Year
1988
Abstract
ROBART II is a battery powered autonomous robot being used by the Naval Ocean Systems Center in San Diego as a testbed in research which seeks to provide a multi-sensor detection, verification, and intelligent assessment capability for a mobile security robot. The intent is to produce a robust automated system that exhibits a high probability of detection with the ability to distinguish between actual and nuisance alarms. An architecture of nine distributed microprocessors onboard the robot makes possible advanced control strategies and realtime data acquisition. Higher level tasks (map generation, path planning, position estimation, obstacle avoidance and statistical security assessment) are addressed by a Planner (currently a remote 80386-based desktop computer). Numerous sensors are incorporated into the system to yield appropriate information for use in position estimation, collision avoidance, navigational planning, and assessing terrain traversability [1]. The robot is also equipped with a multitude of sensors for environmental awareness in support of its role as an intelligent sentry. These monitor both system and room temperature, relative humidity, barometric pressure, ambient light and noise levels, toxic gas, smoke, and fire. Intrusion detection is addressed through the use of five passive true-infrared body heat detectors, four passive optical motion detectors, ultrasonic motion detectors, microwave motion detectors, video motion detection, vibration monitoring, and discriminatory hearing. The realtime security software computes a composite threat assessment by summing the weighted scores of alarmed sensors within a given zone. If the zone composite threat exceeds a dynamically computed threshold, a true alarm condition exists. An adaptive network system (ANS), or \"neural net\" architecture is also under development for temporal security assessment. It will employ a Gaussian classification based on adaptive kernel estimation, a temporal subsystem, and an attention subsystem to allow the system to learn the sequences corresponding to threats.