Situation & Threat Assessment

Background

To perform any difficult control task such as driving, flying an aircraft, operating a chemical plant, controlling a power grid, or exercising military command it is necessary to have a grasp of the overall situation.  In fields including man-machine interaction, human interface design, industrial psychology, supervisory control and C³I, attempts have been made to clarify the meaning of the term situation awareness¹.  Much work has been done on human situation awareness, but development of automatic algorithms that are able to build an overall appreciation of a situation has lagged.  We tend to term automatic situation awareness situation assessment.  Researchers in the Business Intelligence Group have since 1999 explored a probabilistic approach to providing situation algorithms.  The initial work was published.  Since then considerable expertise has been built up within the Group, and there have been applications in the military domain.  The speed and scalability of the algorithms has proved advantageous.

The great advantage that a situation picture provides in complex control problems is that it provides sufficient context to make decisions to allocate resources.  If these decisions are taken without considering the wider context, results could be disastrous.  In the military domain, it is crucial that assets are only risked if their loss would not open up a critical weakness.  Situation awareness can prevent adventitious decision-making.  We have considered resource allocation problems involving both surveillance decisions and substantive actions.  An example of a surveillance decision would be deciding to send a drone aircraft to make photographs of a site, and a corresponding surveillance decision would be whether or not to launch an attack.

       1.  sometimes situational awareness

A dynamic program typical of those found in hard template situation assessment.

Goals

The aim of this line of research is to provide and commercialise the algorithms and application skills in the military and civilian domain.

• construction of a layered overall model of interacting time series

• derivation of fast scalable algorithms

• algorithms for situation assessment

• provision of mechanisms for capturing situation templates from human operators

• situation awareness using hard and soft templates

• self-generating situation templates

• multi-stage multi-object optimal resource allocation

• application in the military and civil domains

Many of these goals have already been achieved.

Research Issues

From the algorithmic perspective, both situation assessment and the ensuing resource allocation problems are searches.  An initial examination of their complexity is somewhat discouraging.  So besides system modelling, the main issue in this research is how to construct very fast searches of the most likely areas of the situation and decision spaces.

Publications

A statistical approach to situation assessment, Fusion 99, Sunnyvale CA. (early open paper) 

Links

• The Orincon Corporation

• The Silk Road Group - a  conventional view of of data fusion and situation assessment 

To top

last updated December 18, 2003 11:47 AM
Geoff.Jarrad@csiro.au