A Hierarchical Shape Representation for Vision-Guided Robotics

Using an adequate representation is often the key to solve complex problems in Artificial Intelligence. Hierarchical shape representations are very convenient in domains-such as vision-based robotics- that require a trade-off between ef-ficiency and accuracy. In this paper, we present new tech-niques and results concerning a hierarchical shape represen-tation based only on spheres. We focus on the global aspects of the model that are relevant to AI applications and present a summary of its main features and the employed proce-dures. Since the underlying object model is the generalized cylinder, the representation lends easily itself to its utiliza-tion in conjunction with a vision system