Robotic sensor networks are used to monitor physical environments that require measurement of physical variables in real time to prevent disaster by detecting and locating anomalies. In this paper we propose a model based on intelligent agents to control a robotic sensor network that detects and traks the perimeter of anomalies in a physical environment. In oder to define a model, a methodology was applied to multi-agent systems Mase and the implementation was done on the JADE platform-OSGi. Finally, the model was validated by testing in a specialized simulator for robotic sensor networks.