A cost measure for efficient scheduling in intelligent machines

A model of the coordination level of Saridis's intelligent machine paradigm is presented. The role of the coordination level is to translate a high-level command into a schedule of low-level primitive activities. The model presented is based on hierarchical Petri nets and stochastic learning algorithms. An analytic cost measure of the scheduling structure of the coordination level is presented. For each high-level command, there are numerous ways to schedule the primitive activities that achieve that command. It is desired that the schedule that results in the minimum execution time of the intelligent machine. The cost measure derived, called structural complexity, predicts the expected execution time of an intelligent machine given a particular schedule of its primitives.<
>