Planning in the Tactical Air Domain
Randolph M. Jones, Robert E. Wray, Michael van Lent · 2001
TacAir-Soar is a reactive system that uses recognition-driven problem solving to plan and generate behavior in the domain of tactical air combat simulation. Our current research efforts focus on integrating more deliberative planning and learning mechanisms into the system. This paper discusses characteristics of the domain that influence potential planning solutions, together with our approach for integrating reactive and deliberative planning. TacAir-Soar (Jones et al. 1993; Rosenbloom et al. 1994) implements artificial, intelligent agents for use in tactical flight training simulators. The overall goal of the project is to create automatic agents that generate behavior as similar as possible to humans flying flight simulators. These agents will help provide relatively cheap and effective training for Navy pilots. In order to accomplish this task, we need not only to acquire and encode a large amount of complex knowledge, but also to address a number of core research issue...