Recognition of Crowd Behavior from Mobile Sensors with Pattern Analysis and Graph Clustering Methods

摘要

Mobile on-body sensing has distinct advantages for the analysis andunderstanding of crowd dynamics: sensing is not geographically restricted to aspecific instrumented area, mobile phones offer on-body sensing and they arealready deployed on a large scale, and the rich sets of sensors they containallows one to characterize the behavior of users through pattern recognitiontechniques. In this paper we present a methodological framework for the machinerecognition of crowd behavior from on-body sensors, such as those in mobilephones. The recognition of crowd behaviors opens the way to the acquisition oflarge-scale datasets for the analysis and understanding of crowd dynamics. Ithas also practical safety applications by providing improved crowd situationalawareness in cases of emergency. The framework comprises: behavioral recognition with the user's mobiledevice, pairwise analyses of the activity relatedness of two users, and graphclustering in order to uncover globally, which users participate in a givencrowd behavior. We illustrate this framework for the identification of groupsof persons walking, using empirically collected data. We discuss the challenges and research avenues for theoretical and appliedmathematics arising from the mobile sensing of crowd behaviors.