Developing a Deterministic Patrolling Strategy for Security Agents

Developing autonomous systems that patrol environments for detecting intruders is a topic of increasing relevance in security applications. An important aspect of these systems is the patrolling strategy; namely, the determination of where to move in order to conveniently detect intrusions. While a large part of patrolling strategies proposed so far adopt some kind of random movements, deterministic strategies can be useful in some situations of interest. In this paper, we propose an approach to find a deterministic strategy that allows the patrolling agent to always detect an intruder that attempts to enter an environment. The problem is formulated as the determination of a cyclic path that visits, under temporal constraints, all the vertexes of a graph representing the environment. We propose a solving algorithm, study its properties, and experimentally evaluate it.