Concept for a Genetic Algorithm for Packing Three Dimensional Objects of Com- plex Shape

In this paper we describe a concept and a research project under way to develop a packing method for three dimensional objects of complex shape. Selective Laser Sintering (SLS) is a rapid prototyping (RP) technology which allows manufacturing of three dimensional objects directly from a solid CAD model by sintering fine plastic or metal powder in thin layers. When more than one part is to be manufactured at the same time, an operator has to build a batch out of CAD files making sure that parts do not intersect each other, they are totally inside the build cylinder and all other constraints are satisfied. This packing problem has several features which make it different from classical bin packing problems or other similar problems studied so far: (i) there are no restrictions on part shape or orientation, (ii) smaller parts can be built inside cavities of larger parts and (iii) no part is required to be underneath a part to provide support (parts “float” in powder). Developed packing algorithm is based on a genetic algorithm (GA) search. An order–based two dimensional chromosome represents both, the order and orientation of parts to be packed. Part geometries are included in the decoder which simulates the packing in order to evaluate each individual’s fitness. Packing schemes violating hard and soft constrains are penalized to reduce their fitness evaluation.