Intrusion Detection from a Moving Platform

Intruder detection using a mobile platform with multiple sensor suites is difficult because of the ever-changing warehouse environments in which the robots operate. Doppler shifts caused by platform motion or intruder motion also add to the problem. ORINCON proposes a novel approach to intruder detection that uses a statistical differencing technique on multiple-sensor-suite digitized snapshots at preprograumied locations in the environment. Image differencing of digitized video camera frames has been used successfully in industry to detect intruders by detecting changes in the background from frame to frame. ORINCON takes this notion a step further. First, during initialization, a robot takes several digitized sensor snapshot measurements along a preprogrammed path and stores the data and the measurement locations. On subsequent path passes, the snapshot measurements-taken at the same locations-are compared to the initialization measurements by differencing. If "significant" discrepancies exist, then an intruder detection is determined to have occurred. If not, then the most recent snapshot is stored for the next pass. The main thrust of the proposed work will be to determine what "significant" discrepancies are and how they are related to probability of detection and probability of false alarm.