Efficient shape and pose recovery of unknown objects from three camera views

摘要

Knowing the shape and pose of objects of interest is critical information when planning robotic grasping and manipulation maneuvers. The ability to recover this information from objects for which there is no a priori knowledge is a valuable behavior for an autonomous or semi-autonomous robot. In this paper, we present methods and algorithms for the shape and pose recovery of unknown objects using no a priori information. Three camera views are taken of the object from three mutually orthogonal directions. Using shape from silhouettes, a point surface reconstruction of the object is performed and a superquadric is fit to the resulting array of points. We show through simulation and experiment that this method yields reconstruction accuracy sufficient for planning and executing grasping maneuvers. In addition to novel contributions, we have made performance improvements to recently published algorithms.