Territorial security deals with the prevention, detection and response to unauthorized persons and/or goods from crossing a perimeter. It deals with large territories of strategic importance, such as international borders, transportation and critical infrastructure. Multi-agent systems provide flexibility, fault-tolerance, high sensing fidelity, low-cost and rapid deployment. In this paper, we concentrate on the challenges presented in applying the concepts of multi-agent systems to those presented by territorial security. We first introduce the overall system as well as prevalent agent architectures. We then briefly present our novel agent architecture, its experimental embodiment and the virtualized reality model that accepts physical sensor data and updates a global model of the environment in real-time.