SBIR-STTR Award

Automatic aids for situation assessment, general information fusion, decision making and planning are vulnerable to deception in adversarial situations. Counter-deception capabilities to detect possible deceptive actions and obtain the correct assessment are needed to preserve the effectiveness of these aids. The proposed research will create technological support so that the effects of deception on problem-solving systems are recognized and mitigated as much as possible. The specific objectives for Phase I are to elucidate the problem of providing automated support for counter deception and to design a general problem-solving support system that will perform a problem-solving task, and also assist the user in dealing with deception relative to the problem-solving task. The four categories of counter deception support are: explicit representation of deception concept, multiple reasoning methods, system wide representation, and system control. The Phase I effort will result in a design of a deception-resistant problem solving support system from which a prototype will be built in Phase II .

The data, knowledge, inference, and reflective capabilities of expert systems are limited. We describe a collection of tools that address the problem that deception poses for expert systems and for organizations that employ them to reason about observations, plan, or make decisions. We have completed a Phase I SBIR in which we explore the ways that deception, of deceiving, and of the components of expert system. For Phase II we proposed to build a deception planner and use the deceptive plans to detect the weaknesses of particular expert systems and determine how they can be mitigated or eliminated. Specifically, our objective is to develop a collection of tools to suppon counter-deception reasoning in support of the situation assessment task for the Navy's Outer Air Battle mission. Our deception planning methodology calls for the deceiver to behave in a way that manipulates the beliefs that the target forms based on his (the deceiver's) emissions and actions. Our deception planner integrates case-based reasoning, sophisticated planning technology, and belief formation and revision technologies.