Date of Award

2004

Publication Type

Master Thesis

Degree Name

M.Sc.

Department

Computer Science

Keywords

Computer Science.

Supervisor

Goodwin, S.

Rights

info:eu-repo/semantics/openAccess

Abstract

Planning has been an important subject in the area of Artificial Intelligence (AI) for over three decades. Planning is the problem of seeking a series of actions (that is, a plan) that will accomplish a desired goal. Most planning approaches rely on a single processor or a single-agent paradigm. Unfortunately, in a complex world, a single agent may not be sufficient to optimally solve the problem. Distributed Planning is a sub-field of Distributed AI that involves multi-agents working together to solve large planning problems. Distribution may speed up the traditional planning system through parallelism. Hierarchical Task Network (HTN) planning is an AI planning methodology that creates plans by task decomposition. SHOP (Simple Hierarchical Ordered Planner) is a domain-independent HTN planning system designed by Dana Nau et al. that plans for tasks in the same order that they will later be executed. This thesis aims at designing and implementing a distributed version of SHOP (that is, DSHOP) and running it on a high performance distributed system called SHARCNET. The implementation is based upon Message Passing Interface (MPI), that is, a library of functions used to achieve parallelism via message-passing. We investigate two approaches to share work between processors: state-copying and state-recomputation. We implemented a state-copying based DSHOP system (DSHOPC), and a state-recomputation based DSHOP system (DSHOPR). We compared these two implementations of DSHOP with the Java version of SHOP on a set of randomly generated artificial domains. A set of experimental results has been used to evaluate the performance of the DSHOP algorithm.Dept. of Computer Science. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2004 .L83. Source: Masters Abstracts International, Volume: 43-01, page: 0240. Advisers: Scott Goodwin; Froduald Kabanza. Thesis (M.Sc.)--University of Windsor (Canada), 2004.

Share

COinS