UNIT -III

CHAPTER
INTRODUCTION TO
PLANNING 6
O What is Planning?
3 Types of Planning
O The Blocks World Problem
OComponents of Planning System
What is Uncertainty in AI?
3 Reasons for Uncertainty in Artificial Intelligence
Nonmonotonic Logics
3 Probabilistic Reasoning in Al -Away todeal with Uncertainty
3 What is Bayes Theorem in AI?
3 Challenges to probabilistic approaches:
3 Fuzzy Logic
S Architecture

Membership function :
Advantages of Fuzzy LogicSystem
Disadvantages of Fuzzy Logic Systems
5 Application
Review Questions
6.2 Artificial Intelligence

6.1 What is Planning?

Planning in the context of artificial intelligence (Al) refers to the process of determining a sequence of
actions or decisions to achieve a specific goal or solve a problem. It involves generating a plan, which
is a structured representation of the intended course of action, considering the initial state, desired
goal, available resources, and constraints:
1. Initial State:
The initial state defines the starting conditions or the current state of the world or system.
> It includes information about the environment, objects, their properties, and their
relationships.
The initial state serves as the foundation for the planning process, determining the
context from which the plan willbe developed.
2. Goal:
The goal specifies the desired state or the outcome that the planning system aims to
achieve.
It represents the condition or set of conditions that the plan should satisfy or accomplish.
º The goal provides the direction and purpose for the planning process, guiding the
selection of actions and decisions.
3. Actions:
> Actions represent the executable steps or operations that can be taken to modify the
state of the system.
Each action typically has preconditions that must be satisfied for the action to be
applicable or executable.
Actions also have effects, which describe the changes they induce on the state of the
system once executed.
The set of available actions and their properties define the action space or the range o'
possible operations in the planning process.
4. Transition Model:
The transition model defines the dynamics of the system, specifying how the state
changes when an action is applied.
> It describes the relationship between the current state, the selected action, andthe

resulting new state.

The transition model is used to simulate the Consequences of actions and predictthe
resulting state in the planning process.
5. Search Algorithm:
The search algorithm determines the strategy for exploring the space of i possible plansor
sequences of actions.
while
> It systematically explores the search space to find a plan that satisfies the goal
considering constraints and optimizing certain criteria (e.g. plan length, cost).
 Introducion to Planning 6.3
Common search
search, A*
algorithms used in
search, and heuristic search.planning include depth-first search,
6. Plan Representation: breadth-first
Anlan is the output of the
executed. planningsystem and represents thesequence of actions to be
> Plans can be represented in various
graphs. formats, such as action sequences, decision
trees, or
The plan
representation should be structured, unambiguous, and
execution phase. executable to guide the
7, Plan Execution and Monitoring:
> Oncea plan is
generated, it needs to be executed in the real or
> During execution, the simulated environment.
actual state of the system is
expected states described in the plan. monitored and compared to the
> Deviations or
unexpected events may require plan adaptation or replanning to
dynamic environments. handle
Planning systems find applications in diverse domains,
including robotics, logistics, scheduling,
resource allocation, and autonomous systems. They enable Alsystems to autonomously reason
about actions, make decisions, and generate effective
and uncertain environments. plans to achieve desired goals in complex

6.1.1 Types of Planning

we have Forward State Space Planning (FSSP) and
level. Backward State Space Planning (BSSP) at the basic

Types of Planning in AI

Forward State Backward State

Space Planning (FSSP) Space Planning (BSSP)

Forward State Space Planning (FSSP)

FSSP behaves in the same way as forwarding state-space search. It saysthat given an initial
state Sin any domain, we perform some necessary actions and obtain a new state S' (which
also contains some new terms), called a progression. It continues until we reach the target
Position. Action should be taken in this matter.
Disadvantage: :Large branching factor
Advantage: The algorithm is Sound
6.4 Artificial Intelligence

" Backward State Space Planning (BSSP)

BSSP behaves in asimilar fashion like backward state space search. In this, we move
from
the goal state gtowards sub-goalg' that is finding theprevious action tobe done to achi
that respective goal. This process is called regression (moving back to the previous g0al s
sub-goal). These sub-goals have to be checked for consistency as wel. The actions have to b.
relevant in this case.

Disadvantage: Not asound algorithm (sometimes inconsistency can be found)

Advantage: Small branching factor (very small compared to FSSP)

6.2 The Blocks World Problem

One of the most famous planning domains is known as the blocks world. This domain consists of a set
of cube-shaped blocks sitting on a table.The blocks can be stacked, but only one block can fit
directy
on top of another. A robot arm can pick up ablock and move it to another position, either on the
table
or on topof another block. The arm can pick up only one block at a time, so it cannot
pick up a block
that has another one on it The goal will always be to build one or more stacks of
blocks, specified in
terms of what blocks are on top of what other blocks. For example, a. goal might be to
Band block B on C.
get block Aon

B A C

Start State
Goal State

The actions it can perform include

UNSTACK (A,B)
remove block A from block B
the arm must be empty
> block Amust have no blocks on top of it.
" STACK (A,B)
put block A from block B
º the arm must already be holding A
the surface of Bmust be clear
PICKUP(A)
º pick up block A from the table
> the arm must be empty and there must be nothing on top of A
 Introduction to Planning 6.5

. PUTDOWN(A)
the table.
> put block A on
blockA
> the arm must have holding
used: Ontable(A) Block A is on the table
Predicates canbe
On(A,B) Block A is on Block B
Clear(A) nothing stands on A
Armempty robot hand is empty
Holding(A) robot hand holds A

Logical notationscan be used:

1. [Bx: HOLDING(X) ]> ARMEMPTY
If the arm is holding anything, then it is not empty.
2. Vx:ONTABLE(x) -> y:0N(x,y)
If a block is on the table, then it is not also on another block.
3. Vx:[By :ON(yx) ]> CLEAR(X)
Any block with no blocks on it isclear
Init (Ontable (A) AOntable (B) AOn(C,A)
C
Goal (On(A,B) A On(B,C))
B A

Action
(a) Move block C on the Table
"Holding( C) C
B A
"Ontable ((C)
"Arm empty
"Clear (A)
(b) Move block Bon block C
"Ontable (B)
B

Holding (B)
. On (B,C) A C

" Armempty
(c) Move block Aon
block B A
"Ontable (A)
" Holding(A) B
" On
(A, B) C
" Armempty
6.6 Artificial Intelligence

6.3 Components of Planning System

Inproblem solving systems, it is necessary to perform each of the following functions:
Choose the best rule for applying the next rule based on the best available heuristics
Apply the chosen rule for computing the new problem state.
" Detect when a solution has been found.
" Detect dead ends so that they can be
abandoned and the system's effort is directed in mor.
fruitful directions.
Detect when an almost correct solution has been found.
Choose the best rule for applying the next rule based on the best available
heuristics.
" Finding the difference between the current and goal
states.
Choose the best rules that reduce these differences most effectively.
Example- Means - End analysis
Wishing to travel by car to visit a friend
+ First -> fill up the car with fuel.
+ If no car ->acquire one.
" Largest difference must be tackled first.
Apply the chosen rule for computing the new
problem state.
" In simple systems, applying rules is easy.
Each rule simply specified the problem state that
would result from its applications.
In complex systems, we must be able to deal
with rules that specify only a small part of the
complete problem state.
. One way is to describe, for each
action, each of the changes it makes to the state
" Anumber of approaches have been used for this description.
task.
Green's Approach
Basically, a given state was described by a set of
true in that state. Each state was predicatesrepresenting the facts that were
represented explicitly as part of the predicate.
Green' Approach for Simple block world description
A
B
ON(A, B,S) ^ON TABLE(B, S,) ^CLEAR (A, S,)
Thisfigure show the state called S, ofa simple blocks
world problems could be represented.
. If wanted to UNSTACK(A,B), the operation is expressed as :
CLEAR(X, S) ^ON(X, y, S)] -’
[HOLDING(x, Do(UNSTACK(X, y), S)) ^
CLEAR(y, Do(UNSTACK(x, y), S))]
 Introduction to Planning 6.7
blocks
> X,y- any
- any state

Do)- specifies that a new state result from the given action
The result ofapplying this to state S, to give S, by applying UNSTACK is
HOLDING(A, S1) ^ CLEAR(B, S1)
Oneproblem with this approach is, Bis still on the table.
This needs to be encoded into frame axioms that describe
affected by the operator. components of the state not
So, it must be,

ONTABLE(Z, S) ’ ONTABLE(Z, Do(UNSTACK(X,y), s)

. If want to color the blocks, an axiom used is,
COLOR(x, c, s) ’ COLOR(X, c, Do(UNSTACK(Y, z), s)
STRIPS Approach
STRIPS ’ Stanford Research Institute Planning System
" Amechanism that doesn't require a large number of explicit frame axiom
" Basically each operator has 3 lists of predicates associated with it.
" Each operation is described by a,
> List of new predicates that become TRUE called ADD
> List of old predicates that becomne FALSE called DELETE
>List contains those predicates that must be true for the operator to be applied called as
PRECONDITION
> Anything not in these lists is assumed to be unaffected by the operation.
SIRIPS style operators that correspondence to the block world operations are :
STACK(X,y)
P:CLEAR(Y) AHOLDING(X) (precondition)
D:CLEAR(Y) AHOLDING(X) (delete)
A:ARMEMPTY AON(x,y) (add)
UNSTACK(x,y)
P ON(x,y) ACLEAR(X) AARMEMPTY
D: ON(x,y) A (precondition)
ARMEMPTY (delete)
A:HOLDING(X) A
CLEAR(Y) (add)
PICKUP(X)
P:D:ONTABLE(x)
CLEAR(X) ONTABLE(X)
A AARMEMPTY (precondition)
AARMEMPTY (delete)
A: HOLDING(X) (add)
6.8 Artificial Intelligence

PUTDOWN()
P: HOLDING(X) (precondition)
D: HOLDING(X) (delete)
A:ONTABLE(x) AARMEMPTY (add)

A A

B B B

A B C A C C C

On(A, Table) On(A, Table) On(A,B) On[AB)

On(B, Table) On(B,C) On(B,C) On(B,C)
On( C), Table) On( C), Table) On( C), Table)
Clear(A) Clear(A) Clear(A)
Clear(B) Clear(B) Clear(Table)
Clear(C) Clear(Table)

Move(B, Table, C) Move(A, Table, B)

S. S, S,
OK!

Detect when a solution has been found.

" Aplanning system willsucceed in finding a solution to a problem only when it has found d
sequence of operators that transforms the initial problem state into the goal state. But how could
a system willknow it has found the solution. This could be done by comparing the problem s
goal state with the current state if it matches it we could straightaway say that a solution state
has been reached. But on the other hand if the entire states are not represented explicitlybut
rather described in term of some properties then detecting a solution becomes more conp
This could be solved depending on the way the state descriptions are represented.
" For any representational scheme that is used it must be possible to reason with representation
reasoning
to discover whether one matches the other or not. Then the corresponding
mechanism could be used to discover when a solution has been found.
representation
Predicate Logic serves as the basis for many of the planning system as a
technique. Due to its deductive mechanism the researchers often find it more appealing
Detecting Dead Ends :
sequence
APlanning system might sometime could land into dead-ends when searching for a usedfor
of operators to solve a particular problem. The same reasoning mechanism that is
reasoning can also be used for detecting the dead ends.
 Introduction to Planning 6.9

eoarch process is reasoning torward from the initial state, it can

prune any paths that
leads to a state from which the goal state cannot be reached
" Ifthe search
process is reasoning backwards from the goal
or hecause it is sure that the initial state cannot be
state it can also terminate a path
reached or because little progress is
being made.
n reasoning backward each goals is decomposed into
subgoals where each of them in turn
ay lead to a set of additional subgoals. Sometime it may be
easy to
way that all the sub-goals in a given set can be satisfied at once. or identify that there is no
example the robot arm
cannot be both empty and holding a block. If any path that isattempting to make both of
those
goals true can be simultaneously pruned immediately.
Detect when an almost correct solution has been found
While solving a nearly decomposable problem the sub solutions when combined may not
acomplete solution. So when such situation arises there are yield
certain approaches to handle it.
The simplest is just to throw out the solution, look for another one, and hope that is is
better.
This approach looks simple but it may lead to a great deal of wasted effort.
" A slightly better approach is to compared the desired solution and derived solution and if
there is a difference then the problem solving system can be called again and asked to
find a
way of eliminating this new difference. The first solution could be combined with the second
one to form a solution to the original problem.
" An even better way to patch up an almost correct solution is to appeal to specific
about what went wrong and then apply a direct patch. knowledge
" Astill better way to patch up incomplete solutions is not really to patch them up at but rather
to leave them incompletely specified until the last possible moment. Then when as
much
Intormation as possible is available, complete the specification in such a way that no conflict
arise. This approach is called as Least Commitment Strategy.

6.4 What is
"
Uncertainty in AI?
When we talkabout
perceiving information fromthe environment,then the main problem that
aises Is that there is always some uncertainty is our observations. This is because the world is
an
So, enormous entity and the surroundings that we take under study is not always well defined.
there needs to be an estimation taken for getting to any conclusion.
Human being face this uncertainty daily that too many times. But still, they manage to take
successful decisions. This is because humans have strong estimating and decision making
power and their brains function in such a way that every time such asituation arises, the
alternative with the maximum positive output is chosen. But, the artificial agents are not
able to take proper decisions while working in such an environment. This is because, if the
information available to them is not accurate, then they
their knowledge base. cannotchoose the right decision from
6.10 Artificial Intelligence

Uncertainty Example
Taking a real life example, while buying vegetables, humans can easily distinguish
different kinds of vegetables by their color, sizes, textures, etc. But there is uncertainty inbetm
ween thethe
right choices here, because the vegetables may not be exactly the same as described. Some ofe making
may be distorted, some may vary than the usual size and there can be many such variations.
spite of allof them,humans do not face any problem in situations like these. Butin
But, this same thing becomes a hurdle when the decision is to be made by a computer based aoe.
Because, we can feed the agent by the information that how the vegetables look, but still we can.
accurately define the exact shape and size of each of them, because all of them have some variation.
So, as a solution to this, only the basic information is provided to the agent. Based on this yen,
information,the agent has to make certain estimates to find out which vegetable is kept in front of t
Not only in this agent, but in designing almost every Al based agent, this strategy is followed. So, ther:
should be a proper method so that the agent can make certain estimations by itself without any heln
or input from human beings.

6.4.1 Reasons for Uncertainty in Artificial Intelligence

The following are the reasons for uncertainty in Artificial Intelligence:
1. Partially Observable Environment : The entire environment is not always in reach of the
agent. There are some parts of the environment which are out of the reach of the agent and
hence they are left unobserved. So, the decisions that the agent makes do not include the
information from these areas and hence, the result drawn may vary form the actual case.
2. Dynamic Environment: As we all know that the environment is dynamic, i.e. there are always
some changes that keep taking place in the environment. So, the decision or calculations made
at any instant may not be the same after some time due to the changes that have occurred in
the surroundings by that time. So, if the observations made at any instance are considered
later, then there can be an ambiguity in the decision mking.
3. Incomplete Knowledge of the Agent: If the agent has incomplete knowledge or insufficent
knowledge about anything, then it cannot produce correct results because the agent itselt doe
not know about the situation and the way in which the situation is tobe handled.
ar
4. Inaccessible Areas in the Environment : There are areas in the environment which
made
observable, but not in reach of the agent to access. In such situations. The observation wil
parts
is correct, but the as an agent cannot act on these parts of the environment, these P can
but
remain unchanged by the actions of the agent. This will not affect the current decision
affect the estimations made by the agent in the future.

6.5 Nonmonotonic Logics

when
the truthfulness of a conclusion does not change
" Areasoning system is monotonic if monotonically grow"
new information is added to the system the set of theorem can only the
set

when new axioms are added. In contrast, in a system doing non-monotonic reasoning
of conclusions may either grow or shrink when new information is obtained.
 Introduction to Planning 6.11

Nonmonotonicclogicssaare used to formalize plausible reasoning, such as the following inference

"
step:
Birds typically fly.
Tweety is a bird.

Tweety (presumably) flies.

Such reasoning is characteristicof commonsense reasoning, where default rules are applied
when case-specific information is not available.
The conclusion of nonmonotonic argument may turn out to be wrong. For example, if Tweety
is a penguin, it is incorrect to conclude that Tweety flies. Nonmonotonic reasoning often
requires jumping to a conclusion and subsequently retracting that conclusion as further
information becomes available.
" All systems of nonmonotonic reasoning are concerned with the issue of consistency.
Inconsistency is resolved by removing the relevant conclusion(s) derived previously by default
rules. Simply speaking, the truth value of propositions in a nonmonotonic logic can be classified
into the following types:
1. facts that are definitely true, such as "Tweety is a bird"
2. default rules that are normally true, such as "Birds fly"
3. tentative conclusions that are presumably true, such as "Tweety flies"
When an inconsistency is recognized, only the truth value of the last type is changed.
Arelated issue is belief revision. Revising a knowledge base often follows the principle of
minimal change: one conserves as much information as possible.
" One approach towards this problem is truth maintenance system, in which a "justification"
Tor each proposition isstored, so that when some propositions are rejected, some others may
need to be removed, too.
Major problems in these approaches:
conflicts in defaults
COmputational expense: to maintain the consistency in a huge knowledge base is hard, if
not impossible
Advantages of Non-monotonic Reasoning
1. For real-world
systems such as Robot navigation, we canuse non-monotonic reasoning.
2.. In Non-monotonic reasoning, we can choose probabilistic facts or can make assumptions.
i
1. In Disadvantages of of Non-monotonic Reasoning
2. It non-monotonic
cannot be
reasoning, the old facts may be invalidated by
used for theorem proving.
adding new sentences.
6.12 Artificial Intelligence

6.6 Probabilistic Reasoning in AI - A way to deal with Uncertainty

" we knowthat there are many cases where the answer to the problem is neither
true nor completely false. For example, the statement- "Student will pass in the board eva completely
We cannot say anything about a student's result before the results are declared. Howeyer s
can draw some predictions based on the student's past performances in academics.
" In these types of situations, probabilistic theory can help us give an estimate of how much an
event is likely to occur or happen? In this theory, we find the probabilities of all the alternative.
that are possible in any experiment. The sum of allthese probabilities for an experiment is
always because all these events/alternatives can happen only within this experiment.
Basic idea: to use probability theory to represent and process uncertainty. In probabilistic
reasoning, the truth value of a proposition is extended from {0, 1} to [0, 1], with binary logic as
its special case.
" Justification: though noconclusion is absolutely true, the one with the highest probability is
preferred. Under certain assumptions, probability theory gives the optimum solutions.
To extend the basic Boolean connectives to probability functions:
negation: P(-A) = 1 - P(A)
> conjunction: P(AAB) = P(A) * P(B) if Aand Bare independent of each other
> disjunction: P(AVB) = P(A) +P(B) if Aand Bnever happen at the same time
Furthermore, the conditional probability of Bgiven Ais P(B|A) = P(BAA) / P(A), from which
Bayes' Theorem is derived, and it is often used to update a system's belief according to new
information: P(H|E)=P(E|H) *P(H) / P(E).

6.6.1 What is Bayes Theorem in AI?

Baves theorem is a method to find the probability of an event whose occurrence is dependent on
some other event's occurrence. In simple words,using the Bayes theorem, we can find the conditional
probability of any event. Bayes theorem, given by Reverend Thomas Bayes and thus named after him
Bayes Theorem Mathematical Equation
The Bayes Theorem, also known as Bayes law or Bayes equation is amathematical equation which 5
given as follows:
P(B|A) P(A)
P(B) 0.
P(B) Where, A and B are events and P(B)
Here,
Bhas already occurred
" P(A|B): Conditional probability of occurrence of event Awhen event
Ahas already occurred
" P(B|A): Conditional probability of occurrence of event Bwhen event
other events.
" P(A): Probability of occurrence of event Aalone without any dependence on
events.
" P(B): Probability of occurrence of event Balone without any dependence on other
 Introduction to Planning 6.13

Theorem
DerivationofBayes
P(A ^B)
P(B)-= P(B)

Similarly
P(A ^B)
PB|A)= P(A ^B) = P (B|A) P (B)
P(B)

Puttingthe value of P(A^B) in equation (1), we get

P(B|A) P(A)
P(B)=
P(B)
Which is our required Bayes equation.
It should be noted that in the Bayesian equation,we need not find the probability of both the events
occurring simultaneously, i.e. P(A^B). We can simply calculate the conditional probability of an
event if we know the conditional probability of the event on which it is dependent and the individual
probabilities of both the events without any dependency on each other.
Baves Theorem is applicable only in those experiments where we have only two events. It is not
applicable to the cases where the number of events is more than two.

6.6.2 Chalenges to probabilistic approaches:

" unknown probability values
" inconsistent probability assignments
computational expense
6.7 Fuzzy Logic
Fuzzy logic is a generalization of classical logic, and reflects the impression ofhuman language
and reasoning, Examples of fuzzyconcepts: "young", "furniture", "most" "cloudy". and so on.
hcCording to fuzzy logic, whether an instance belongs to a concept is usually not a matter of
"yes/no", but a matter of degree. Fuzzy logic uses a degree of membership, which is a real
number in [0, 1]. Example :
+ Yes/1
Boolean
Is it hot water? Logic

+No/0

+ Very much/0.9
Fuzzy
Is it hot water? +Little/0.25
Logic
+ Very less/0. 1
6.14 Artificial Intelligence

" In summary, Fuzzy Logiccis

i a mathematical method for representing vagueness and
uncertainty
in decision-making, it allows for partial truths, and it is used in awide range of applicatione
is based on the concept of membership function and theimplementation is done
rules. using Fuzy
6.7.1 Architecture
Its Architecture contains four parts:

Rule Base

Crisp Fuzzification Defuzzification Crisp

Rules
input Module Module outpue

Inference
Fuzzy Enginee Fuzzy
Input Output

" RULE BASE: It contains the set of rules and the IF-THEN conditions provided by the experts to
govern the decision-making system, on the basis oflinguistic information. Recent developments
in fuzzy theory offer several effective methods for the design and tuning of fuzzy controllers.
Most of these developments reduce the number of fuzzy rules.
FUZZIFICATION: It is used to convert inputs i.e. crisp numbers into fuzzy sets. Crisp inputs
are basically the exact inputs measured by sensors and passed into the control system tor
processing, such as temperature, pressure, rpm's, etc.
" INFERENCE ENGINE: It determines the matching degree of the current fuzzy input with
respect to each rule and decides which rules are to be fired according to the input field. Next,
thefired rules are combined to form the control actions.
" DEFUZZIFICATION: It is used to convert the fuzzy sets obtained by the inference engine
into
one is
a crisp value. There are several defuzzification methods available and the best-suited
used with a specific expert system to reduce the error.

6.7.2 Membership function :

The membership function isa function which represents the graph of fuzzy sets, and allows us
to quantify the linguisticterm. It is agraph which is used tor mapping each element of x to the valu
between 0and 1. There can be multiple membership tunctions applicable to fuzzify anumerk
value. Simple membership functions are used as the complex functions do not add precision in
output. The membership functions for LP. MP, S, MN, and LN are:
 Introduction to Planning 6.15

Membership
Function
LN MN
1 MP LP

0.5

Input Voltage
-10 -5 +5 +10

The triangular membership function shapes are most common

among various other membership
function shapes. Here, the input to 5-level fuzzifier varies from -10 volts to +10 volts. Hence the
corresponding output also changes.

6.7.3 Advantages of Fuzzy Logic System

Advantages of Fuzzy Logic System

1. This system can work with any type of inputs

whether it is imprecise,
distorted or noisy input
information.
2. The construction of Fuzzy Logic Systems is easy and
understandable.
3. Fuzzy logiccomes with mathematical concepts of set theory and the
reasoning of that is quite simple.
T t proVides a very efficient solution to complex problems in all fields of life as it
resembles human
reasoning and decision-making.
J. The algorithms can be described with little
data, so little memory is required.

6.7.4
Disadvantages of Fuzzy Logic Systems
Disadvantages of Fuzzy Logic Systems
1. Many
researchers proposed different ways to solve a given problem through fuzzy logic
which leads to ambiguity. There is no systematic approach to solve agiven problemthrough
2.
fuzzy logic.
Proof of its characteristicsis difficult or impossible in most cases because every time we do
not get a
3. As mathematical description of our approach.
fuzzy logic works on precise as well as imprecise data so most of the time accuracy is
Compromised.
6.16 Artificial Intelligence

6.7.5 Application
spacecraft and satellites.
" It isused in the aerospace field for altitude control of
control, traffic control.
" It has been used in the automotive system for speed
the large combany
It is used for decision-making support systems and personal evaluation in
business.
the pH, drying, chemical distillation
" It has application in the chemical industry for controlling
process.
various intensive applications in
Fuzzy logic is used in Natural language processing and
Artificial Intelligence.
systems such as expert systems.
" Fuzzy logic is extensively used in modern control
mimics how a person wouldmake decisions
" Fuzzy Logic is used with Neural Networks as it
changing it into more meaningful data
only much faster. It is done by Aggregation of data and
by forming partial truths as Fuzzy sets.
Summary
of determining a
Planning in the context of artificial intelligence (AI) refers to the process
problem.
sequence of actionsor decisions to achieve a specific goal or solve a
Goal, Actions, Transition Model, Search
" Different stages of planning are : Initial State,
Algorithm, Plan Representation &Plan Execution and Monitoring.
Backward State Space
Types of planning:Forward State Space Planning (FSSP) &
Planning (BSSP)
" Forward State Space Planning (FSSP) : FSSP behaves in the same way as forwarding
stale"
necessaly
space search. It says that given an initial state S in any domain, we perform some
actions and obtain a new state S(which also contains some new terms), called a progress
Backward State Space Planning (BSSP): BSSP behaves in a similar fashion like finding backNa
the
fromthe goal state gtowardssub-goal g'thatis
state space search. In this, we move
previous action to be done tto achieve that respective goal. This process is
called regression
donain
This
" One of the most famous planning domains is known as the blocks world. stacked.buton
consists of a set of cube-shaped blocks sitting on atable. The blocks can be and move it
block
one block can fit directly on top of another. Arobot arm can pick up a onlyone
pickup
another position, either on the table or on top of another block. The arm can goalwillalways
block at atime, so it cannot piCk up a bl0Ck that has another one on The ofwhat
ontop
be to build one or more stacks of blocks, specified in terms of what blocks are
other blocks. For example, a. goal might be to get block Aon Band block Bon C.
Components of planning system: heuristics
º Choose the best rulefor applying the next rule based on the best available
º Apply the chosen rule for computing the new problem state.
found.
r Detect when a solution has been
 Introduction to Planning 6.17

Detect dead ends so that they can be abandoned and the

more fruitful directions. system's effort is directed in
> Detect when an almost correct solution has been found.
oimportant approaches of problem solving are : Green's
Approach &STRIPS Approach.
Green's Approach :Basically, a given state was described by a set of
the facts that were true in that state. Each state was
predicates representing
predicate.
represented explicitly as part of the
eTpIPS is a planning algorithm that was developed by Stanford AI Lab in
cTRIPS is an acronym for "STanford Research Institute Planning System" the early 1970s.
The STRIPS
aloorithm works by breaking down a planning problem into a series of smaller
each of which can be solved independently. The algorithm then combines the sub-problems,
sub-problems to find a solution to the overall problem. solutions to the
Uncertainty in AI:When we talk about perceiving information from the environment, then
the main problem that arises is that there is always some uncertainty is our
This is because the world is an enormous entity and the observations.
surroundings that we take under
study is not always well defined. So, there needs to be an estimation taken for
conclusion.
getting to any
" Reasons for uncertainty in Al : Partially observable environment,
Dynamic Environment.
Incomplete knowledge of the agent, Inaccessible areas in the environment.
" Areasoning system is monotonic if the truthfulness of a conclusion does not change
when
new information is added to the system the set of theorem can only
when new axioms are added. In contrast, in a system doing non-monotonic
monotonically grows
of conclusions may either grow or shrink when new
reasoning the set
information is obtained.
" Probabilistic Reasoning in Al - In probabilistic reasoning,
probabilistictheory can help us
g1ve an estimate of how much an event is likely to occur or happen? In this
theory, we find the
probabilities of allthe alternatives that are possible in any experiment. The sum of all these
Pobdbities for an experiment is always because all these events/alternatives can happen
only within this
experiment.
s
special case.value
, of a proposition is extended from {0, 1}to [0, 1]. with binary logic as its
" Bayes
on sometheoremi
is a method to findthe probability of an event whose occurrence is dependent
other event's occurrence. In simple words, using the Bayes theorem, we can find the
conditional probability of any event. Bayes theorem, given by Reverend Thomas Bayes and
thus named after
him
Fuzzy Logic is a mathematical method for representing vagueness and uncertainty in

"
decision-making,concept
is
based on the
rules.
it allows for partial truths, and it is used in a wide range of applications. It
of membership function and the implementation is done using Fuzzy
The
users
xto
tOmembership function i is afunction which represents the graph of fuzzy sets, and allows
quantify the linguisticterm, It is a graph which is used for mapping each element of
the value
between 0 and 1.
6.18 Artificial Intelligence

6.8 Review Questions

Short Answer Questions

1. Define planning in context of AI.

2. List the stages of Planning in Al.
3. Briefly define two types of Planning in Al.
4. List the components of Planning system in AI.
5. Define STRIDE approach of Planning system in Al.
6. What is Uncertainty in Al?
7. List any 4 reasons for Uncertainty in Al.
8. Define Nonmonotonic logics.
9. List any twoadvantages and disadvantages of Nonmonotonic logics.
10. Justify- "ProbabilisticReasoning in Al -Away to deal with Uncertainty"
11. Define Bayes Theorem in AI.
12. List any two challenges to probabilistic approaches.
13. Define Fuzzy logic.
14. Define member functions.
15. List any two advantages of Fuzzy logic.
16. List any two disadvantages of Fuzzy logic.
17. Name any 4applications of Fuzzy logic.
Long Answer Questions

1. Explain different stages of planning in AI. Diferentiate between FSSP &BSSP.

2. With aneat diagram, Explain The block world problem.
problem. bloc
3. Define Green's Approach. Explain Green's Approach for Simple block world tothe
4. Define STRIPS approach. Explain STRIPS style operators that correspondence
world operations.
reasou.
5. With an example explain Uncertainty in Artificial Intelligence. Explain any four
6. Write a note on Nonmonotonic logics with advantages &disadvantages.
7. What is Bayes Theorem in AI? Explain.
8. Define Fuzzy logic. Explain its architecture.