LINGUISTIC METHODS FOR HIERARCHICALLY INTELLIGENT CONTROL
Abstract
The purpose of this research is to develop a new, hierarchical, linguistic based, learning controller for complex systems. The complete system will be able to interact with a human operator in a limited natural language at the highest level of the hierarchy, but will be able to control detailed motions of some complex physical system at the lowest level of the hierarchy. Each level of the heirarchy is defined by a formal grammar which can generate exactly the class of admissible control actions at that level. A new linguistic structure, the linguistic decision schema, is proposed to model the translation or mapping of commands between levels of the hierarchy. In the most general form, the decision schema incorporates a learning algorithm to yield asymptotically optimal mappings for control under stochastic environments. Variations of the general linguistic decision schema are presented for solving specific classes of linguistic mappings. The computational requirements of the decision schema are analyzed, and a special digital computer architecture is proposed to perform the assignment and update functions of the schema. Also presented is a case study of the application of the linguistic based, hierarchically intelligent control method to a manipulative system for incapacitated hospital patients.
Degree
Ph.D.
Subject Area
Electrical engineering|Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.