Cognitive Models for learning to control dynamic systems

The development of fast and robust learning models which can work in non-stationary environments or scenarios with rapidly changing goals is becoming a critical task in both military and civilian applications. The objective of this proposal is to develop new mathematical/computational models based on cognitive science principles that are capable of rapid learning for command and control problems. In Phase I, the focus is on the integration of current cognitive models, such as DFT, with practical methods in nonlinear systems and control theory. Stochastic resonance techniques will be applied for the first time to dynamic signal detection and dynamic decision-making tasks. The efficiency of the proposed cognitive learning algorithms will be tested and evaluated based on previously established experimental research with human decision tasks. Phase II will focus on the development of software for application of these novel learning models to some Air Force missions of critical importance. This phase involves both computer simulations and experimental validations with human decision tasks.