ijcai-tutorial
119 Pages
English
Downloading requires you to have access to the YouScribe library
Learn all about the services we offer
Downloading requires you to have access to the YouScribe library
Learn all about the services we offer
119 Pages
English

Description

J6XG5HQDWL8FHEEQZDWLS9LHZSQLILHKDP3GYDODKDP6ELQ.DPJ.D$DUDRHV8LQQLILDUDR9LHE$$,5DQQDQF $GY5HFHQW $GYDQFHV LQ $, 3ODQQLQJ$ 8QLILHG 9LHZ,-&$, 7XWRULDO6XEEDUDR.DPEKDPSDWL$UL]RQD 6WDWH 8QLYHUVLW\rao@asu.eduhttp://rakaposhi.eas.asu.edu/ijcai99 tutorialHF HQW W$ QFHV L XE PEQ$, 3O QQLHGJ6XG5HQDWL8FHEEQZDWLS$SQLILHKDP6GYDODKDP5ELQ.DPJ.D$DUDRHV8LQQLILDUDR9LHZE9LH$,3DQQFDQ $GYPlanning is hot...26% of the papers in AAAI 99. 20% of papers in IJCAI-99. New People. Conferences. Workshops. Competitions.Inter planetary explorations. Why the increased interest?* Significant scale up in the * Significant strides in ourlast 4 5 years understanding– Before we could – Rich connections betweensynthesize about 5 6 planning and CSP(SAT)action plans in minutes OR (ILP)– Now, we can synthesize » Vanishing separation100 action plans in between planning &minutes SchedulingFurther scale up with New ideas for heuristic» –domain specific control of plannerscontrol– Wide array of approachesfor customizing plannerswith domain specificknowledgeHF HQW W$ QFHV L XE PEQ$, 3O QQLHGJ6XG5HQDWL8FHEEQZDWLS$SQLILHKDP6GYDODKDP5ELQ.DPJ.D$DUDRHV8LQQLILDUDR9LHZE9LH$,3DQQDQF $GYOverview* &ODVVLFDO SODQQLQJ SUREOHP† 0RGHOLQJ 3URYLQJ FRUUHFWQHVV* 5HILQHPHQW 3ODQQLQJ )RUPDO )UDPHZRUN* &RQMXQFWLYH UHILQHPHQW SODQQHUV† +HXULVWLFV* ...

Subjects

Informations

Published by
Reads 17
Language English

Exrait

J
6X
G
5H
Q
DWL
8
FH
EE
Q
Z
DWL

S
9LHZ
S
QLILH
KDP
3
GYD
OD
KDP
6
E
LQ
.DP
J
.D
$
DUDR

HV
8
LQ
QLIL
DUDR
9LH
E
$
$,
5
DQQ
DQF $GY
5HFHQW $GYDQFHV LQ $, 3ODQQLQJ
$ 8QLILHG 9LHZ
,-&$, 7XWRULDO
6XEEDUDR.DPEKDPSDWL
$UL]RQD 6WDWH 8QLYHUVLW\
rao@asu.edu
http://rakaposhi.eas.asu.edu/ijcai99 tutorial
HF HQW W$ QFHV L XE PE
Q$, 3O QQL
HGJ
6X
G
5H
Q
DWL
8
FH
EE
Q
Z
DWL

S
$
S
QLILH
KDP
6
GYD
OD
KDP
5
E
LQ
.DP
J
.D
$
DUDR

HV
8
LQ
QLIL
DUDR
9LHZ
E
9LH
$,
3
DQQ
FDQ $GY
Planning is hot...
26% of the papers in AAAI 99. 20% of papers in IJCAI-99.
New People. Conferences. Workshops. Competitions.
Inter planetary explorations. Why the increased interest?
* Significant scale up in the * Significant strides in our
last 4 5 years understanding
– Before we could – Rich connections between
synthesize about 5 6 planning and CSP(SAT)
action plans in minutes OR (ILP)
– Now, we can synthesize » Vanishing separation
100 action plans in between planning &
minutes Scheduling
Further scale up with New ideas for heuristic
» –
domain specific control of planners
control
– Wide array of approaches
for customizing planners
with domain specific
knowledge
HF HQW W$ QFHV L XE PE
Q$, 3O QQL
HGJ
6X
G
5H
Q
DWL
8
FH
EE
Q
Z
DWL

S
$
S
QLILH
KDP
6
GYD
OD
KDP
5
E
LQ
.DP
J
.D
$
DUDR

HV
8
LQ
QLIL
DUDR
9LHZ
E
9LH
$,
3
DQQ
DQF $GY


Overview
* &ODVVLFDO SODQQLQJ SUREOHP
† 0RGHOLQJ 3URYLQJ FRUUHFWQHVV
* 5HILQHPHQW 3ODQQLQJ )RUPDO )UDPHZRUN
* &RQMXQFWLYH UHILQHPHQW SODQQHUV
† +HXULVWLFV
* ’LVMXQFWLYH UHILQHPHQW SODQQHUV
† 5HILQHPHQW RI GLVMXQFWLYH SODQV
† 6ROXWLRQ H[WUDFWLRQ IURP GLVMXQFWLYH SODQV
“ ’LUHFW &RPSLOHG 6$7 &63 ,/3
* &XVWRPL]LQJ 3ODQQHUV
† 8VHUDVVLVWHG &XVWRPL]DWLRQ
† $XWRPDWHG FXVWRPL]DWLRQ
* 6XSSRUW IRU QRQFODVVLFDO ZRUOGV
HF HQW W$ QFHV L XE PE
Q$, 3O QQL
HGJ
6X
G
5H
Q
DWL
8
FH
EE
Q
Z
DWL

S
$
S
QLILH
KDP
DUDR
GYD
OD
KDP
6
E
LQ
.DP
J
.D
$
DUDR

HV
8
LQ
QLIL
5
9LHZ
E
9LH
$,
3
QDQ
FDQ $GY
Planning : The bi g picture
* Synthesizing goal directed behavior
Planning involves
*
– Action selection; Handling causal dependencies
– Action sequencing and handling resource
allocation (aka SCHEDULING)
Depending on the problem, plans can be
*
– action sequences
– or “policies” (action trees, state action mappings
etc.)
HF HQW W$ QFHV L XE PE
Q$, 3O QQL
HGJ
6X
G
5H
Q
DWL
8
FH
EE
Q
Z
DWL

5
$
S
QLILH
KDP
S
GYD
OD
KDP
6
E
LQ
.DP
J
.D
$
DUDR

HV
8
LQ
QLIL
DUDR
9LHZ
E
9LH
$,
3
QDQ
FDQ $GY
Planning & (Classical Plannin g)
(Static)
Environment
(Observable)
Goals
perception
action
(perfect)
(deterministic)
What action next?
I = initial state G = goa l state
(prec) O (effects)
i
[ ] [ ]
I O O O O G
i j k m
HF HQW W$ QFHV L XE PE
Q$, 3O QQL
HGJ
6X
G
5H
Q
DWL
8
FH
EE
Q
Z
DWL

S
$
S
QLILH
KDP
3
GYD
OD
KDP
6
E
LQ
.DP
J
.D
$
DUDR

HV
9LHZ
LQ
QLIL
DUDR
9LH
E
8
$,
5
QDQ
FDQ $GY
Why care about classical Plannin g?
* Many domains are approximatel y classical
– Stabilized environments
* It is possible to handle near classical domains
through replannin g and execution monitorign
* Classical planning techniques often shed li ght on
effective ways of handlin g non classical plannign
worlds
– Currently, most of the efficient techniques for handling
non classical scenarios are still based on
ideas/advances in classical planning
* Classical planning poses man y interesting
computational challen ges
HF HQW W$ QFHV L XE PE
Q$, 3O QQL
HGJ
6X
G
5H
Q
5
8
FH
DWL
Q
Z
S

KDP
$
KDP
QLILH
E
3
GYD
OD
DWL
6
.DP
LQ
.D
9LHZ
DUDR
$
DUDR

HV
8
LQ
QLIL
E
9LH
EE
S
$,
J
QDQ
FDQ $GY
The (too) many brands of classical planners
Planning as Theorem Provin g
Planning as Search
(Green’s planner)
Search in the space of States
(pro gression, regression, MEA)
(STRIPS, PRODIGY, TOPI)
Search in the space of Plans
(total order, partial order,
Search in the space of
protections, MTC)
Task networks (reduction
(Interplan,SNLP,TOCL,
of non primitive tasks)
UCPOP,TWEAK)
(NOAH, NONLIN,
O Plan, SIPE)
Planning as (constraint) Satisfaction
(Graphplan, IPP, STAN, SATPLAN, BLackBOX)
HF HQW W$ QFHV L XE PE
Q$, 3O QQL
HG
8QLI\LQJ
)UDPHZRUNJ
9LHZ
G
5H
Q
DUDR
8
FH
KDP
Q
E
E

EE
$
.D
QLILH
6X
3
GYD
OD
6
9LH
Z
LQ
DWL
J
DWL
$
S

HV
8
LQ
QLIL
S
.DP
KDP
DUDR
$,
5
DQQ
FDQ $GY
Advantages of the Unified View
To the extent possible, this tutorial shuns brand names
and reconstructs important ideas underlying those
brand names in a rational fashion
* Better understanding of existing planners
Normalized comparisons between planners

– Evaluation of trade offs provided by various
design choices
* Design of novel planning algorithms
Hybrid planners using multiple refinements

– Explication of the connections between planning,
CSP, SAT and ILP
HF HQW W$ QFHV L XE PE
Q$, 3O QQL
HG