Generating Automated Layout Design using a Multi-population Genetic Algorithm

The problem of space layout planning, constrained by a number of functional and non-functional requirements, not only challenges architects in coming up with a good solution, but is more difficult to give an alternative. Genetic algo-rithms (GAs) have been found suitable for solving the problem of providing alternative solutions. However, GAs have been found to be susceptible to the problem of local maxima and plateau conditions. To overcome these prob-lems, the multi-population genetic algorithm (MPGA) improves the diversity of the population, thereby improving the quality of the solution. Algorithms are employed to automatically generate layout designs in best-connected ways, either rectangular or square. The area of the floor plans is optimized to minimize the extra area in the layout. The layouts are divided into four groups and these groups are related to each other based on highest proximity. Layout designs have been simulated using GA and MPGA algorithms and MPGA has shown significant improvement in computation time as well as quality over alternative solutions. In addition, the algorithm also provides the architect with the facility to interactively modify the dimensions and adjacent criteria during the design phase. The system works on clouds and shows the result for inputs passed by an architect.