Intelligent Security Assessment for a Mobile Robot

Abstract

In many robotics applications the data from different types of sensors must be fused to provide useful information about the surrounding environment. The system described here uses sensor fusion to determine the probability that an intruder is present in the vicinity of a remote robot. Several of these robots are employed in an indoor security scenario, where each robot monitors a different region of a large building and reports back intruder information to a single operator. This paper examines the realtime security assessment algorithm used by each robot to fuse information from 82 sensors of five different types, and return a single composite threat value to the operator. The algorithm described has been successfully tested on a single robot with 99% probability of detection achieved. No nuisance alarms were recorded.