百貨公司廣場行人移動自我組織現象之研究

Abstract

過去行人相關研究多以聚集人群為單位，侷限於描述規律性的移動行為，但無法預測人群因個體間的互動所帶來的多變化結果；行為者基礎模型由下往上的模擬方式，已經成功地模擬多種突現現象，惟行人自我組織的形成條件與原因仍有待深入探討。有鑑於此，本研究藉由錄影調查台北車站新光三越百貨之行人廣場，探討該廣場的行人移動特性與現象，並從觀察中找到互動規則，以建構行為者基礎模型模擬現況，並探討其自我組織現象。經研究發現，該廣場之行人具有資訊不完全、有知性、密度適中的特性，行人之間以『吸引力』與『進入阻力』作為傳遞資訊，在不同時段觀察發現行人分佈均收斂成某一固定形態，另外，駐足人群的群落規模與數量的關係呈現冪次趨勢，為自我組織臨界的特徵現象。本研究進一步以NetLogo作為模擬軟體，構建模擬模型以分析行人之移動行為，其中環境狀態變數為『吸引力』與『進入阻力』，行人會選擇吸引力最大與進入阻力最小的方向。模擬模式經以廣場現況在低、中、高三種行人密度的駐足人群分佈驗證，顯示模擬模型具有良好的詮釋能力。本研究利用模擬模型分析行人密度與延遲時間的關係發現，穿越行人延遲時間會隨著密度的增加而增加，並行人移動系統會從秩序變成自我組織再轉變到混亂。本研究另分析廣場內設置咖啡座的影響，該應用以增加平均駐足吸引力與避免激增穿越行人延遲時間作考量原則，分析得咖啡座之適當設置位置及其對廣場行人移動之影響情形。由本研究之應用分析案例，顯示出該模擬模型可作為輔助廣場空間配置，分析廣場家具與行人移動互動影響之工具。

Literature review shows that pedestrian movement models were usually built in an aggregative scale, and were restricted within regular movement behavior. But they can’t predict varied results cause of interaction of individuals. Agent-based models, built from the bottom-up, have successfully simulated several emergent phenomena, but the formed conditions and reasons need be discussed more. The study is to investigate a living example – a department store plaza, and analyses its self-organization phenomenon, complex characteristics, interaction rules, and builds a pedestrian agent-based model. The paper discovers that pedestrians are restricted rational, reactive, medium density. “Attraction” and “walkability” are their communication information. Their distribution always tends to some form, and the relationship between the scale and frequency of stationeries shows the power law, one of the evidences of self-organization. These complex characteristics create a nice environment for the interaction and evolution between pedestrians. This model uses three modules to direct pedestrian movement. Pedestrian will take the way of maximum attraction and minimum walkability. This model can simulate the actual situation. It can be seen from the case study 1 that the delay time of pedestrian will increase with higher density, and system changes from order to self-organization, and to chaos. Case study 2 simulates the influence of adding a coffee shop in the plaza, and gets the proper location of coffee shop. The model can be a good tool for predicting the influence of space and furniture on pedestrian movement.