ARG-US Remote Area Modular Monitoring: Digital Camera Enhancing Safety and Security

ARG-US Remote Area Modular Monitoring (RAMM) is as an expandable, adaptable system for monitoring critical nuclear fuel cycle facilities, such as nuclear power plants, fuel production and reprocessing facilities, spent fuel dry storage systems, decommissioned plants and facilities, and underground repositories. ARG-US RAMM is being developed under the auspices of the U.S. Department of Energy Packaging Certification Program, Office of Packaging and Transportation, Office of Environmental Management. The ARG-US RAMM architecture is designed on a modular platform to accommodate an expandable array of sensors that may include external thermocouples for temperature, humidity, and radiation (gamma and neutron) sensors, as well as an accelerometer and electronic loop seal. A digital camera, or optical sensor, is the latest sensor that has been successfully incorporated into the RAMM platform. The digital camera greatly enhances RAMM capabilities, providing an essential set of “eyes” covering selected areas of continual operations. The digital-camera-equipped RAMM units enhance safety by providing views in areas that may be inaccessible during a disruptive event. Switching automatically to battery backup and wireless sensor network (WSN), the RAMM units can transmit images, in addition to other sensor data, from areas where the landline-based surveillance assets are lost, such as during the crucial periods of the Fukushima accident. The digital camera can also enhance security during normal facility operation by applying algorithms to detect motion and/or events and, in subsequently, triggering an alert/alarm for any abnormal activities and enabling timely response. This paper describes the development and preliminary testing of distributed RAMM units, each equipped with a digital camera and other sensors in separate buildings at Argonne, focusing on image processing and video archives that pose challenges on information management and data storage. We believe our image processing can be further enhanced by incorporating artificial intelligence to allow for visual accounting of assets in the vision space, a capability similar to using computer vision for facial recognition. The pixel-based images accumulated over time—coupled with process knowledge in facilities—will be used as data input for machine learning, thus further improving algorithms for identifying potential abnormalities and reducing false alarms.