Crowd-Z: The user-friendly framework for crowd simulation on an architectural floor plan Machi Zawidzki a,⇑ , Mohcine Chraibi b , Katsuhiro Nishinari c a Department of Architecture, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA b Jülich Supercomputing Centre, Wilhelm-Johnen-Straße, 52428 Jülich, Germany c Research Center for Advanced Science and Technology, The University of Tokyo, Komaba 4-6-1, Meguro-ku, Tokyo, 153-8904, Japan article info Article history: Available online 1 November 2013 Communicated by Rita Cucchiara Keywords: Pedestrian dynamics Agent based modeling Design support Digitized floor plan abstract This paper introduces Crowd-Z (CZ): a framework that provides a user-friendly platform where architects can perform simple crowd simulations on floor plans. A simple but robust and flexible agent-based system is used for modeling of the crowd dynamics. Such simulations can be performed at any stage of design – from rough sketches to the final blueprints. CZ allows acquiring the layouts for the simulations in a number of ways: freehand sketches, importing already prepared images and appropriat- ing preprocessed images from commercially available Computer Aided Design programs. These three methods are illustrated with practical examples, followed by a number of simulations compared with the literature or other commercially available programs. Ó 2013 Elsevier B.V. All rights reserved. 1. Introduction In the past, several models for pedestrian dynamics were devel- oped and used for various purposes, e.g. planning of buildings, organization of mass events, etc. In general, these models can be classified in two major categories: space continuous models, which are based on the Newtonian approach (Hirai and Tirui, 1977; Hel- bing and Molnár, 1995; Yu et al., 2005; Chraibi et al., 2011) and rule-based models discrete in space that describe the dynamics of crowd on a regular grid by means of transition probabilities (Gipps, 1986; Blue and Adler, 1999; Rajewsky et al., 1998; Kretz and Schreckenberg, 2006). Particularly relevant to the concept pre- sented in this work are agent-based systems (Kerridge et al., 2001; Bandini and Vizzari, 2007; Shi et al., 2009). Most of these models describe rather well pedestrian dynamics quantitatively and qualitatively. Based on these models, several commercial tools are available, for example: VISSIM (Kretz et al., 2008), ASERI, (Schneider and Könnecke, 2002), PedGo (Klüpfel et al., 2002). However, in the common architectural and urban planning practice they are rather rarely used. One of the reasons is that designers often do not feel necessity for such experimenta- tion, but also they may get discouraged by the price, complexity or ‘‘unfriendliness’’ of the available software. The motivation for this work is to introduce an intuitive environment for designers, in which they can comfortably perform crowd simulations (CSs). The holistic approach presented here also contributes to the academia so the researchers can have an overview of what design- ers expect from such tools and models. For the same reason – it is also addressed to the software engineering community. Crowd-Z (CZ): the user-friendly framework brings the idea of CS to broader audience than before, and hopefully will encourage the architects and planners to use it for designing safer and more convenient spaces. The main objectives of CZ are: 1. To simulate the crowd dynamics (CD) on a given floor plan, in particular – to find the problematic areas where the traffic does not flow properly or conversely, the areas that are visited too sparsely. 2. To allow for experimentation by alterations to the geometry of the environment, in particular by introducing or erasing ele- ments or their parts, changing the locations of the start termi- nals, exits, etc., and to immediately observe the influence of these changes on the overall CD. 3. To study the emergent behavior of agents determined by differ- ent behavioral scenarios. The next section briefly describes the model used for CSs, fol- lowed by three practical applications of CZ in architectural design and a number of simulations validating the CZ model. 2. Crowd dynamics component This work focuses on the geometrical processing rather than the mathematical modeling of pedestrian dynamics. Although modern models produce more realistic results in terms of CD, for testing 0167-8655/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.patrec.2013.10.025 ⇑ Corresponding author. Tel./fax: +81 3 5452 5286. E-mail address: zawidzki@gmail.com (M. Zawidzki). Pattern Recognition Letters 44 (2014) 88–97 Contents lists available at ScienceDirect Pattern Recognition Letters journal homepage: www.elsevier.com/locate/patrec