TutorialsArena

Navigating the Wumpus World: Knowledge Representation and Reasoning in AI

Explore the Wumpus World, a classic AI problem, and learn how knowledge-based agents use logic and reasoning to navigate a dangerous environment, overcome challenges, and achieve their goals. Discover the principles of knowledge representation and intelligent decision-making in AI.

Navigating the Wumpus World: A Knowledge-Based Agent

Introduction to the Wumpus World

The Wumpus World is a classic AI problem used to illustrate knowledge representation and reasoning in intelligent agents. Based on the game "Hunt the Wumpus" (Gregory Yob, 1973), it involves an agent navigating a cave system with dangers (Wumpus, pits) and a goal (finding gold). The agent must use its senses and logical reasoning to survive and succeed.

The Wumpus World Environment

The Wumpus World is typically a 4x4 grid of interconnected rooms. The agent starts in a safe room ([1,1]). The environment contains:

  • Wumpus: A creature that kills the agent if encountered.
  • Pits: Bottomless pits that trap the agent.
  • Gold: The agent's goal is to find the gold and exit the cave.

The Wumpus is stationary. The agent has one arrow to kill the Wumpus. The agent receives a reward for finding the gold and penalties for being killed by the Wumpus or falling into a pit.

Agent's Sensory Inputs

The agent uses these sensory inputs to understand its environment:

  • Stench: Indicates the presence of a Wumpus in an adjacent room.
  • Breeze: Indicates the presence of a pit in an adjacent room.
  • Glitter: Indicates the presence of gold in the current room.
  • Bump: Indicates the agent has walked into a wall.
  • Scream: Indicates the Wumpus has been killed.

These percepts are typically represented as a five-element list (e.g., [Stench, Breeze, None, None, None] means the agent detects stench and breeze but nothing else).

PEAS Description of the Wumpus World

Here's a PEAS (Performance, Environment, Actuators, Sensors) description:

  • Performance Measure: +1000 for exiting with gold; -1000 for death (Wumpus or pit); -1 for each action; -10 for using the arrow.
  • Environment: A 4x4 grid of rooms, Wumpus and gold placed randomly (excluding [1,1]), pits have a 0.2 probability in each square (excluding [1,1]).
  • Actuators: Turn left, turn right, move forward, grab, release, shoot.
  • Sensors: Stench, breeze, glitter, bump, scream.

Exploring the Wumpus World: An Example

(A step-by-step example of the agent navigating the Wumpus world using logical reasoning is described in the original text. This example, including diagrams showing the agent's movement and inferences, is omitted here for brevity, but would be included as text and diagrams in the HTML.)

The example illustrates how the agent uses its sensory inputs and logical deductions (e.g., inferring the Wumpus's location based on stench) to safely navigate the cave, avoid dangers, and find the gold.