EAGER: Model-Driven Sensor Management in Support of Decision Making

点击翻译按钮获取中文摘要

Imagine a scenario in which illicit nuclear material may be in transit. How can we tell in a matter of minutes whether a truck, boat, railway car, or even a person with a backpack in a crowed street is a carrier of (possibly shielded) radioactive material? This project focuses on the underlying science behind the problem of timely detection of weak radioactive sources from a distance, by steering robots (ground and/or aerial) equipped with radiation counters. The proposed theory promotes national security objectives by providing improved decision-making algorithms. These algorithms can support both existing and future sensor technology, and help develop new detection tools which can form an effective layer of protection against radiological attacks at home, and contribute a useful tool for nuclear nonproliferation abroad.Novel features of the methodology in this project include the model-based approach to active sensing, and the formulation of the detection problem in a way that is more compatible with real-world scenarios. This approach departs from the classical active sensing approaches by querying measurement models, as opposed to merely measurement data, in order to improve the performance of decision making. The approach signifies a paradigm shift for active sensing in sensor management: instead of relying on sensor measurements (data) to decide how to sample, here the approach is to rely on measurement physics (models).