Evolutionary search techniques application in automated layout-planning optimization problem

The purpose of this paper fits in optimization of apparel shapes layout. The aim is to mark all the shapes to be cut onto the sheet by minimizing unoccupied spaces. As the problem involves many sub-optimal solutions or many local optima, genetic-like algorithm is used to handle the layout process. The application of genetic algorithm (GA) needs parameter encoding. At first, an individual is encoded by combs coordinates of shape description. Strings are of variable length and genetic operators are created to this domain specific encoding. As genetic algorithm used alone is not very efficient, we made attempt to hybridize GA with simulated annealing (SA). Results provided are compared with GA alone and with our previous results by tree search.