應用於大規模人群移動之視覺化與分析及建模技術之研究---子計畫二：結合互動式視覺化技術之交通與人群動態分析

Abstract

大規模的人群移動分析對於都市規劃、交通預測、避難疏散以及流行病毒傳 播等的研究上均扮演重要腳色。隨著各類型可攜式全球定位系統(GPS)的普及以及 大規模交通監測網路的設置，我們可以更輕易的量測到各種人群移動資料。這些 移動資訊可能是精確的個人移動軌跡(例如手機GPS、車輛GPS)或僅是量測的平均 統計量(例如車輛偵測器測得的平均流量或車速、捷運進出口人數)。如何從不同 類型的大規模即時資訊或長期資料來分析人群移動，是極具重要性與挑戰性的研 究課題。 在本子計畫中，我們將以人群在道路或大眾運輸系統移動之視覺化動態分析 為主，並探討人群移動與車輛交通的互動關係。針對現今車輛交通及人群移動資 料的大資料特性，利用互動式可視化技術來協助專家有效呈現及分析交通及人群 移動資料。主要原因在於人類的視覺感知及分析能力，目前仍遠高於現代電腦所 能達到的範圍。互動視覺化分析緊密結合了人類的視覺感知能力以及電腦的運算 搜尋能力，透過兩者的互補，使用者可以有效率的分析大量資料。互動視覺化分 析不僅提供資料關聯性，也提供模型來闡述資料的模式。 我們相信本計畫將會對視覺化、計算機圖學、大資料分析視覺化、社會學及 運輸科學等跨領域研究帶來具體貢獻與實質影響。本計畫預期貢獻如下：(1)提出 以互動式視覺化分析技術探討大規模長時間的交通動態與人群移動及其互動關 係；(2)發展適合多類型交通動態與人群移動視覺化分析與互動建模技術；(3)所 發展的模型及分析結果對於大規模的車輛與人群的電腦模擬將有關鍵影響，可產 生更逼真的電腦動畫、更精確的交通預測、更完善的運輸規劃與避難疏散措施。

Large-scale and long-term human mobility analysis plays an important role in urban planning, traffic forecast, disaster prevention and evacuation, and epidemic propagation. With the popularity of portable GPS devices and the installation of large traffic sensor networks, we can easily measure all kinds of human movement data. These movement data can be individuals’ accurate moving trajectories, such as cellular phone GPS and vehicle GPS, or just aggregated mean statistic, such as the average flow or speed measured by vehicle detectors or the average number of passengers through an entrance/exit of an MRT station. How to analyze human mobility data from large-scale streaming sources or long-term historic database has become a very important yet challenging research topic. In this project, we will develop visual analytic techniques for analyzing human mobility that mainly travels via roads or public transportation systems. In addition, we will also develop visual analytic techniques for studying the interaction between crowd movement and vehicle traffic. Our motivation is that a human’s visual perception and analytic capability are far beyond those of computers. Visual analytics multiplies the analytic capability of a user and the computational power of a computer. As visualization and computation interplay and complement each other, the user can analyze big data effectively and efficiently. Visual analytic methods enable not only discovery of patterns in data but also building of formal models representing the patterns. We believe this project will make concrete contributions and great impact to the research fields of visualization, computer graphics, big data analysis and visualization, social science and transportation research. The expected contributions are: (1) proposing to apply interactive visual analytic techniques to study large-scale long-term vehicle traffic and human movements as well as their interaction; (2) developing visual analytics and interactive modeling approaches for analyzing multi-type traffic and human mobility; (3) the model obtained from our visual analysis approach will have big impact to large-scale vehicle and crowd computer simulation, which results in more realistic computer animation, more accurate traffic forecast, better transportation planning and evacuation procedures.