Planning strategies for decision-theoretic robotic surveillance

Abstract

In this paper, a decision theoretic view of the goal of surveillance is adopted. Subsequently a very simple formal model of the environment is introduced with the purpose of defining and comparing surveillance strategies. Several existing surveillance strategies are discussed. Then our decision-theoretic strategy is proposed and compared with the existing strategies on a theoretical level. In order to further experimentally compare surveillance strategies, a simulator has been implemented. Finally, some time complexity issues related to the implementation are mentioned.