AN ADAPTIVE REAL-TIME AGENT FOR RESOURCE PLANNING WITHIN TACTICAL NAVAL DEFENCE

PDF

Authors
  1. Berger, J.
  2. Loranger, F.
  3. Sauve, M.
Corporate Authors
Defence Research Establishment Valcartier, Valcartier QUE (CAN)
Abstract
Reactive planning systems operating in a dynamic environment work in real-time but cannot guarantee efficient solutions, whereas deliberative systems are able to provide optimal solutions conditional on a perfect knowledge of the world and the use of unbounded computational resources. However, a real agent evolving in a dynamic world must deal with uncertainty, an incomplete information domain, finite and limited computational power, imprecise sensing, and must regard planning time as another degree of freedom. Accordingly, a tradeoff between reactive and deliberative planning must be reached considering planning as a base-level action. An adaptive real-time agent for resource planning within tactical naval defence is proposed. The single utility-driven agent model focuses on real-time issues related to planning, execution and control for closed-loop resource allocation. Real-time planning models and algorithms relying on anytime computation are briefly described and characterized. A computational experiment based on object-oriented discrete-event simulation to carry out performance analysis is presented in order to assess the value of the anytime planning approach for this application domain.
Report Number
DREV-9623 —
Date of publication
01 Aug 1997
Number of Pages
37
DSTKIM No
97-04150
CANDIS No
505695
Format(s):
Hardcopy;Document Image stored on Optical Disk

Permanent link

Document 1 of 1

Date modified: