Article Preview
TopIntroduction
The mobility of citizens in an urban area is the source of various problems: traffic congestion, environmental pollution, overcrowding of public transportation systems and spreading of diseases, among others. For these reasons many efforts have been put to understand the mobility behavior of individuals in space, understand space itself, and understand the use people make of the urban space as a way to reduce and possibly eliminate these difficulties.
The dynamics associated with mobility in urban areas have always two components, Time and Space, rising new challenges on how to capture, represent and visualize these dynamics. While capturing the presence and mobility of people in urban spaces has evolved enormously in recent years, movement representation and visualization still faces many challenges. Actually, the huge size of datasets being collected these days is creating more challenges to representation, modeling and visualization rather than solutions (in spite of their great potential for mobility analysis).
As referred by Yu and Shaw (Yu & Shaw, 2004) existing Geographic Information Systems (GIS) are structured to represent the spatial component of data but lack good support for the temporal component. For this reason, some authors have developed platforms that provide support for spatio-temporal data, such as SECONDO (De Almeida, Güting, & Behr, 2006), with the aim of representing the time dependence of mobile artifacts through the use of abstract definitions to represent the positions or shapes of objects over time and space, respectively (Erwig, Güting, Schneider, & Vazirgiannis, 1998). On the other hand there are also challenges in how this information could or should be presented for human movement analysis. Thus, several studies presented recently explore different forms of visual representation of movement data, such as the iSPOTS project (Sevtsuk & Ratti, 2005) which illustrates the occupation of space or, for example, the visualization of human travel behavior based on the trajectory of money bills (Brockmann & Theis, 2008).
However, regardless of how the dynamics of an urban space is represented, most of these works focus only on one, homogeneous, type of mobility data. Although our research work is focused on the analysis of the dynamics of urban space, our initial aim is to create a flexible and comprehensive conceptual framework for the representation of movement processes that allows the same concepts to be applied to different types of data from different sensors, such as GPS, Wi-Fi, GSM, ticketing systems, as well to the different modalities of urban mobility.
Our approach to capture the dynamics of the urban space is based on merging the individual mobility profiles of people. This approach aims to benefit from the current capability of smartphones and other personal devices to be used as proxies to observe human spatio-temporal behavior. In our approach, the first step in the analysis of the urban dynamics is, then, the automatic creation of personal mobility profiles from multi-sensor data. On a previous paper we set the foundations for the above mentioned framework, and proposed a method for place learning based on a probabilistic model applied to observations obtained from GPS, Wi-Fi and GSM sensors (Peixoto & Moreira, 2012). That previous work has been useful to assess the effectiveness of the proposed concepts and to identify different perspectives on the analysis of human mobility. In this article we intend to extend that work and discuss some of the open issues that were previously identified, namely the concepts of Place and Trajectory.
The next sections describe: some of the work developed in the field of analysis and visualization of urban mobility; the concepts that are the basis of our proposed framework for the representation of spatial-temporal data; one dataset composed of data from three types of sensors and how it is mapped into the proposed concepts. Finally, in the last section, some conclusions and open questions are discussed.