An architecture for distributed knowledge based-systems
ACM SIGMIS Database
A resource based framework for planning and replanning
Web Intelligence and Agent Systems
| Hi-index | 0.00 |
TO COOPERATE EFFECTIVELY, INTELLIGENT AGENTS MUST BEHAVE PREDICTABLY IN ORDER TO COORDINATE THEIR PLANNED ACTIVITIES. IN DYNAMIC DOMAINS WHERE AGENTS CANNOT FULLY PREDICT FUTURE SITUATIONS, HOWEVER, PLANS ABOUT POSSI- BLE FUTURE ACTIONS AND INTERACTIONS MUST BE TENTATIVE AND MODIFIABLE, SO THAT AGENTS ARE FREE TO RESPOND TO UNEXPECTED SITUATIONS. COORDINATION IN DYNAMIC DOMAINS THUS INVOLVES BALANCING PREDICTABILITY AND RESPONSIVENESS: AGENT MUST BE PREDICTABLE ENOUGH TO ANTICIPATE AND PLAN FUTURE INTERACTIONS WHILE BEING RESPONSIVE ENOUGH TO REACT TO DOMAIN DYNAMICS. THE PARTIAL GLOBAL PLANNING APPROACH TO COORDINATION PROVIDES A FRAME- WORK FOR FLEXIBLY BALANCING PREDICTABILITY AND RESPONSIVENESS. IN THIS APPROACH, EACH INDIVIDUAL AGENT PLANS ITS ACTIVITES INCREMENTALLY: IT PRE- DICTABLY PURSUES A SEQUENCE OF MAJOR PLAN STEPS BUT RESPONSIVELY ADDS DE- TAILED ACTIONS TO ACHIEVE THOSE STEPS, INTERLEAVING PLANNING AND EXECUTION. THE AGENTS COMMUNICATE SUMMARIES OF THEIR LOCAL PLANS TO BUILD UP `PARTIAL GLOBAL PLANS'' (PGP''S) THAT SPECIFY COOPERATIVE ACTIONS AND INTERACTIONS. IN DYNAMIC DOMAINS WHERE LOCAL PLANS CHANGE, AGENTS MUST DETERMINE WHEN RESPONDING TO CHANGES IS WORTH THE TIME AND EFFORT OF REFORMULATING PGP''S, AND WHEN WORKING PREDICTABLY WITH SLIGHTLY OUT-OF-DATE PGP''S IS MORE COST- EFFECTIVE. IN THIS PAPER, WE BRIEFLY OUTLINE THE PARTIAL GLOBAL PLANNING APPROACH, DISCUSS HOW IT FLEXIBLY BALANCES PREDICTABILITY AND RESPONSIVENES