This demo paper serves to describe a research platform for autonomous surveillance by mobile robots. System components from various technical disciplines (including robot navigation, computer vision, discrete event systems, and human-computer interaction) have been combined together into a cohesive, modular system architecture. We employ our architecture to experiment with state-of-the-art methods in planning and decision making under uncertainty, which we demonstrate enable robots to effectively patrol and respond to surveillance events in a large indoor environment.@en