Brief Announcement: Configuration of Actuated Camera Networks for Multi-target Coverage

Abstract

In various domains, including public safety, first-responder, and security applications, an important task is monitoring a public space for events of interest, using sensors of various types, including e.g. networks of cameras installed on the ground or unmanned aerial vehicles (UAVs), which may be either autonomous or not. In the settings we consider here, cameras are characterized by a family of parameters, some fixed, some settable, including location, viewing range, and so on. The task is to deploy the sensors, i.e., set the applicable parameters, in order to optimize an objective function capturing the quality with which we observe a set of targets. In a dynamic scenario, we may wish to observe a large set of targets (or a continuous region) with observation quality passing some low threshold, sufficient for surveillance; upon request (i.e., when events are detected), we may then be obliged to observe a small number of distinguished targets to a higher level of quality, sufficient for identification and localization. An example objective function might maximize the total observation quality of all targets, conditioned on the hard constraint that each target is observed to the appropriate minimum quality threshold.