A MULTI-AGENT EPISTEMIC PLANNING MODEL FOR MANUFACTURING MANAGEMENT

THIS EFFORT ADDRESSES THE NEED FOR ADEQUATE REPRESENTATIONS OF MANAGEMENT AND MANUFACTURING KNOWLEDGE AND ADEQUATE REASONING STRATEGIES UTILIZING THIS KNOWLEDGE TO SUPPORT AUTOMATED PLANNING PREDICTION AND EXPLANATION OF ORGANIZATIONAL ACTIVITIES. MANAGEMENT DECISION MAKING IS OFTEN BASED ON INCOMPLETE AND UNCERTAIN INFORMATION AND ON BELIEFS AND KNOWLEDGE THAT ARE CHANGING OVER TIME. MANAGEMENT INVOLVES THE INTERACTION OF MULTIPLE AGENTS WITH DIFFERENT PERSPECTIVES WHOSE CONCLUSIONS MAY BE CONFLICTING AND CONTRADICTORY. UNDERSTANDING OF PERSONAL INTERACTIONS IN MANAGEMENT INVOLVES CONCEPTS SUCH AS ABILITY, PERMISSIBILITY AND OBLIGATION. THE TRADITIONAL AI PLANNING PARADIGM REQUIRES PRECONDITIONS AND POSTCONDITIONS OF ACTIONS BE KNOWN. THIS PARADIGM IS NOT USEFUL IN THE MANAGEMENT PLANNING DOMAIN WHERE KNOWLEDGE IS INCOMPLETE, UNCERTAIN, AND BASED ON CHANGING BELIEFS ABOUT INTENTIONS, CAPABILITIES, OBLIGATIONS, ETC. OUR APPROACH TO THIS PROBLEM IS TO REPRESENT EPISTEMIC, MODAL, AND INTENTIONAL CONCEPTS AS EXTENSIONS OF FIRST ORDER LOGIC AND TO FORMULATE AUTOMATED PLANNING MODEL BASED ON DEDUCTIVE, PREDICTIVE AND HYPOTHETICAL REASONING STRATEGIES FOR THIS EXTENDED LOGIC.