Sponsored by
IEEE Singapore
SMC, R&A, and
Control Chapters
Organized and
invited by
Professor Sam Ge,
NUS
Wireless Sensor Networks for Monitoring Machinery,
Human Biofunctions, and BCW Agents
F.L. Lewis, Assoc. Director for Research
Moncrief-O’Donnell Endowed Chair
Head, Controls, Sensors, MEMS Group
Automation & Robotics Research Institute (ARRI)
The University of Texas at Arlington
F.L. Lewis, Assoc. Director for Research
Moncrief-O’Donnell Endowed Chair
Head, Controls, Sensors, MEMS Group
Automation & Robotics Research Institute (ARRI)
The University of Texas at Arlington
Discrete Event Control & Decision-Making
http://ARRI.uta.edu/acs
Discrete Event Control
$75K in ARO Funding for Networked Robot Workcell Control
$80K in NSF Funding for research and USA/Mexico Network
Objective:
Dr. Jose Mireles- co-PI
•
•
•
•
•
•
•
•
Develop new DE control
algorithms for decision-making,
supervision, & resource
assignment WITH PROOFS
Apply to manufacturing
workcell control, battlefield C&C
systems, & internetworked
systems
Patent on Discrete Event Supervisory Controller
New DE Control Algorithms based on Matrices
Complete Dynamic Description for DE Systems
Formal Deadlock Avoidance Techniques
Implemented on Intelligent Robotic Workcell
Internet- Remote Site Control and Monitoring
USA/Mexico Collaboration
Exploring Applications to Battlefield Systems
Intelligent Robot Workcell
Man/Machine User Interface
Texas
USA/Mexico Internetworked Control
Matrix Formulation: Definition
Based on Manufacturing Bill of Materials
DE Model State Equation:
x  Fv vc  Fr rc  Fu u  FDuD
Where multiply = AND & addition = OR
x is the task or state logic
where
Fv is the job sequencing matrix (Steward)
Fr is the resource requirements matrix (Kusiak)
Fu is the input matrix
FD is the conflict resolution matrix
Job Start Equation:
Resource Release Equation:
Product Output Equation:
Vs  Sv x
rs  S r x
y  Syx
Meaning of Matrices
Resources required
Prerequisite jobs
Next
job
Fv
Steward’s Task Sequencing Matrix
Conditions fulfilled
Next
job
Sv
Next
job
Fr
Kusiak’s Resource Requirements Matrix
Bill of Materials (BOM)
Conditions fulfilled
Release
resource
Sr
EXAMPLE
ARRI Intelligent Material Handling (IMH) Cell
3 robots, 3 conveyors, two part paths
Layout of the IMH Cell
IBM robot
R1
Conveyor
bidirectional
B3
Conveyor
unidirectional
X1
X9
A
B
A
B1
B
X8
X2
X3
M2
X4
ADEPT robot
R3
M1
machine
X7
machine
X5
X6
B2
conveyor
PUMA robot
R2
Part A job 1
Part B job 1
Part A job 2
Part B job 2
Part A job 3
Part B job 3
Construct Job Sequencing Matrix Fv
Used by Steward in
Manufacturing
Task Sequencing
Part A job 1
Part A job 2
Part A job 3
Prerequisite
jobs
Next
jobs
Part B job 1
Part B job 2
Part B job 3
Contains same information
as the Bill of Materials
(BOM)
Part A job 1
Part A job 2
Part A job 3
Robot 1- IBM
Robot 2- Puma
Robot 3- Adept
Conveyor 1
Conveyor 3
Fixture 1
Construct Resource Requirements Matrix Fr
Used by Kusiak in
Manufacturing
Resource Assignment
Contains information
about factory resources
Prerequisite
resources
Part B job 1
Part B job 2
Part B job 3
Next
jobs
More About Fv
J2
J5
J6
J3
J1
J3
J5
J4
1 0 0
0 1 1 


1 0 0
J4
Two 1’s in same row = Assembly
J2
DECISION
NEEDED!
J1
J6
Two 1’s in same col. = Routing (Job Shop)
More About Fr
R1
J2
J5
J6
R2
R2
R3
1 0 0
0 1 1 


1 0 0
Two 1’s in same row
= Job needs multiple res.
Two 1’s in same col.
= Shared Resource
J5
R3
J2
R1
J6
DECISION
NEEDED!
Controller based on Matrix Formulation
Resource allocation, task planning,
task decomposition, Bill of Materials
Dispatching
rules
Matrix Formulation
Discrete Event Controller
External Events
Start jobs
Start resource release
Task complete
Workcell
External events present
Jobs completed
Resources released
Tasks completed
uc
Dispatching rules
Rule-Based Real-Time Controller
Controller state monitoring logic
x  Fv  v  Fr  r  Fu  u  Fu c  uC
.. .
Job start logic
v S  SV  x
Resource release logic
rS  S r  x
 S y  x logic
Tasky complete
Parts
present u
Parts in pin
Start tasks vs
Start resource
release rs
Output y
Plant commands
Work Cell
Tasks
completed vc
Resource
released r c
Products pout
Plant status
Advantages of the Matrix Formulation
•
•
•
•
•
•
•
•
•
Formal rigorous framework
Complete DE dynamical description
Relation to known Manufacturing notions
Formal relation to other tools- Petri Nets, MAX-Plus
Easy to design, change, debug, and test
Formal deadlock analysis technique
Easy to apply any conflict resolution (dispatching) strategy
Optimization of resources
Easy to implement in any platform (MATLAB, LabVIEW, C,
C++, visual basic, or any other)
Relation to Petri Nets
Jobs complete
Resources available
Trans.
Trans.
Fv
Transition
Next
jobs
Sv
Fr
Transition
Release
resource
Sr
Example
r1
t1
pinA
p1
t2
p2
t3
poutA
r2
pinB
t4
p1 p2 p3 p4
t1
t2
t3
t4
t5
t6
0
1

0
Fv  
0
0

0

0
0
0
0
1
0
0
0
0
1
0
0
p1 p2 p3 p4
t1
t2
t3
t4
t5
t6
Sv
T
1
0

0
 
0
0


0
t5
p3
0
0
1
0
0
0
0
1
0
0
0
0
0
0

0

0
0

1

0
0

0

0
1

0

t6
p4
r3
pinA pinB
r1 r2 r3
Fr
1
0

0
 
0
0

0

0
1
0
0
1
0
0
0

0

1
0

0

Fu
Sr
T
0
0
1
0
0
1
0
0

0

0
1

0

1
0

0
 
0
0

0

0
0

0

1
0

0

poutA poutB
r1 r2 r3
0
1

0
 
0
0


0
poutB
Sy
T
0
0

1
 
0
0


0
0
0

0

0
0

1

OR/AND Algebra- Locating transitions firing from current marking
r1
t1
pinA
p1
t2
p2
t3
poutA
r2
pinB
t4
t5
p3
t6
p4
poutB
r3
0
1

0
x  
0
0


0
x=
0 
0 
 
1 
 
0 
1 
 

1 

Fv
0
0
0
0
1
0
0
0
0
1
0
0
0
0

0

0
0

1

1 
0 
 
0 
 
0 
0 
 

0 

v
1
0

0

0
0


0
0 
1 
 
1 
 
1 
0 
0 
 
0 
 
0 
0 
 

0 

Fr
0
1
0
0
1
0
=
r
0
0

0

1
0

0

1 
0 
 
1 
 
0 
1 
 

1 

,
1 
0 
 

0 

Fu
1
0

0

0
0


0
so x =
u
0
0

0

1
0

0

0 
1 
 
0 
 
1 
0 
 

0 

0 
0 
 
i.e. fire t2 and t4
Complete DE Dynamic Formulation
Activity Completion Matrix F:
F  [ Fu Fv Fr Fy ]
Activity Start Matrix S:
S  [ Su
T
Sv
T
Sr
T
T
Sy ]
PN Incidence Matrix:
M  S  F  [Su  Fu , Sv  Fv , Sr  Fr , S y  Fy ]
T
T
T
T
T
PN marking transition equation:
m(t  1)  m(t )  M T x  m(t )  [ S T  F ]x
Allowable marking vector:
xk  F  mk  [ Fu Fv Fr Fy ]  [ PI v r PO]k
Petri Net Marking Transition Equation-need to add Job Duration Times
m(t )  ma (t )  m p (t )
PN Marking Vector
Split transition equation in two steps
m p (t  1)  m p (t )  S T x(t )
Add tokens
ma (t  1)  ma (t )  F x(t )
Subtract tokens when job complete
T  [O, vtimesT , rtimes T , O]T
Add Time Duration Vector
Tpend (t  1)  diag{ m p (t ) }[Tpend (t )  t sample]  diag{ S T x(t )}T
m p (t )  m p (t )  m finish(t )
ma (t )  ma (t )  m finish (t )
Corresponds to Timed Places
Allows Direct Simulations- e.g. MATLAB
c.f. DE version
of ODE23
Jobs completed
by Robot 1
Robot 1
busy or idle
Conflict Resolution for Shared Resources
r1
t1
pinA
p1
t2
p2
Which one to fire?
pinB
t4
t3
poutA
r2
t5
p3
t6
p4
poutB
r3
0
1

0
x  
0
0


0
Fv
0
0
0
0
1
0
0
0
0
1
0
0
0
0

0

0
0

1

Fr
v
1
0

0

0
0


0
0 
1 
 
0 
 
1 
0
1
0
0
1
0
r
0
0

0

1
0

0

1 
0 
 

1 

Fu
1
0

0

0
0


0
u
0
0

0

1
0

0

0 
0 
 
Shared Resource- Two entries in same column
0 
0 
 
1 
 
0 
0 
 

1 

1 
0 
 
0 
 
1 
0 
 

0 

0 
0 
 
0 
 
0 
0 
 

0 

=
1 
0 
 
1 
 
1 
0 
 

1 

, so x =
0 
1 
 
0 
 
0 
1 
 

0 

But gives negative
marking!
Cannot fire both.
Conflict resolution, add extra CR input and new matrix Fuc:
r1
p1
t1
pinA
t2
p2
r2
pinB
poutA
r2
p3
t4
t3
t5
t6
p4
poutB
r3
0
1

0
x  
0
0

0

Fv
0
0
0
0
1
0
0
0
0
1
0
0
0 
0 
 
1 
 
0 
0 
 

1 

0
0

0

0
0

1

1 
0 
 
0 
 
1 
0 
 

0 

Fr
v
0 
1 
 
0 
 
1 
0 
0 
 
0 
 
0 
0 
 

0 

1
0

0

0
0

0

0
1
0
0
1
0
r
0
0

0

1
0

0

0 
1 
 
0 
 
0 
0 
 

0 

=
1 
0 
 

1 

1 
1 
 
1 
 
1 
0 
 

1 

Fu
1
0

0

0
0

0

u
0
0

0

1
0

0

, so x =
0 
0 
 
0 
0 
 
0 
 
0 
1 
 

0 

Fuc
0
1

0

0
0

0

0
0

0

0
1

0

r2
1 
0 
 
Now only t5 fires
Application- Intelligent Material Handling
Machine 1
Adept
Puma
Machine 2
CRS
12 Sensors!!
ARRI Intelligent Material Handling (IMH) Cell
3 robots, 3 conveyors, two part paths
Layout of the IMH Cell
IBM robot
R1
Conveyor
bidirectional
B3
Conveyor
unidirectional
X1
X9
A
B
A
B1
B
X8
X2
X3
M2
X4
ADEPT robot
R3
M1
machine
X7
machine
X5
X6
B2
conveyor
PUMA robot
R2
Multipart Reentrant Flow Line
PART B OUT
PART A
PART A OUT
A(1)R2
A(1)R1
CRS
ROBOT 1
PART B
c.f. Kumar
PUMA
B(1)R1
B(1)R2
A(2)R1
ROBOT 2
B(2)R1
A(2)R2
Machine 1
ADEPT
B(1)R3
ROBOT 3
Machine 2
B(2)R3
A(1)R3
Petri Net flow chart
B1AA
PAI
X1
R1U1
X2 B1AS
X3
R2U1 X4
B3AA
B2AA
M1A
M1P
X5
R2U3
X6
B2AS
R3U1
X10 PAO
X8
B3AS
X9
R1U3
X17 R3U3 X18
B3BS
X19
R1U4 X20
X7
R3A
R2A
R1A
PBI
X11 R1U2 X12
B1BA
B1BS
X13 R2U2 X14
B2BS
B2BA
X15
R3U2 X16
M2A
M2P
B3BA
PBO
PC with High Level Controller
c.f. Saridis
Jim Albus
Rule -Based Real -Time Controller
Controller state monitoring logic
x  Fv  v  Fr  r  Fu  u  FD  u D  FuC  u C
Start tasks/jobs
uc
Dispatching
rules
To Generate uc
Job start logic
vS =S v  x
Resource release logic
rs  S r  x
Task complete logic
y =S y  x
Tasks: vSA
Medium Level Tasks Controllers
Robot 1
Task 4
Task 3
Task 2
Task 1
Robot 2
Task 3
Task 2
Task 1
Robot 3
v~SA
Task 3
Task 2
Task 1
Workcell data
gathering
u
v
r
p
Parts out
~
p in , ~
rS , v~SB
DAQ -card
RS232 -1
RS232 -2
RS232-3
Analog & digital I/0
Jobs
vSB
rSB
vSB
Low level PD & PID controllers
CRS
controller
Puma 560
controller
ADEPT One
controller
rSA, pin
Machines
Sensors
Robots
LabVIEW diagram of Controller
LabVIEW Controller's interface:
Resources
Fv
Fr
Results of LabVIEW Implementation on Actual Workcell
R1u1
R1u2
R1u3
R1u4
R2u1
R2u2
R2u3
R3u1
R3u2
Discrete events
Compare with MATLAB simulation!
We can now simulate a DE controller and then implement it,
Exactly as for continuous state controllers!!
U.S.-Mexico shared research
DE control via internet
Texas
Using Matrix DEC in
LabVIEW