Ambient intelligent agent to control crowd dynamics

Abstract

Human behaviour within a crowd is chaotic and unstructured normally controlled by the individual or group interests and needs within the same crowd. The fact that smart phones have become so popular throughout society, together with the potential of unleashing a suite of intelligent applications on the same device, leads to a plethora of added value services provided by the input from the mobile’s sensors into the system developed and deployed on the phone. This paper presents a system, where users are guided through a crowded environment, like a museum, in such a way that their path and experience does not interfere with that of other users where possible. The system uses a client-server model where the server is used to monitor the current crowd density and potential congestion locations. The clients are users who are equipped with an Android application which communicates with the server to generate a tour according to the user’s interests. The server uses Histograms of Oriented Gradients (HOG) to detect human figures and build a topology of the environment using this information. Users are able to request their next destination, prioritising factors and crowd dynamics which may affect their tour. The client and server components communicate using a web service. The system has been evaluated thoroughly and both components delivered satisfactory results.