Group Title: Department of Computer and Information Science and Engineering Technical Reports
Title: An Object-oriented multimodel approach to integrate planning, intelligent control and simulation
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00095191/00001
 Material Information
Title: An Object-oriented multimodel approach to integrate planning, intelligent control and simulation
Series Title: Department of Computer and Information Science and Engineering Technical Reports
Physical Description: Book
Language: English
Creator: Lee, Jin Joo
Norris, William Dean II
Fishwick, Paul A.
Publisher: Department of Computer and Information Sciences, University of Florida
Place of Publication: Gainesville, Fla.
Copyright Date: 1993
 Record Information
Bibliographic ID: UF00095191
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.

Downloads

This item has the following downloads:

1993103 ( PDF )


Full Text












An Object-Oriented Multimodel Approach to Integrate Planning,
Intelligent Control and Simulation


Jin Joo Lee


William Dean Norris II


Paul A. Fishwick


Department of Computer and Information Sciences
University of Florida
Gainesville, FL 32611


Abstract

The areas of planning, intelligent control and sim-
ulation have each spawned their own representational
structures. Deliberative planning approaches found
within AI are often rule based and simulation is of-
ten function based. Intelligent control approaches and
the perceived need to integrate reactive control with
deliberative planning has suggested an integration of
modeling techniques; however, well-formed integration
techniques are scarce. We propose using the multi-
model method for large scale systems containing in-
lelligent and non-intelligent objects. Our initial re-
sults with planning and control within a truck de-
pot show that, with multimodeling, current delibera-
live planning methods are extended to include a richer
set of model types besides production rule systems.
Moreover, reactive control is modulated by deliberative
planning model structures so that the entire model is
well-formed and easy to comprehend. [Key Words:
Autonomy, Intelligent Control, Multimodeling,
Planning, Object Oriented Simulation]


1 Introduction

The object oriented approach to simulation is dis-
cussed in different literature camps. Within computer
simulation, the system entity structure (SES) [20] (an
extension of DEVS [17, 16]) defines a way of organiz-
ing models within an inheritance hierarchy. In SES,
models are refined into individual blocks that contain
external and internal transition functions. Within the
object oriented design literature [18, 1], the effort is
very similar in that object oriented simulation is ac-


complished by building 1) a class model and 2) dy-
namic models for each object containing state infor-
mation. Harel [10, 11] defines useful visual model-
ing methods in the form of -I ,I.. 1. 11 so that the
dynamics may be seen in the form of finite state ma-
chines. From our perspective the object oriented ap-
proach provides an excellent starting point when de-
ciding how to organize information about dynamical
systems:

1. Start with a concept model of the system.

2. Create a class model using a visual approach such
as OMT [18]. This phase should involve creating
all relationships among classes.

3. Specify the dynamics for each class instance
where state transition is a factor. Note that
some classes will not contain state information
and some relations may not be of a dynamic na-
ture.

4. Construct a multimodel to build a network of
models each of which defines a part of the overall
system.

In the usual object oriented approach, phase three
translates to creating methods for an object that alter
the state of that object. The problem is that phase
three can be quite complex depending on the scale of
the system being modeled. There needs to be a way of
developing multi-level models that specify the phase
three dynamics. Our approach is to use multimod-
els [8, 4, 5, 6, 13] for this purpose. Multimodeling is a
paradigm for designing and executing models. We use
several well defined model types and connect them,
so that the lower levels refine the higher levels. Due










CIS TR93-022 TBP: Conf. on AI Simulation and Planning in High-Autonomoy Systems


to the hierarchical structure of the multimodel [4, 5]
approach, the object oriented paradigm is natural for
implementation. Each of the model types are exe-
cuted using the same methods, Initialize(, Input),
Output, State( and Update). Therefore, by having
a method of communication between each model type,
the models are executed regardless of which model
types are used.



2 Integrating Simulation and Planning

The idea of integrated simulation and planning de-
veloped from our efforts to overcome the many prob-
lems inherent in artificial intelligence (AI) and simula-
tion. Traditionally in AI planning, researchers are con-
cerned in building a planner that produces a symbolic
plan (order of primitive actions) which will achieve
certain tasks. However, the planner is unable to con-
trol the execution or modify the plan in order to guar-
antee success of the proposed plan. Traditional AI
planners were therefore unrealistic. Basically, these
deliberative planners assumed perfect knowledge of a
very static world. Taking the other extreme, other re-
searchers developed purely reactive planners that can
react to the environment by executing actions without
extensive reasoning.
The next logical step is to combine planning and
control to produce a planner that is both deliberative
and reactive. Dean [3] provides a good overview of
the various problems and techniques available in these
two areas. Most of the traditional planners which
have been built so far are either purely deliberative
or purely reactive. Recently, there have been some ef-
forts to develop a combined planner [12, 2, 9]. Due
to the divided research between deliberative and reac-
tive planners, the technology of the two fields has also
been divided. We believe the major difficulty in trying
to build a combined planner is integrating the different
methods of each area. Thus, multimodeling allows the
integration of these different techniques as submodels.
With planning and control combined, we now want
to integrate the different modeling types that exist in
AI and simulation. Some previous work [14, 15] has
been done in the integration of AI and Simulation.
The integration has two major advantages. Because
both the intelligent and non-intelligent objects are be-
ing modeled and simulated under one simulation, we
are able to test and evaluate the performance of the


Figure 1: Concept model of truck depot (Aerial view)


overall system. From the planning perspective, test-
ing, evaluation and modification can be done without
connecting the planner to an actual physical device or
object.
The Truck Depot problem was originally taken from
[3]. Since the problem contains both non-intelligent
objects (e.g. basin, trucks, valves) and intelligent ob-
jects (e.g. robots or people) in equal emphasis, the
problem inspired us to find a solution to optimization
by combining the two fields of simulation (simulating
non-intelligent objects) and AI (simulating intelligent
objects) under a unifying modeling paradigm.


3 A Truck Depot Example

Fig. 1 shows the aerial view of the truck depot,
which represents the concept model of system. The
depot contains one basin with two input pipes P1,P2
carrying two different chemicals and one output pipe
P3. Each pipe has a valve (VI, V2 and V3) which con-
trols the flow. Each valve has a servo motor attached
enabling the human operator to remotely close or open
the valves. Empty tanker trucks arrive at the depot
and wait until they can move under pipe P3 to be
filled with the mixture from the basin. When each
truck leaves the basin, its cargo is tested. If the truck
has been filled with an acceptable mixture, it leaves
the system; otherwise it dumps the cargo and returns
to be refilled. All mixture which overflows from the
basin or the truck is treated as waste. In our version
of the problem, the capacity of each tanker trucks is
constant.











CIS TR93-022 TBP: Conf. on AI Simulation and Planning in High-Autonomoy Systems


Figure 2: Instance model of truck depot


Empty Full
TEncks ITucks
Truck Depot


%
.


Non-Intelligent Object
(Basin, Valve, Truck)

Intelligent-Object
(Human)


Figure 3: Top view of system



The intelligent objects control the non-intelligent
objects (see Fig. 2). In our case, the human opera-
tor plans and controls the valves in the depot, while
achieving the goal of maximizing the profit of the de-
pot by maximizing the number of trucks filled while
minimizing the cost. The depot is charged for the to-
tal amount of chemicals that flow through the input
pipes during the period in which it is open. The trucks
are independent objects which arrive according to an
exponential distribution over the period of time when
the depot is open.
We also have the notion of time to consider in our
simulation. The start of simulation time corresponds
to an opening time of the depot and the end of simu-
lation time corresponds to the closing time in the real
world. Thus, any remaining mixture in the basin af-
ter closing time will also be waste. Fig. 3 shows the
control system of our simulation.


LEAST CRITICAL
V1,V2
StateofBasn OIMAL
CONTROL

--~--~--~~--~--~~--~---------------------






Level4 Intelligent Objects
V1LV2
MIXTMRn
State off Bas CONTROL

Levels. Be
put to be suppressed by a higher level for one time------------------

MOST CRmVCAL

Figure 4: Multimodel planner


4 Intelligent Objects


Because the arrival of the trucks is dynamic and
there is no determined number of trucks a prior, no
type of offline planning is possible. As illustrated in
Fig. 4, we have adopted Brook's subsumption archi-
tecture [2] to integrate the different level modules. Be-
cause our simulation is discrete, we will allow the out-
put to be suppressed by a higher level for one time
step until the next event arrives and causes another
output. The multimodel planner has multiple levels
that are divided based on how critical the reaction of
each level is to the overall success of the planner. The
levels also reflect the reactivity of the control in that
the lowest level module is the most reactive module
whereas the highest level is the least reactive. In gen-
eral, however, this may not always be the case.

4.1 Exception Control

At the lowest level in the hierarchy the Exception
Control module exists, which is the most reactive and
the most critical. In our problem, there are two crit-
ical situations. First, overflow either from the basin
or the truck must bed voided at any expense, since
spilled mixture can never be recovered. Second, un-
less it is close to the end of simulation time, the plan-
ner should avoid having an empty basin since no truck
can be filled. The Exception Control takes the state
of the basin as input, which is the volume of the mix-
ture in the basin, Bo,, and the volume of the mixture
in the truck, To,,i. With Bo,,, fuzzy logic [19] is used
to infer whether the basin is in an OVERFLOW or
EMPTY state. With To,,, fuzzy logic is used to de-
cide if the truck is in an OVERFLOW state. Only
when these conditions arise, does Exception Control










CIS TR93-022 TBP: Conf. on AI Simulation and Planning in High-Autonomoy Systems


Figure 5: FSA for Mixture Control


react and send an output command (CLOSE or OPEN
valve V,). NO-CHANGE which corresponds to a
null signal is sent otherwise.
At this level, the output fuzzy set is identical for
both valves V1 and V2. Because when an overflow
(or empty) occurs or is about to occur, both of the
valves need to be closed (or opened) at the same time.
To control the overflow of the truck, the volume of
the mixture in the truck is monitored and inferred
by fuzzy logic to be either in the state NOTFULL
or FULL. The output set for valve V3 is similar to
the one for valves V1 and V2. Fuzzy logic seems well
suited since we are not concerned precisely when these
events occur, but rather the appropriate time to start
monitoring and controlling to prevent overflow. The
fuzzy logic model best mimics the actual performance
of the human operator at this level.


4.2 Mixture Control


The Mixture Control module is responsible for
maintaining the correct ratio of the two chemicals in
the basin. Initially, the planner is given a ratio R
which is to be maintained throughout the simulation.
The mixture is considered acceptable and to be of the
correct ratio if the actual ratio r falls between the
range R- 5% and R+ 5%. The type of model used for
Mixture Control is a finite state automaton as shown


in Fig. 5. In the figure, the current state of the valves is
represented by the tuple (V1, V2) where V, can be one
of the following Open, Closed, Opening, Closing. De-
pending on the current state, the next state is reached
by choosing the appropriate transition. If the next
state reached is a transition state (state that contains
one or more valve settings ending with ing), an output
command is sent that will actually create the event to
change the physical state of the valves. However, the
output command may be suppressed by the higher
level, therefore never reaching the basin model. This
may also be the case for the Exception Control mod-
ule. For more detailed explanation of the suppressor
function, refer to [2].

4.3 Optimal Height Control

Finally, the Optimal Height Control module con-
trols the height in order to maximize the profit. Be-
cause this module is less reactive and involves more
symbolic knowledge and reasoning (heuristics) than
the other lower level modules, rule-based reasoning
is best suited for the task. The notion of optimal
height is time dependent. Since the simulation has
a start and an end time, the intelligent object can
have different strategies for maintaining the height at
different times. Another consideration that Optimal
Height Control module takes into account is the speed
of chemical flow. The flow rate of valve V3 depends
on the height of the mixture in the basin. Since our
Optimal Height Control module uses heuristics, op-
timality is not guaranteed. Included in this module,
is the Evaluator which evaluates the overall profit of
the system, during and after the end of simulation.
The formula used is Profit = N(Vgt) C(V) where
N is the amount of reward per unit of volume and C
is the amount of money charged per unit of volume.
Vgt represents the total volume of good trucks and V
represents the total volume of input.


5 Simulating Multimodels

5.1 What are Multimodels?

Models that are composed of other models, in a net-
work or graph, are called multimodels [4, 5, 7, 6, 8].
Multimodels allow the modeling of large scale systems
at varying levels of abstraction. They combine the ex-
pressive power of several well known modeling types










CIS TR93-022 TBP: Conf. on AI Simulation and Planning in High-Autonomoy Systems


such as FSAs, Petri nets, block models, differential
equations, and queuing models. By using well known
models and the principle of orthogonality we avoid
creating a new modeling system with a unique syn-
tax. When the model is being executed at the highest
level of abstraction, the lowest level (representing the
most refined parts of the model) is also being executed.
Each high level state duration is calculated by execut-
ing the refined levels that have a finer granularity.

5.2 Why use Multimodels?

For the Truck Depot example the question arises,
"Why use multimodels?' The non-intelligent objects
in the system could be modeled using control func-
tions since classical control theory would provide an
optimal solution. Therefore, using a multimodel may
not seem necessary. However, multimodels offer sev-
eral advantages over classical control theory.

Extensibility: A multimodel is able to be ex-
tended. Making a change to the model, (e.g.
adding evaporation of the mixture while in the
basin or having a variable filling capacity for the
tanker trucks) is difficult, if not impossible, to
implement when using classical control theory.

Replaceability: Any of the objects in the system
can be replaced by another object that accepts
the same input and gives the same output. For
example, the two input valves can be replaced by
three input valves.

Reusability: The planner, the basin or the entire
truck depot could be used within the context of
a much larger model containing the depot as a
component.

Comprehensibility: Any physical system can be
modeled using classical control theory, but it has
drawbacks. For example, the equations for large
scale systems become prohibitively complex and
unsolvable when small changes are made to the
system.

5.3 How to use Multimodels

Each type of model (e.g. Petri net, block, differ-
ential equation, FSA, queuing) has similar features:
input, output, state, and a transition from one state


to the next. Some model types have lower level com-
ponents that encode information about the model. In
an FSA, each state holds all of the information that
is needed to answer any question about the model. In
contrast, a transition and place in a Petri net each
only have information about a subset of the system,
although when combined, they describe the complete
system.
To simulate a multimodel, it is necessary to have
the input from one model type accept the output from
another type of model. Each model must be able to
recognize the output of any refining model as valid
input.
To synchronize the system at its highest level, a
coordinator is used to process the external inputs. The
coordinator also creates and initializes each model and
its components, and then organizes the models into
the specified hierarchy. The coordinator executes the
refining models, but allows only external output from
the levels above the specified level. The coordinator
uses a future event list (FEL) to keep track of: 1) the
next event, 2) to which model or model component the
event should be sent, 3) which token caused the event,
and 4) the global time of the simulation. Each level of
the model must wait for an event to begin execution,
then it posts its new state to the FEL.
The following methods are used to simulate each
of the model types. ReadModel( reads the model
specifics and creates the model instance. Initialize(
describes how the model and its refining models are
initially set. Inputi collects all of the input data that
a model receives during execution. State( returns the
current system state. Update) causes the input or
event to be applied to the current state and the next
state to be scheduled on the FEL. Output) returns
the output of the current state.



6 Non-Intelligent Objects

6.1 Model Design

Modeling the basin poses several challenges. The
model state includes continuous and discrete variables,
constraints, and functional relationships.
The volume of the mixture in the basin changes con-
tinuously throughout the simulation. The input sig-
nals that control the three valves give continuous out-
puts, but they change at discrete times. The tanker











CIS TR93-022 TBP: Conf. on AI Simulation and Planning in High-Autonomoy Systems


VALVE CONTROL SIGNALS FULL TRUCKS
Basin


Fill Basin Fill Tnick

Figure 6: Petri net model of the basin


V, Control Signal


Mixture Volume in Basin

V2 Control Signal V



_-_-_-------------_-___---_-_----__-_---_-_--_-_-_-_-_--_-_-_-_-_-

Figure 7: Refinement of transition Fill Basin


trucks move through the system as discrete objects,
and are constrained when waiting to be filled because:
the signal for valve V3 must open the valve; the basin
must have enough mixture to fill a truck; and the fill-
ing area must be empty.
We chose a Petri net to model the top level, see
Fig. 6. The inputs to the model are tanker truck ar-
rivals and control signals for opening and closing each
of the valves. The output from the model includes
statistics showing the number of trucks that were filled
properly, and the volume of each chemical that was
poured into the basin during the simulation.
The Truck Queue transition is refined by an S/S/1
queuing model. It was chosen to model the tanker
trucks waiting to be filled. The queue is used to main-
tain the trucks arrival order.
The transition Fill Basin is refined by a block
model, see Fig. 7. The control signals are passed to
the refining models Valve 1 and Valve 2 respectively.


Figure 8: FSA model of valve V1

Vbasm > ck Vbasm >= 2Vtuck Vbasin =3Vuck Vbasn > 3Vtck


0 2 3 Overflow


Vbasm Vtck Vbasm 2Vtck Vbasm< 3Vtck Vbasin 3Vmck

Figure 9: FSA model of the volume of mixture in the
basin


Each refining model returns a value representing the
control to be applied. The function block, Mixture
Volume in Basin takes the control input and returns
a value that states how many tanker truck loads the
basin currently holds.
Fill Truck is also refined by a block model (sim-
ilar to Fig. 7) that shows the relationship between
valve V3 and the Mixture Volume in Truck function
block. Valve V3 takes as input the control signals,
OPENV3 and CLOSEV3, then returns the amount
of control to be applied due to the current state of
valve V3. The function block Fill Truck takes the con-
trol value as input and returns a value when the truck
has been filled.
Each of the function blocks are refined by a finite
state automaton. There are five FSAs, but the three
FSAs that refine the valves are similar to the refine-
ment of valve V1.
The function block Valve V1 is refined by the FSA
shown in Fig. 8. The input for this FSA is the control
signal for the valve V1, (Vi = 1 or V1 = 0). The FSA
will change state if an internal transition is detected,
(0 = 1 or 01 = 0).
The Fill Basin function is refined by the FSA dis-










CIS TR93-022 TBP: Conf. on AI Simulation and Planning in High-Autonomoy Systems


played in Fig. 9. The input for this FSA is the amount
of control that is being applied to the system by valves
V1 and V2. The output is how many truck loads of
mixture the basin contains.
The function block Mixture Volume in Truck is re-
fined by an FSA with three states. The input to this
FSA is the amount of control applied by valve V3, the
system output is the state of the truck, whether the
truck is FILLING, FULL or OVERFLOWING.
The control equation x = Ax +Bu is used to model
the continuous state variables in the system. Where x
is the subsystem state, u is the control from the valves,
and B is the amount of control being applied.

6.2 Model Execution

A coordinator is used to create and execute the mul-
timodel system. After each level of the model is cre-
ated, the levels are connected to their refining models,
then each level is initialized. Our initial conditions for
the truck depot example are the basin is empty, the
valves are closed, and no trucks are waiting to be ser-
viced. The coordinator's task during execution is to
dequeue events from the FEL and direct the event to
the model specified.


7 Conclusions

Through our truck depot example, we demon-
strated how to integrate simulation and planning tasks
under the object-oriented multimodel framework. The
designing of the model for our system is performed
through the 3 phases 1) concept model, 2) class model
and 3) instance model with the relationships specified
as discussed in section 1. As shown, multimodeling
cannot only be used to integrate different type models
in a hierarchy but also used to integrate model types
coming from completely different background or dis-
ciplines.
For future work, we would first like to extend our
truck depot example to include more dynamic proper-
ties such as varying tanker truck capacity and allowing
the planner to reorder the trucks after they arrive. We
would also like to experiment with different intelligent
objects, such as a mobile robot and adding more con-
straints to the problem such as asynchronous control
of the valves. Using adaptive control (e.g. adaptive
fuzzy systems) will be another extension. For multi-
modeling in general, we will add a graphical interface


for creating the models and a distributed persistent
object database to store all of the objects which have
been created. Then a user could load any object from
the Internet.

Acknowledgments

We would like to thank the Institute for Simula-
tion and Training (IST) at the University of Cen-
tral Florida for partial funding of this research in
connection with the \1--.... Planning" sub-contract
#307043.


References

[1] G. Booch. Object Oriented Design. Benjamin
Cummings, 1991.

[2] R. A. Brooks. A robot layered control system for
a mobile robot. IEEE Journal of Robotics and
Automation, 2:14 23, 1986.

[3] T. L. Dean and M. P. Wellman. Planning and
Control. Morgan Kaufmann, 1991.

[4] P. A. Fishwick. Heterogeneous Decomposition
and Coupling for Combined Modeling. In 1991
Winter Simulation Conference, pages 1199
1208, Phoenix, AZ, December 1991.

[5] P. A. Fishwick. An Integrated Approach to Sys-
tem Modelling using a Synthesis of Artificial In-
telligence, Software Engineering and Simulation
Methodologies. AC if Transactions on Modeling
and Computer Simulation, 2(4):307 -330, 1992.

[6] P. A. Fishwick. A Simulation Environment for
Multimodeling. (accepted for publication), 1993.

[7] P. A. Fishwick. Computer Simulation Model De-
sign C Execution. Prentice Hall, 1993. (to be
published as a textbook).

[8] P. A. Fishwick and B. P. Zeigler. A Multimodel
Methodology for Qualitative Model Engineering.
AC if Transactions on Modeling and Computer
Simulation, 1(2):52 -81, 1992.

[9] I. Futo and T. Gergely. Artificial Intelligence in
Simulation. Ellis Horwood Limited/John Wiley
and Sons, 1990.










CIS TR93-022 TBP: Conf. on AI Simulation and Planning in High-Autonomoy Systems


[10] D. Harel. On Visual Formalisms. Communica-
tions of the AC i1, 31(5):514 530, May 1988.

[11] D. Harel. Biting the Silver Bullet: Toward a
Brighter Future for System Development. IEEE
Computer, 25(1):8 -20, January 1992.

[12] L. P. Kaelbling. An architecture for intelligent
reactive systems. In Reasoning About Actions and
Plans, pages 395 -410. Morgan Kaufmann, Los
Altos, CA, 1 i.

[13] V. T. Miller. Heterogeneous Hierarchical Mod-
elling for Knowledge-Based Autonomous Sys-
tems. PhD thesis, University of Florida, 1993.

[14] N. R. Nielsen. Applications of AI Techniques to
Simulation. In P. Fishwick and R. Modjeski, ed-
itors, Knowledge Based Simulation: Methodology
and Application, pages 1 19. Springer Verlag,
1991.

[15] R. M. O'Keefe. The Role of Artificial Intelligence
in Discrete Event Simulation. In L. E. Widman,
K. A. Loparo, and N. R. Nielsen, editors, Arti-
ficial Intelligence, Simulation & Modeling, pages
359 -379. John Wiley and Sons, 1989.

[16] H. Praehofer. Theoretic Foundations for Com-
bined Discrete Continuous System Simulation.
PhD thesis, University Linz, Austria, 1991.

[17] H. Praehofer, G. Auernig, and Reisinger G. An
Environment for DEVS-Based Modeling in Com-
mon Lisp/CLOSS. (accepted for publication),
1993.

[18] J. Rumbaugh, M. Blaha, W. Premerlani, E. Fred-
erick, and W Lorenson. Object-Oriented Modeling
and Design. Prentice Hall, 1991.

[19] L. A. Zadeh. Fuzzy Logic. IEEE Computer, pages
83 -93, April 1988.

[20] B. P. Zeigler. Object Oriented Simulation with
Hierarchical, Modular Models: Intelligent Agents
and Endomorphic Systems. Academic Press,
1990.




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs