Networked Mobility-Enabled Detection of Mobile Weak Radiological Signatures
Abstract
This project develops a novel approach to reconfiguring networks of passive detectors mounted on unmanned aerial vehicles in order to timely detect weak nuclear sources moving in an urban environment. The proposed work informs on how to actively reconfigure and manage in real-time the mobile network of radiation sensors for the purpose of detecting a moving weak source as quickly and as accurately as possible. It does not critically depend on particular sensor modalities or technologies; in fact, it can directly incorporate developments made in the sensor technology front, only to improve the overall performance. With off-the-shelf hardware but new state-of-the-art algorithms, the proposed approach promises improvement in detection performance in outdoor environments that can be an order of magnitude better compared to stationary sensor alternative solutions. The proposed approach targets a number of key questions related to short-range search and detection of radiological material using sensors mounted on off-the-shelf robotic vehicles. These robots will be exchanging information and navigating in outdoor environments with obstacles, without relying exclusively on GPS to track their motion. To realize such a system using existing hardware, we need new algorithms for robotic vision and motion estimation, motion planning and navigation, and networked decision-making regarding the presence or absence of nuclear material. The estimation algorithms will enable the sensors to map the space around them, find their own position in it, and track the location of potential threat; the motion planning algorithms will steer the sensors along paths on which they collect information that supports an optimal decision as to whether their target is benign or not; and the networked decision-making algorithms inform as to how the sensors can exchange information between them in order to increase the confidence on their decision. The key insight is that a common underlying mathematical foundation can link the estimation, control, and decision-making problems, allowing them to be formulated so that each is framed within the constraints of the others. So if by co-design one succeeds in meshing these different algorithmic components in a way that they serve each other, the potential order-of-magnitude improvement in decision-making performance (for example, detection accuracy) can be realized. The new methodologies will significantly advance our capabilities to remotely detect and intercept nuclear material in transit that may be shielded and concealed. And while no defense system can ever be 100% effective, such improved detection methodologies not only add another layer of security against the deployment of radiological weapons of mass destruction, but will also force adversaries to employ more elaborate strategies, which are even more difficult to escape detection. Decision-making ultimately plays a pivotal role in translating information to action. Human decision makers can be significantly assisted in determining a time-critical plan of action if provided with concise, dependable information about the state of the world, rather than being buried under a ton of data. While the technical approach in this proposal is particularized in the context of nuclear detection, the impact of the methods proposed here can each a wide range of application domains, from natural disaster early warning systems, to optical communications, radar, and acoustic detection, to medicine, and physics.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 26, 2016
- Source ID
- HDTRA11610039
Entities
People
- Herbertglenn Tanner
Organizations
- Defense Threat Reduction Agency
- University of Delaware