Automated Generation of Optimally Directed Three Dimensional Component Layouts
Abstract This research introduces a novel approach towards automating the generation of three dimensional component layouts. Three dimensional component layout tasks are typically highly combinatorial, and exhibit objective function spaces that can be nonlinear and/or discontinuous due to discrete allowable locations for component placement. We present an approach to the three dimensional component layout problem that employs shape annealing, a design generation technique combining concepts from shape grammars and simulated annealing, to produce optimally directed designs. The shape annealing paradigm sacrifices global optimality in exchange for the ability to find good designs in a reasonable amount of time, given very large, ill-behaved objective function spaces.