A Multiagent Based Model for Tactical Planning
Abstract
The classical paradigm in planning consists of making a precise plan taking into account the whole set of variables that may intervene in the project and under the supposition that none of the variables will change during ;the project development. In case there is an unexpected change the project manager will decide which action to carry out. He will base his decision on his own experience, since he will have less time to study the complete area of new possibilities. For these reasons it is vital to make a project plan as accurate as possible. many times this objective is not easy to achieve, due to the different variables that intervene in the planning process. From a general point of view the planning process is a combination of tasks, resources and objectives in order to achieve a goal. In the planning phase a project team must define the different tasks and how long each task will take; the resources that can be used in order to do tasks and the goal or goals of the project. This paper presents a multiagent based model that permits to develop two prototypes in different contexts. One of these prototypes was introduced in the NMSG symposium held in Breda (The Netherlands) in 2001. The conceptual model has been improved in order to be applied in another different context. A second prototype has been developed under the same conceptual model and using similar Artificial Intelligence Tools. We use four different stimulus/response agents in order to solve specific functions, such as classifying, quantifying, assigning and finally optimizing the response of the computer. The first prototype is able to solve an artillery preparation or counter-preparation plan by using as few artillery units as possible. The second prototype solves the planning process in project management.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 2002
- Accession Number
- ADA425872
Entities
People
- Fernando De Arriaga
- Jos M. Castillo