Control of problem solving: principles and architecture

摘要: This paper presents an approach to the design of autonomous, real-time systems operating in uncertain environments. We address issues problem solving and reflective control reasoning under uncertainty terms two fundamental elements: 1) a set decision-theoretic models for selecting among alternative problem-solving methods 2) general computational architecture resource-bounded solving. The provide principles choosing based on their relative costs benefits, where benefits are characterized value information provided by output activity. may be estimate some quantity or recommendation action. architecture, called Schemer-II, provides interleaving communication various subsystems. These subsystems approaches gathering, belief refinement, solution construction, execution. In particular, mechanism interrupting response critical events. decision theoretic account scheduling elements critical-event-driven interruption activities such as Schemer-II.