Integrating GIS, 3D-Isovist, and an NSGA-II multi-objective optimization algorithm for automation of design process in urban parks and public open spaces

Advanced digital tools in landscape architecture are mostly limited to visualization and presentation of alternatives. However, they can potentially be used in different design stages. In this paper, we propose a method to approach a design problem as a multi-objective problem (MOP) and integrate advanced digital techniques into an automated landscape design framework to exploit their superior computational capabilities. We combined geographic information system (GIS) tools for mapping of the site, 3D Isovists for analyses, and a meta-heuristic method (constrained Non-dominated Sorting Genetic Algorithm-2 or NSGA-II), to search in the continuous solution space (fitness landscape). The case study was an urban park in Tehran, Iran, and the focus was on spatial-visual characteristics of the green space. The results showed that the NSGA-2 was able to solve the complex design problem with 185 trees and 66 observers. The algorithm produced a Pareto-frontier consisting of four optimal solutions that, compared to the existing state of the park, showed more than 18% and 12 % improvement according to Tree View and Building View, respectively. These results confirmed the applicability of our proposed semi-automated design framework. This study is of interest to both professional practitioners and academics of landscape architecture since it can help bridge the gap between scientific assessment and its application in real-world design studies. The proposed method can be further developed to take other design considerations into account and also has the potential to be of use in other related design fields.