into UAVs in order to realize autonomous flight. The quality of these sensors, in terms of accuracy, depends on the model of UAV and its purpose. The navigation solution (position and attitude) is estimated by the internal UAV sensor and can be employed to directly georeference the images, in order to produce an quick and easy description and analysis of the overlooked area. This is a good condition for semi-automatic procedures using a bundle-block photogrammetric approach. But is it possible to realize a direct photogrammetry? And what are the limits?

Several tests were carried out over different areas and in different conditions using three different UAVs belonging to the Geomatics group of the Politecnico di Torino.

In this paper, an investigation of the limits of some commercial UAVs is reported, defining a dedicated procedure to valuate their performance, especially considering the use of UAVs for direct photogrammetry.