Keywords

1 Introduction

To design a building is to create a number of environmental settings for its users’ activities. For this reason, architectural designers should always consider how situations change when crafting environmental settings that support users’ behaviors. The human factor is one of the most important standards for judging the quality of design alternatives [1].

Recently, many architectural features with non-Euclidian shapes have appeared in the urban environment, in response to social demands for innovative building shapes. In the educational field of architectural design, non-Euclidian design has become an important part of the curriculum, because generating these shapes is not an easy task for students who have been trained in conventional design education. In this challenging pedagogical situation, one significant problem involves the difficulty that students may simply be attracted to the notion of creating unusual forms while ignoring important factors like human behaviors during the non-Euclidian architectural design process, because many instructors focus primarily or even entirely on the technical methods needed to create non-Euclidian shapes [2].

To resolve this issue, we suggest a technical solution. This research develops a human behavior simulation system, called PlayGA, which can be used in the non-Euclidian architectural design process. PlayGA is a type of export system that can simulate human behaviors while taking into account design geometry. After geometrical shapes are designed using a geometrical design tool, PlayGA can deploy non-player characters (NPCs) in the design alternatives. These NPCs play freely and behave in a variety of different ways in the simulation. The autonomous behavior of NPCs is based on the intelligence of recognizing geometric affordance. In this paper, we introduce how to develop a technology that simulates geometric affordance.

2 3D Object Recognition Techniques

3D object recognition is a basic requirement for comprehending the NPCs’ reactions to geometrical shapes. There are two commonly known methods for 3D object recognition: the pattern recognition approach and the feature-based approach. Pattern recognition approaches use pre-captured (or pre-computed) images of an object to recognize that object. However, they cannot consider an object’s 3D context and do not handle occlusion [3, 4]. On the other hand, feature-based approaches work well in recognizing distinctive featured objects that contain vivid edge features, blob features, etc. However, it is not appropriate for recognizing the smooth surfaces objects without texture [5].

Several architectural objects like doors, windows, and furniture contain typical patterns, and previous approaches might well be able to recognize them successfully. However, those methods are not appropriate for this research, the goal of which is to extract geometrical affordances from non-Euclidian shapes, which are not easily specified as patterns. In order to extract NPC behaviors according to objects of various shapes, we developed a method using those objects’ basic attributes: vertices, edges, and surfaces. All architectural shapes have those attributes, and humans generally display certain behaviors in the context of a given shape. Thus, we have introduced a strategy to match possible behavioral information related to the properties of basic objects and reproduce it according to the characters’ traits. Figure 1 shows NPC behavior allocation strategy in PlayGa.

Fig. 1.
figure 1

NPC behavior allocation strategy in PlayGA

Full size image

3 PlayGA: A Human Behavior Simulation System Based on Geometry Affordances

In order to perform human behavioral simulation in the design process of non-Euclidian shaped buildings, a human behavioral simulation module (PlayGA) has been developed, based on the commercial modeling tools SketchUp Make 2017. Figure 2 shows the process of using PlayGA as a plug-in with SketchUp. PlayGA is developed on Unity3D, a game-based platform. After architects design non-Euclidian shapes, PlayGA can be executed in a real-time manner. It generates a virtual place based on the designed geometries and automatically deploys NPCs in that space. NPCs behave freely according to the geometric shapes, which are associated with pre-assigned behaviors. NPC behavior is also constrained by the characteristics of the human NPC figure and the object’s configuration status. Figure 3 shows NPCs’ reactions according to the direction and height of the contextual geometry.

Fig. 2.
figure 2

The process of using PlayGA

Full size image
Fig. 3.
figure 3

NPCs’ reactions according to the direction (left) and height (right) of geometry

Full size image

The behaviors of automatically generated NPCs provide architects with valuable information regarding the usability of the designed shapes. With PlayGA, architects can evaluate functionality in the midst of the non-Euclidian architectural design process and produce the optimal alternative (Fig. 4).

Fig. 4.
figure 4

Execution images in PlayGA

Full size image

4 Conclusion and Discussion

This research presents PlayGA, a human behavior simulation system that can be used in the non-Euclidian architectural design process. In PlayGA, the autonomous behavior of NPCs is based on the intelligence of recognizing geometric affordances. In this paper, we have introduced how to develop a technology that simulates geometric affordances. PlayGA appears to be helpful in evaluating the functionality of design alternatives in a non-Euclidian architectural design process. However, in further research, the effect of PlayGA should be evaluated through rigorous field tests, and further development is required to stabilize its functions.