Proceedings of the 1991 lEEE
Intematiod Conference on Robotics and Automation
Sacramento, California - April 1991
COORDINATION STRUCTURES FOR SPECIFICATION OF
INTEGRATION IN INTELLIGENT MACHINES
FEI-YUE WANG GEORGE N SAEUDIS
Department of Systems and Industrial Eng
University of Arizona, Tucson, AZ 85721
NASA CIRSSE, Cll
RPI, Troy, NY 12180
ABSTRACT
This paper presents an analytical model for the
Coordination Level of Intelligent Machines. The
framework of the CL is a tree structure consisting
of a dispatcher and a number of coordinators.
Petri net transducers have been used as the basic
module in the theory to describe individual task
translating processes of the dispatcher and co-
ordinators. Coordination structures are intro-
duced as a formalism for the specification of
integration in the CL. The task precedence rela-
tionship in the coordination process is presented
by the Petri nets derived from the coordination
structures. These Petri nets also provide us a
formal approach of using the concepts and an-
alysis methods in the Petri net theory to investi-
gate the properties of the coordination structures.
1. INTRODUCTION
An Intelligent Machine is composed of three
levels: 1) the Organizaion Level represents the
brain of the system with functions dominated by
AI to reason, plan, and make decisions about
task organization; 2) the Coordination Level
defines the interface between high and low levels
with functions dominated by OR that coordinate
the activities of the hardware; and 3) the
Execution Level is the lowest level with high
requirement in precision and with functions
dominated by Control Theory to execute the
specified tasks [7- 151 (Fig. 1).
The focus of this paper is on a formal theory
for the CL with a tree topology consisting of a
dispatcher as the root and a set of coordinators as
the the subnodes (Fig. 1). Specifically, the fol-
lowing features should be accomplished by this
theory: 1) a formal description of individual pro-
cess within each system unit; 2) a formal speci-
fication of the cooperation and connection among
system units; 3) a mechanism of control and
communication for the processes in system units.
Some results have been obtained along these
guidelines [ 12-14] based on the linguistic
approach developed in [lo]. In this approach, the
task processes of the dispatcher and coordinators
have been considered as the process of translat-
ing the high level task plans to the lo level control
actions by using vocabulary decision schemata.
Since the schemata are grammars, Petri net
transducers have been introduced to implement
them. The reason for choosing Petri net
transducers over the other automata is mainly due
to the facts that it has been shown in both
theories and applications, especially in the model-
ing of manufacturing processes [ 1,3-4,6,18],
that Petri nets can be used to specify connections
among the system units, to describe concurrence
and conflict in the system processes, and to
perform various process analysis.
The specification for the integration of the
dispatcher and coordinators has been achieved by
coordination structures. A coordination structure
is obtained by connecting a set of Petri net trans-
ducers with a set of connection points through
the connection mappings. Specifically, the use of
this coordination structure can enable us to: 1)
describe the task translation of the dispatcher and
coordinators; 2) represent the individual process
within the dispatcher and coordinators; 3) specify
the cooperation and connection among the
dispatcher and coordinators; 4) perform the
process analysis and evaluation; and 5) provide a
control and communication mechanism for the
coordination process.
The results obtained here can also be used
for the specification of other distributed systems,
e.g., the software system for the Manufacturing
Message Specification [ 16- 171. The work of
using Petri nets to formalize the decision struc-
ture of human organizations by Levis and his
colleagues [2,5] is quite similar to ours.
2690
CH2969-4/91/0000/2690$01.00 0 1991 IEEE