This paper considers solving a problem in combinatorial search: the automated arrangement of irregular-shaped objects for industrial 3D printing. The input is a set of triangulated models; the output is a set of location and orientation vectors for the objects. The proposed algorithm consists of three stages: (1) translation of the models into an octree; (2) design of an efficient test for pairwise intersection based on sphere trees; and (3) computation of an optimized placement of the objects using simulated annealing. We compare several sphere-tree construction methods and annealing parameter settings to derive valid packings.
展开▼