Planning

Planning
 In Artificial Intelligence (AI), planning refers to the process of
deciding a sequence of actions that a system must take to
achieve a specific goal. Just like humans plan their daily tasks
with a goal in mind, AI systems use planning algorithms to
break down a problem into smaller, manageable tasks and
determine the best course of action.
 Planning is crucial for AI applications because it allows
machines to:
Think ahead: AI systems evaluate multiple possible outcomes
before selecting the optimal one.
Adapt to changes: When faced with uncertain environments, AI
systems can revise their plans based on new information.
Act autonomously: Whether it’s an autonomous robot or a smart
assistant, AI planning enables machines to perform tasks
2
independently.
  Blocks World

 Predicates describing the initial state:

 On(C, A), On(A, Table), On(B, Table), Clear(C), Clear(B)
 Predicates describing the target state:
 On(A, B), On(B, C)
 ACTIONS:
 Move(X, Y)
 Precond: Clear(X), Clear(Y)
 Effect: On(X, Y)
 Move(X, Table)
 Precond: Clear(X)
3
 Effect: On(X, Table)
 Choosing Action

 We can move C to the table

 This achieves none of the goal predicates
 We can move C to top of B
 This achieves none of the goal predicates
 We can move B to top of C
 This achieves On(B, C)

4
  Partial Solution

5
  Partial Solution

6
  Ordering Partial Solutions

7
  Ordering Partial Solutions

8
 Sometimes Partial Order may stay

9
10
  Example Planning Problem: Making Tea
 suppose you have a robot that can serve tea think about what
the robot must do to make tea, e.g.:
 it must put water in the kettle
 heat the kettle
 get a cup
 pour hot water into the cup (after the water is hot enough)
 get a tea bag
 leave the tea bag in the water for enough time
 remove the tea bag
 add milk
 add sugar
 mix the tea
 serve the tea

11
 There are many, many actions to consider! and most
actions consist of many smaller sub-actions and in
some situations very long sub-plans might be
triggered.
e.g. if there is no milk, the robot might go to the store,
or ask if creme would be an acceptable substitute
e.g. if the robot drops the spoon on the floor, then it
must pick it up and clean it before continuing
time and sensing is important as well
how long should the robot mix the tea for?

12
 Types of Planning
 Two types of planning
 Classical planning environments
Fully observable, deterministic, finite, static and discrete.
 Non-classical planning is for partially observable or
stochastic environments.
The non-classical planning involves a different set of algorithms
and agent designs.
 Classical planning is a type of planning in artificial
intelligence (AI) that involves creating a series of actions to
achieve a goal in a predetermined setting.
 In classical planning, the goal is to move a system from an
initial state to a goal state by applying actions with known and
deterministic effects.
13
 Planning Problem
 Agent environment states are represented as valuations of
state variables.
 An action can be represented as a procedure or a program.
 The procedures are used to compute values of state
variables.
 After the execution of procedure (i.e. after the action), the
environment state will be changed toward the goal.
 A planning problem in artificial intelligence (AI) involves
choosing a sequence of actions to change the world's state
to meet a goal. For example, building a house involves
deciding the order of construction

14
 Types of Planning in Artificial
Intelligence (AI)
 There are several types of planning approaches in AI, each
suited to different tasks and environments:

15
 Classical Planning
 Classical planning is the traditional form of AI planning and
assumes a static and fully observable environment where all
actions are deterministic.
 The AI agent has complete knowledge of the world and
operates with a fixed goal, attempting to find a sequence of
actions that leads from an initial state to a goal state.
 Key Concepts
 Initial State: The starting point of the problem.
 Goal State: The desired outcome or target state.
 Actions: The possible steps or moves that can be taken.
 State Transitions: The changes in the state resulting from
actions.
16
 Classical Planning
Components
 Planning Domain: Description of the environment, including:

Objects (e.g., blocks, robots)

Actions (e.g., move, pick-up)
Constraints (e.g., block A cannot be on top of block B)
 Problem Instance: Specific initial state and goal:
Initial state: Description of the initial environment (e.g., block A
is on the table)
Goal: Desired outcome (e.g., block A is on block B)
 Actions: Operators that transform the state:
Preconditions: Conditions that must hold for an action to be
applicable
Effects: Changes to the state resulting from an action
17  Goal: Desired outcome:
 Planning Process
 Problem Formulation:
Define planning domain and problem instance
 Planning: Generate a plan (sequence of actions)
Planning algorithms (e.g., Graphplan, Plan Space Planning)
 Planning strategies (e.g., forward search, backward search)
 Plan Validation: Verify plan correctness
Check plan feasibility
Check plan optimality
 Plan Execution: Execute plan

18
 Example 1
 Block Stacking
 Planning Domain:-
Blocks (A, B, C)
Actions:
Pick-up (block)
Put-down (block)
Stack (block1, block2)
Constraints:
Block A cannot be on top of block B
 Problem Instance:-
Initial State: Block A is on the table, Block B is on the table
Goal: Block A is on Block B
 Plan:
Pick-up (Block A)
19
Put-down (Block A, Block B)
  Robot Navigation Planning Domain:- Example 2
Robot-
Locations (L1, L2, L3)
Actions:
Move (robot, location)
Pick-up (object)
Deliver (object)
Constraints:
Robot cannot move to occupied location
Problem Instance:-
Initial State: Robot is at L1, Object is at L2
Goal: Object is at L3
Plan:1.
Move (Robot, L2)
Pick-up (Object)
Move (Robot, L3)
20 Deliver (Object)
  Example
 Let’s consider the Robot Navigation problem.
 Scenario:
 A robot needs to navigate from a start point to an end point in
a grid. The robot can move up, down, left, or right.
 Initial Setup:
 Initial State: Robot is at (1,1).
 Goal State: Robot needs to reach (3,3).
 Grid:

 Here S is initial state and G is the goal state.

21
  Actions:
Move Up: If not at the topmost row.
Move Down: If not at the bottommost row.
Move Left: If not at the leftmost column.
Move Right: If not at the rightmost column.
 Plan:
Move Right: (1,1) -> (1,2)
Move Right: (1,2) -> (1,3)
Move Down: (1,3) -> (2,3)
Move Down: (2,3) -> (3,3)
 Steps:
Move Right: The robot moves from (1,1) to (1,2).
Move Right: The robot moves from (1,2) to (1,3).
Move Down: The robot moves from (1,3) to (2,3).
Move Down: The robot moves from (2,3) to (3,3).
22  By following these actions, the robot reaches the goal state at (3,3).