In today's society, patrolling has become a non-negligible part of office buildings, home environments, factories, and a city. Security surveillance systems are widely adopted and accompanied by the intervention of patrol personnel. Most surveillance cameras are set in a fixed location and have difficulty reaching all corners. In addition, hazardous environments prevent humans from being involved. In these situations, automated guided vehicles (AGVs) can be used not only for moving goods but also helping patrol personnel ensure the safety of humans and goods. In this paper, we develop a patrolling system for automated guided vehicles that integrate with wireless sensor network (WSN) technology to recognize objects and persons at a very low cost. We also show how to integrate current recognition mechanisms into one system. The system is composed of the front end of an automated guided vehicle and the monitoring back end and automatically sends a notice in case of something wrong. An authorized person accesses the system through the Web from a mobile phone, a laptop, and a computer. The system is validated by experiments and shows its feasibility and high potential.