Reactive Data Gathering

Designing autonomous systems to perform reconnaissance and surveillance missions is especially challenging. Platforms must coordinate, to make on-line decisions about what to sense, what to communicate and how to manoeuvre to achieve their mission goals. A first step in achieving consistent, coordinated decision making is to provide a means for rapid and efficient distribution of information between platforms. We have developed an efficient algorithm for this task in systems that use probabilistic method (Bayesian networks) for reasoning in uncertain domains. We demonstrate that this algorithm is more efficient in communication constrained environments (e.g. EMCON limited environments) than the alternative approach of communicating observations directly.