An Execution Architecture for Multiagent Anytime Scheduling


  1. Havens, B.
  2. Goebel, R.
  3. Berger, J.
Corporate Authors
Defence R&D Canada - Valcartier, Valcartier QUE (CAN);Alberta Univ, Edmonton ALTA (CAN);Simon Fraser Univ, Burnaby BC (CAN)
Dynamic real-time scheduling applications abound yet traditional optimization methods are inadequate. These tasks are characterized by rapidly changing requirements, the need for timely response and the desire for near optimal solutions. Recent work has indicated that anytime algorithms and multiagent systems could together be applied to real-time scheduling. We are concerned with developing a multiagent anytime scheduling system for helicopter mission planning and dispatching for the Canadian Forces. This document describes the architecture and an early experiment using the prototype system for Tactical Air Mission Planning/Scheduling (TAMP/S). The system uses multiple planning agents to incrementally produce plans for helicopter missions. The agents employ heuristic optimization algorithms to produce a single preferred plan within resource constraints and during dynamic environment changes. The anytime architecture defines various components, namely, a problem representation manager, a solution manager and, an environment manager. The problem representation manager provides an agent with an initial problem instance and processing resources. The solution manager evaluates computed solutions and reallocates resources according to individual agent progress. As for the environment manager, it makes information about external changes available to anytime planning agents. TRUNCATED

Il y a un résumé en français ici.

Adaptive intelligent system;Dynamic planning;Mission planning;TAMP/S (Tactical Air Mission Planning/Scheduling)
Report Number
DRDC-VALCARTIER-TR-2001-222 — Technical Report
Date of publication
29 Oct 2002
Number of Pages
Hardcopy;Document Image stored on Optical Disk

Permanent link

Document 1 of 1

Date modified: