Artificial Intelligence - UNIT - 1 Topic - 8 : Problem Solving Agents

 

PROBLEM SOLVING AGENTS

---

Part A: Introduction to Problem Solving Agents

---

1. Introduction

In Artificial Intelligence, many tasks can be thought of as problems that need to be solved — such as finding a path, playing a game, or completing a task.
A problem-solving agent is a type of intelligent agent that finds a solution through search and decision-making techniques.

---

2. What is a Problem-Solving Agent?

A Problem Solving Agent is an intelligent agent that formulates a problem, searches for a solution, and acts to reach the goal.

It doesn’t just react. Instead, it thinks ahead, evaluates options, and chooses the best possible path to solve a specific problem.

---

3. Steps of a Problem-Solving Agent

A problem-solving agent follows these steps:

Step	Description
1. Goal Formulation	Decide what goal or objective needs to be achieved.
2. Problem Formulation	Define the problem clearly (initial state, actions, goal).
3. Search	Explore different possible paths or solutions using algorithms.
4. Solution Execution	Follow the selected path and reach the goal.

---

4. Example: Maze Navigation

Let’s say an agent needs to find its way out of a maze.

• Goal: Reach the exit.
• Initial State: Start position in maze.
• Actions: Move forward, left, or right.
• Search: Try paths to find the best route.
• Solution: Sequence of moves that leads to exit.

---

Part B: Components of a Problem

---

To solve a problem, the agent needs a well-defined problem which includes:

Component	Description
Initial State	Where the agent starts (e.g., current city in a map).
Actions	Legal moves that the agent can take (e.g., move left, right, drive, etc.).
Transition Model	Description of what each action does (e.g., result of moving from one city to another).
Goal Test	How to check if the goal is reached.
Path Cost	A numeric cost of each step taken (used to find optimal solutions).

---

Part C: Types of Problem-Solving Agents

---

1. Simple Problem-Solving Agent

• Has complete knowledge of the environment.
• Solves problems before acting.
• Used in static environments.

Example: Map navigation using Google Maps.

---

2. Online Problem-Solving Agent

• Doesn’t know the environment fully in advance.
• Learns and explores as it moves.
• Used in dynamic or unknown environments.

Example: Robot vacuum exploring a new house.

---

Part D: Problem Types

---

1. Single-State Problems

• The environment is fully known.
• Only one possibility at each step.

Example: Route planning from city A to city B.

---

2. Multiple-State Problems

• The agent must track all possible situations.
• Used when the environment is partially observable.

Example: Navigating a dark room without light.

---

3. Contingency Problems

• Outcomes may be uncertain.
• The agent must plan for different cases.

Example: Planning a picnic with uncertain weather.

---

4. Exploration Problems

• Agent does not know the state space.
• Learns and discovers as it explores.

Example: Mars rover exploring unknown land.

---

Summary

• A problem-solving agent chooses actions by thinking ahead and searching for the best path to reach a goal.
• It works in four steps: Goal Formulation → Problem Formulation → Search → Solution.
• The problem must be clearly defined with initial state, actions, goal test, and path cost.
• These agents are ideal for applications like game playing, pathfinding, robotic movement, and planning.

In short, a problem-solving agent is a think-before-you-act type of agent