Using superquadrics for the approximation of objects and its application to grasping
Marcus Strand, Zhixing Xue, Marius Zoellner, Rüdiger Dillmann · 2010
Objects grasping and manipulation is a major task for housold and industrial robots. Sensing the objects with 3D laser scanners generating point clouds is a common approach for gathering object information. The abstraction from the raw pointclouds to high level representations is a necessity in the real world domain. This paper describes a method for the approximation of objects from scenes and an application to grasping with multi fingered hands. For this task a solution based on superquadrics is presented.