Title :
Coevolving species for shape nesting
Author_Institution :
Northern Michigan Univ., Marquette, MI, USA
Abstract :
We extend the work of Horn (2002) by applying the resource-based fitness sharing (RFS) approach to nesting of arbitrary, non-convex polygons. The shape-nesting problem now includes rotation, in which the shaped pieces can be placed anywhere on the substrate, at any angle of rotation. The task at hand is to maximize the number of such placed pieces such that no placed pieces overlap with each other or with the substrate boundary. We find that a single population, evolved under RFS, is able to discover a cooperative set of "species" that together "cover" most of the substrate, thus showing that the successful results reported by Horn (2002) carry over to the more general case of non-convexity.
Keywords :
computational geometry; evolutionary computation; resource allocation; arbitrary nonconvex polygon nesting; coevolving species; resource-based fitness sharing; shape nesting; Automotive engineering; Clothing industry; Coils; Fasteners; Genetics; Manufacturing industries; Metals industry; Resource management; Shape; Textile industry;
Conference_Titel :
Evolutionary Computation, 2005. The 2005 IEEE Congress on
Print_ISBN :
0-7803-9363-5
DOI :
10.1109/CEC.2005.1554906