The problem of grasping is widely studied in the robotics community. This project focuses on the identification of object graspable features using images and object structural information. The primary aim is the creation of a framework in which the information gathered by the vision system can be integrated with automatically generated knowledge, modelled by means of fuzzy description logics.