Hybrid Evolutionary Algorithm applied to Automated Floor Plan Generation

Abstract

The article presents the application of Hybrid Evolutionary and Greedy-based algorithms to the problem of Automated Floor Plan Generation. The described optimization issue is part of a wider domain of Computer-Aided Architectural Design. The article covers the extensive description of the representation domain model (architectural canonical guidelines, user design requirements and constraints) and the explanation of proposed approach: problem representation, genetic algorithm operators, and fitness function definition. The research experimental procedures are based on real-world data: the architectural design guidelines being the design constraints and five real-world functional programs introduced and proposed as benchmarks. The article summarizes the implementation of the proposed approach, compares the Hybrid Evolutionary Algorithm experimental results with the Greedy-based algorithm, and suggests possible extensions and future research directions.

A tool for computer-generated dimensioned floorplans based on given adjacencies

“…Considering the automation where user can insert his choices, in 2019, Nisztuk et al [37] built a tool for automated floorplan generation, covering the majority of adjacency and size constraints, but it is limited to rectangular rooms only and generates empty spaces in the layouts. Recently, Shi et al [38] used reinforcement learning based on a heuristic search technique called Monte Carlo Tree Search to generate a closest feasible dimensionless RFP corresponding to any adjacency graph inserted by the user.…”

“…However, it took eight hours to generate 30 candidate solutions only. Nisztuk et al [37] also produced multiple solutions using a greedy approach and thus have very high computational time which clearly shows the efficiency of graph algorithms over greedy search techniques.…”

PROJETO, PADRÕES E TECNOLOGIA: da linguagem de Alexander à programação e inteligência artificial

“…Seja qual for o procedimento, tanto no enriquecimento semântico (ES) como na verificação de código, a definição do método de aprendizado de máquina ou baseado em regras dependerá do investimento no desenvolvimento, que por sua vez seguirá variáveis como o esforço necessário para formular conjuntos de regras, a complexidade da lógica, o grau de interpretação implícita necessário e a disponibilidade de dados para o aprendizado, como ilustra a Figura 3 (SACKS et al, 2019). Evidencia-se que a aplicação da tecnologia ao ambiente construído contribui de diferentes formas no processo desde a concepção do projeto arquitetônico, como por exemplo, na síntese (AS, et al, 2018;NISZTUK;MYSZKOWSKI, 2019), passando pela verificação (HU, 2018;SACKS et al, 2019), gamificação (SAVOV et al, 2016), ou ainda na própria fabricação (TAMKE et al, 2018).…”

Graph-based approach for enumerating floorplans based on users specifications

“…Recently, Egor et al (2020) gave an evolutionary algorithm for inserting circulations. In this paper, we present an algorithm for generating floorplans along with spanning circulations. Time complexity Many existing works can produce residential building layouts for a small number of rooms (Merrell et al , 2010; Wu et al , 2018; Nisztuk and Myszkowski, 2019), but for the complex building structures with a large number of rooms, we need efficient algorithms. The proposed prototype takes space for generating floorplans for any given number of rooms .…”

“…Many existing works can produce residential building layouts for a small number of rooms (Merrell et al , 2010; Wu et al , 2018; Nisztuk and Myszkowski, 2019), but for the complex building structures with a large number of rooms, we need efficient algorithms. The proposed prototype takes space for generating floorplans for any given number of rooms .…”