UNIT-

Afcial nteuigec Planning

he gnteligert waytow thing
INTRODUCTION
atiom
planhing is qired to Lonwent objectives into
Tts th Robot
detision making task betohed by
OY
Comp. dys. athiewe a peeigi Goal.
to

planning th tak ofcoming up wth OSequoncCo

attiond that wn athuwe th 90a
exOrnpl: Goal- Get distintin o tluas ouuu aubyerts
pioning. How many daus tor xam, No,

chaptuc dtud, Notis awailabilüty et

what s AT Planning ?
isa starding aub.aua e A T . t s
Planning Long
inding a phouduAal e o e attin
tho tazk
to &Lach ts goals
a duclanatively duuibed sys.
opimizing ueau peyoimancl measuHLS,
whi
thi ansfomations to
tutomatd plannus tind
Stat oud e thi poscble
apply in 9iven
tath
tsansromationsfor that stat.
Apps) in Sndustn
why is t Inp : planning Ap
Attomation is an emonging kund that Yeguunts
tbuent autonatd plannhit
Many app)t planning i frdusty ( Robts
CutonomO Us dyss), , CoGhruve Cybli seitY,
80wict tomposition .
Aduts). b AL planning Techiqu
whn explaunabilty dutived
whin yAu want to be able to etplain wh
a paticuula cowlse actuon ichosen
Osiqnment uspor0si bilty | blams essential
uutbmation e puseL eq. autonomaUs d
tos
Vng medital ebent ystems
Rapid pototyping shot time to solution
- s h e l plannes auulatbte
yauety o open 0uCl
both TOM opHietauy
even mall
ye pob. is Mguuntly thanging,
nud to t&ange h salution, eny
ehdngis, no

tweok th modu,

when planning ets Deop caning (DL)

6oing cptimizod wbeo. wth tauning hand
but ntegAatung planhing Athngs th htu-
istie uaned y DL wu AUullt suLCes,
quudan
planning (Mont caslo
Go play AtphaG0 ses

tAu eah) with DL (Husistie Guidane) to selut

th nut move.
CognutvNe Assistant viy (Samsung) se knowledga
aph Planurg dL t answel tomplitatd quves
 PLANNING PROBLEM 2

L0gntive NeUAal)
N/W
sstEm
plahning
k n o w l e d g 9 9 -
AL
Roboti
bEnteptuon)

Mc
(NLP) (GLaAnun9

Types o4 planhing
4 Henaswhical plahninq - a pi.ob. is dividd into smalls=
sub piob. is Solued one at a
sub pAoblema, ach

time 's un ustd n Robotes app whet a

out hew to comalut a task
Robot nuds gwu
brEakung t douwn unto Smallts sttps.

9)NOD-Hesaschital plannin a ob, is not dividid

th AI dys.
tnt Amalun ub phoblums, nstead
wiu solwe thi
tes to tind a singt þlan that
ofun Lsedih
Gyamus whai a
evtie bob.
Out
out thi
*hi blut wdy t
Pla
uds to 9ua
comp.
th 9ame,
w that u
Reautve planhing -
ib a ype ot planhing
in Aituations whthi ehuhon. ib ehanging
ed
h AI nuds to be able to rtypond qickly
sys
1t etun ustd in Robotics app whu a Robet
Deids to be able to awoid obstacl o respond=
changs in thi en UiDO,

what planhing PLoblum ?

hoblems that aise uhan an A
PP inAI a
to plan its mOve CouLe
6us. tying to solve
tan be ditilt
ation, Thu? poblems
take into a au th
boz t AT Sus must

O t tomÚ
attions 2choose th On

possible sults.
that wu Lead to th best

Phoblum th hou
th quustien
The Planninq ATat om tho
ntmeve OY 90al
to 9 to
CUUunt stob, The PP i dyied wth
Modul - dyine th actions alon
, Domdu)
wrth th obets.
s
a, Iitual stati - 4ts stat wh any attion

yet to tak platt.

statitot which th plan is itehded
8. Coa
athieve.
COmbon ents Of Planhun9 Sustem
best u l based upon Huunitics
1 choose thu
ubjectits based an
selction qsipl
bY
past peUute
wttuu 0ttChdlud
) Apply this Bul to uatt o nu stat 3
stu dyng tht
subyect, A0 as to achiwe nut
staqg)
3) Detet whin a 80ubON stound
mose
a
(a position is Teached whue hu can &toYe
thah cut- ehmars)
6o that thy can be ayoidlud
4) Dutett diad ends
tind thu oitticut topic wotd them
statu ots P use
&owed
5 Dent when a nasly stafe. ( tn
make it a 8owed
6pial methods to mOks au
than w- o
bosition to get move

Subit ecpt Maths thin aPp

yeeh th 9pal
Plannin9gRepsLentation

tate, Ation P Goal

Consit 8 componends

Actien
OpOkatou Plannin9

1. 8tate > itonunction o Ve teals which (an

contaun uaniabluS invoke e) Ths 4s representd
tatts.
Cond c) thot mus hold butone
2 Action> thtse a

exerdh p th e t oten exLeution

3.Goo 4 is thi most specititied stat1;
q stat is satisty
th qwen goal4 tonsiss al ob e9d. fes th goal.
 Aqunt 4US to Pirdo 0utior e a S¬QMN
ations thot selwes a psoblam
planning Aqent = phoblu 0wing aqent +
Knouwled9e Based Aent
Gennati seure adons to eMtasum tasks
CUnd athieve ebetives)
PAoblm solwing agent > to tonsidu thi Conseqinus
sequarncts uons bejoe ating
Knowledg BOsed Agent> tn stlu detions basd
scþo¬LenHat thu outvurd
th
uplit logial
8tadtuP h ts ations.

PLANN)NGLANGUAGES
PL should be erbhusive P ewey planning long
makis thu epresntad Schama Ao tht tho

alap
alap. tan beopeatud on t
3tandand Rueach snsiutk psoblem
S0We 5TRIPS
a Acuo QuUiption oung ADL)
3. plannng Dormain OLLCpton lang, PDDL)
() STRIPS
deueloped in
4970s t Statovd to thu tSt+
inteuligent nobot
i t tan dusuibe th wod ( 3nitial stat, Goal
Statu) by bavviding objun, Actions precond eHect.
A A STRIPS istance s Composed 11) Stalo
2) o al 3) Action
tenals which cant
Stat ts conTuntion o ve

Contaib vauabls iuoke

(Home) ^ Haue (Banana2)
tenals may tondain-
Goal - these ConTunction e
vaiablus. e Haue (Banana)
At(Home)^
AtX) A Seus x, Banana)
Action- thuse a phsconcd) that rmust hold behore
th excutio).
extution P t s at
Action fy CP» aom, to)
PRE COND: t (p, fom) A plane(p) A Airpot (fum)
Aipot (to)
EFFECT: A CP. rorn) A At (p, to))

ADL (Aduion Destpt LOng.

10ng.
C
mitations oSTRIPS
Ustd t ovextome th

move epehsive than $TRIPS

pOpes) 1. t allouwe-ve eMals

4uanti-fied vanbles wYth th disunction
2. 34 use
th LonTuhcuon,

1POOA fomous V Smatt

. Cond'al post tord't au aewed
time asu
vasiableu wth lit. pes a h a m
allouwed also equatrty p is auailabl
 daving eas with STRIPS
Considu cah

Ation (obive (c,Yuom, to))

puLcond a (t, toomA Canlc)

post tond: at (C, tom) A at (C, 4o)

tonod means remov th
Tn 9tsups, post
'

cond add thu 2r cond

with ADL, ame
action is tbresentid au tallos

Action(dive (c, tsom, to))

at (C, tom) Atanlc) a tom tu)
pxe ond;
pest tond: at CCstom) dt(c to)
^

(3) PDDL 4 td ehtoding Lang. to 'dassial

plannig task.
STRIPS ADL
SupetsG of
4Componints of PDDL planhing task
in thu world thal inteust Us
1) Obect thingg
>
au uteraste
tha4 uwe
a)pvedicates pöopt. eebja)
t COn be false.
&tati Hatu oth woridthat we &tat 0,
Snitiay
tht we wart to be u
4 Goal Speuittation things
stati
Acions opehadons ways ehangin the
th world,
 plannin9 qsks specified in PODL au &tpanoded 5
nto ils
1. Demun file poeditates P outtions
phobley ile to obyeets; initial t t fgoal
9oa
specifieation.
Domain lue looks ke
deine (domai domun namu)
PDDL tode o pacditati
PDDL tLst ttion
PDDL Last

PsLoblem tlesLook ike

deine Päoblem Poblem name
( domain domu-NAMA >
PODL tocu obes
PDDL niial 9tatu>
PDOL Goa

BLOCKSwORLD
also knoun as Susmar Anomaty
bloeks labelid al A '8 cau
n thiu pouob.,
3 3
Ausal. The gwen
allowec to ut on th lat
one block tan be moved at a
cond" is that tnl
athieue the goal
time to
 t's conist otallouwing
wrth urniqw uttes On
table 2) clential bldcks them
onu to Onotheh stack tom
Blocks au put
built with a Aobot aum: The àsm tan
) 8tack s
time
beutoum p' gtung
o
dinglu blok at a
placing it atatus n
piudiat used t duouibe
ampll=

block world aU
A is on
on th table
1. On (A, table); Block

2. On (B, tabl)
3.
3 On B, A): Block is on Bloek AA

4. Cleah (A): Bloek A has notkunq o n t

5. cleanC 6)
6. Hetcling (B) Robot a hoding
Empty Axm : om not holding aything
RepLLentatuorn -

Jnitial &tat n (A, tabli)^ On (8, tablu) a cLas CA)A

cleant) Empty A

On (A , 1abu) A n (B, A) A Ckar (B) A EMpty-

GDal 8tah;
Used to dusoibe tatts n Block
action lop)
wortd aLA
bleck 8 om
1. UNSTACK (B,A); to t t A

22. Ghack (B,A) To place block 8 on

On A
A
3. ugt ) : to ift th block 8 mhr tabl
 H. block B on the fable. 6
place (B): to put th
s chosen bucausI
The locks wOld
weu. behawed
3t's uHhiciently aimbl f
easily nclwstood
enuiwn. -to Study plan
otovids a qood ample
broken nto neat
pobluhs ean he
ub poblems
distinct
GOAL STACK PLANNING hich
in AL in
eaiest methods
one th qoaltati to
thi
we woyKbackwards LOm
tht nitial stat
fo plan genciaticno.
abproath Ses a Stack
thus acudns dus
an contun Aub-bal f.
Th stack
redicates,
oubed using Solwed che by one
One in
In
tan be
thu
uhgoals asy ovdiA.

we stat at th Gbal statt p We toy uiung

Aegd. t achieue th inithal jat1
th pu- tondc
'goals Sub-goals untid
we kup &oluing the
atuve th oticdMade.
nally eoom 4 op'u)
Robot Am can
The
plock X n Block
STACK (X,Y): stackirg
up Block X Awhich
which y Or- s Or
UNSTACK (x,Y): pickirq
of Block
XPICKUP (x) picking up Biouk x which is on t e
th table,
Block X n h tobl
tabla
PUTDOWN (X): pu

Opeyatiorn
Pre-condition
Action
1. PICK UP Cx) ARM-EMPTY
A DN (x, table) Holdirg (x)
AClean CX)
ArOT9-Empty A OnlX
2. PUTDOWN (X) Holding (X)
tablu) ClEan CX)
OnCXY)A clai(x)»
3. STACK (x,Y) Holding (x)^ckan(Y)
Arom Emp
4UNSTACK (X,Y) tn(X,Y) Ueay (X)^ Holding (x)
Avm-Em CieanY)

nitial 8tat A fe
Goou 8Tot A

6oLuton ON A,B)
Oul gool
Stp1 no. 33
no. 30,
0,
explanaticn goo is achievecd by sp
ShOck A, B)
Stup
9ttp3 Holding (A) ^ (B) psecOnd
cLlean
ptecond
-

not ue,
hue,
explaatien t stap 3, Holding (A)
when we se itial stat, 43s expand by
o n0, 1
Sep 4 PICKUP (A)
Tep5Asm êmpty ^ ON (A, table) Uian CA)-poetond
TRUE
b Tsue,&0, ifs also
NOte Hence, he tep 5
thi stup 3,
than ( A,8)
step 2 "StackTRUE
5 ue
S0,
So, whe Step 3 P
we Can aehieue goa
ON(A,8)

pLcond dh au wndudlned wth Rid& actions

aUu mattkid

MEAN,END ANALYSIS t goal cstabli_-

it 8owes obus). by duining
shing thu sight adion plh, Thus tehniquu is
wed in AI psog u). to imt &each,.
Atyat1gauS to
cOmbines tonuand Backuand
sowe COmplu ÞAoblums.
this. tchhigul, sys.
evaluatu thi dit. blwv
T
stati. +
cuunt 0,0
statpositicn &d tahget /9oal
best action to be undstakan
then d euduy
to Yeach the end Qool.
MEA wOks
How
kye. evaluatus tht unt stati to esiabI
4) fit,
sh whtho th pab. 3 aoh8 1jdentifise
means that an action houwd takan
theh
wMictit.
to
290tord sep dyining thu tagt os dusired
UVolwes
that netds to be dchieved,
9001
put t b sub goals, thad ao
3)Tangt 90a
tult split intD Othil &mallt g0als,
4) it invowes establishing th actions op'c
casied ot t ochuewe tho encd
thad wW be Stati
Unkid uwrth
all th dub-90als asu
54 this
Co sbondi rg exeutablattions (o)
intemediat1 stups a0 urdui-
6Atu that s dene,
thr u u d st
sakin solue thi Soebe. in
udur
chosen opehatastu ww be applied
tht cuwut statt 4 end&tat.
the ditt blo
ratking
Uvolves all thi chahqs med to th
actual Atoti changs au Made entiil h toqct
8tati achuewed

End sub &ub sub

Acions
Goolu 90a goals
Goal ActionsS
9al Acti ons

Ahctyons
Soal
Goal Actions
 8
Algo, dt fo MEA
th Atatus o the
th
QssesM
Conduet a study to
macto 0
be done at a
Cwunt stat, he can

MDluwel.
cwvrd Atat &dyuno
2, Captwu th obu), in th
hu
maCn
can also be done
at a
th aHq stat. This
mitd Awe.
stati end
8, Makea combaison blo th Cunt
thu
thue Atats a
&tat that ou olluned,
action. This i ar-
80me, then peuom no wthu
ib dicatton that thu ob, bas been*atklid. +tho
twDtatis a not thu same, thun maue to stap .
Aatts a th tup
. Rcord th dithountS bjw w
aoremehtioned Jevels maero . mico
unto odjustmen to
5, Tsansfom thuse difunas
hu Cuud Atati,
the
6. beeunine th ight action tor únplementing
oduttoents in tep 5.
thi USult witH
.
èiut tht hanges f corma
th tagt 9pal
th 0sh stiRBome ditf. bluw wvunt stat
8.
P t1q &tatu, pertom ouse OTetior n t i th
end goal is athieued.
nitad Atati
 th COept o MEA to establish whthes
Apply
theu any adustments nudid. The Phct step
is t evaludt th initial &tati compau4
wth tht tnd 9wal to eutablish whtthu thu a
b/w h tuv tates,
any di

dnatial tat GOOUstato

that s a diy blo Cusd
Dw uwwt
Abeve imagushows tthlu
that t h u is
inditatt
th0
t h taqt stat, his
uNunt state th
adyustments to
nld to makt
Yeach
the ehd goal.
to ub- tbot arn
qoalL that a
be divicld inty
900l Cah
ThE
The
aetions oaps
Lunkidwith exuuut able R
b SCO
that tah
tho 3 obehatos
The olo. au

sotwe thu bLablum.

1, Delte operator - symbol at th top Aight

dot
n }oal
COAnu in th inial stat doess+ exist
yemeved by applyir
stati. Tfu dot ymbol Can be
dult e y ator
the

Delt

dnitrl &tta
2 MovNC operatoY NOU, tompau thi
nUAtau
w}h g
i th nuu sat1 is
thu ehd tti, The diamond
cirde whil in the end 8tat, t's a the
insidu h tto
move this diamond symbol
tep coihU,
ight
tht moU opesator,
th sig positon by plying
Delute Move
> )
Tnitial Hatr

3Expand OpeAafol aytii yaluatung thi nw 6tat1

thott ths diamond symbol
gentuated stup 2, od
than th On i th ehd datu, tWean
&mallls
th Aize tht mbol by pblying euoUnd
penatO.

Dellt Move
Evpard

Inin statu
thtue twD dtatu
Thesu au no dilf blw
&olued
etich means brob,
Application 04 MEA MEN
OYgorhizational planrung >,ued in orq'U to tari-
tKelps orqa mahagvu to
tat qtnehal mqmt.
Conduct blanning to Qthieue thY abectivt s th
ovg. The mgmt eathis th dusnred gtal y divndng
thi main 0als inb ub- 90als tho nkid wHh
anonabltask, 0
this techniqui s
busines ranstoYmation hers
Used to opltment tsansfoomat brojet, 4
a
i ths urreit Atati
a a oluived changu
iappled to eatabliuh
th
bus otet, MEA
be inplumettid. TF pvUse a u
po0CUses to enhan e uue
Anto Aub
aplit
implmentation.
blw tho
tho
th DMpasuSGn
* ap
G Analysis>i
betomand,
besomone th red,
C Lnt tompcuU
thr
eld to
this
abpled An
MAEA
the dised tech. In ogy.
existi ng techhol094
to istiNg
t h eistiNg
aplied
Vauou epe au

techholog
MACHINE LEARNING
mstiain
S ab»anch
a AL which eably
t-
data ôy expewencLs twthout being
suom pat
piuty brDgrammed.
budictions
a w0mb. , to mak
enablls
M
ML histoaieal data wth
Some ditisions6 tuing
take barammd

out bei upleitey

Jsum histomal dan,
Juans
ML POM
ML d.
buulds th bytdiction modls whunaue it reCtives