Fig. 2 Effect of the push operation on a horizontal capsule- like agent. Fig. 1 Left: sphere-like agent. Right: capsule-like agent. The topology finding process is implemented through a multi- agent system, where rooms are represented as bubble-like agents and connections as strings linking the agents. Bubble agents are divided into two types based on the type of room. Fig. 1 shows that the first type is a sphere-like agent that consists of a center point and a buffer distance. The second type is a capsule-like agent that consists of a segment and a buffer distance. Common rooms are represented as sphere- like agents and linear spaces, such as corridors, staircases, and rooms, which are represented as capsule-like agents across several floors. For horizontal capsule-like agents, the endpoints of their segments are affected by other agents. Thus, the endpoints can adjust their length and orientation automatically during the interaction between agents (Fig. 2). For vertical capsule-like agents, which usually represent staircases and tall rooms, the endpoints of its segment are Fig. 3 Swap operation between Rooms 2 and 4 avoids the incorrect topology caused by the intersection. The swap operation Swap(A1, Az) swaps the positions of agents. The position of a sphere-like agent is the center This is an optional rule that may be involved if minimizing the volume of the entire building is necessary. The defini- tion of this rule is that, for every agent A, the call push operation is Push(A, p, —mag-p), where p is the position of “ig. 4 Example of the multi-agent system. Left: initial status. Middle: status during the interaction. Right: stable status. Fig. 5 Screenshots of the entire process. Left: generated multi-agent layout. Middle: converted grid system. Right: optimized grid system. Evolutionary approach for spatial architecture layout design enhanced by an agent-based topology finding system Fig. 7 Left: generated topology relationships of the three-level houses. Blue points and lines show the internal geometries of bubbles, and red lines are connections between bubbles. Right: optimized grid-based layout. Evolutionary approach for spatial architecture layout design enhanced by an agent-based topology finding syster Fig. 8 Generated spatial layout based on the bubble diagram. The left image is rendered without an exterior wall, and the right images are separated into layers. Fig. 9 Left: generated topology relationships of the office building. Agents marked in the red rectangle are failure cases. Right optimized grid-based layout. Fig. 10 Generated layout of the office building rendered as separated layers. example, the volume of the room in the grid system can be determined by counting the number of cells. It can also generate layouts in different scales without numerous changes on time consumption by adjusting the resolution of the grid. Furthermore, grid-aligned layouts are easier to process, and thus more complex results may be produced within the same amount of time. The model used by the evolutionary optimization is based on a 3-D grid system. This discrete model is easy to implement and does not require many calculations. For Please cite this article as: Guo, Z., Li, B., Evolutionary approach for spatial architecture layout design enhanced by an agen topology finding system. Frontiers of Architectural Research (2017), http://dx.doi.org/10.1016/j.foar.2016.11.003 Fig. 12 Generated layout that contains 40 rooms. Evolutionary approach for spatial architecture layout design enhanced by an agent-based topology finding syster bui d because all their components have the same size. However, the grid system limits the solution space. For inst wit ach ance, non-orthogonal layouts are difficult to generate hin one grid system, and curve-shaped rooms cannot be ieved within this model. An alternative for this is adopting a multiple grid system (Hua, 2012) or a non- orthogonal grid system (Peng et al., 2014) or using the con con tinuous model instead of the discrete one. However, the tinuous model requires much more complex calculations ana sometimes strong Boolean operations, which are diffi- cult to implement. The approach provided by Merrell et al. (2010) and Rodrigues et al. (2013a, 2013b, 2013c) is an exa des mple of adopting the continuous model in a 2-D layout ion. The results of these annroaches are still orthogonal. program may involve a concept of hierarchy, which means that rooms are divided into groups according to topology. The optimization process starts from organizing the positions of groups instead of rooms. After the group positions have been decided, the program modifies the rooms in each group. This step may reduce the difficulty of generating complex archi- tectural layouts, such as hospitals and laboratories. More- over, an interactive design process, especially during the multi-agent system, may be involved. The positions of the agents may be adjusted by architects in real-time. Thus, a powerful tool that assists the architects in the conceptual design process may be produced.
Connect with 287M+ leading minds in your field
Discover breakthrough research and expand your academic network
