Probabilistic surveillance by mobile robot for unknown intruders

This paper focuses on a surveillance mission performed by an autonomous mobile robot in an environment. In this mission, the robot is required to detect as many intruders as possible by utilizing the mobility and monitoring capability. However, the robot does not know where and how many intruders are in the environment. Without the information, it is impossible for the robot to estimate an intrusion trend and detect the unknown intruders. For this challenge, we propose a novel surveillance method in order for the robot to estimate the intrusion trend. For this purpose, Bayes' rule is basically used. This probabilistic approach allows the robot to express the intruders through existence probability and appearance frequency. As the contribution of this paper, the robot is enabled to mainly monitor the places with the intrusion trend, in addition to other minor places. Through simulation experiments, the effectiveness of the proposed method for the unknown intruders is shown. Furthermore, an adaptability of the robot based on the probabilistic approach to various kinds of intrusion trends is discussed.