Please enable JavaScript.
Coggle requires JavaScript to display documents.
Goal-Based Agents & Pathfinding, images (2), SCC1, Mu6VZ - Coggle…
Goal-Based Agents & Pathfinding
Definition
What is a Goal-Based Agent?
Goal-based agents solve problems by setting and reaching goals
Goals help organize behavior by narrowing possible actions
Limitations
Requires complete model or map
Can be computationally expensive
Performance depends on the heuristic
Components of Goal-Based Agent
Goal Formulation
Search/Planning Module
Internal Model (understanding of the environment)
Action Execution
Perception (via sensors)
Advantages
Efficient decision-making
Can adapt to changing environments
Goal-directed behavior
Search Strategies
Uninformed Search
Depth-First Search (DFS) – Explores deep paths first
Uniform Cost Search
Breadth-First Search (BFS) – Explores evenly layer by layer
Informed Search (Heuristic-based)
A* – Combines path cost + estimated cost to goal
Uses heuristics (e.g., Manhattan or Euclidean distance)
Finds optimal path efficiently
Greedy Best-First Search – Chooses closest to goal (heuristic)
Heuristics
Common types: Manhattan distance, Euclidean distance
What is a heuristic?
Used to estimate remaining cost to goal
References
https://www.simform.com/blog/types-of-ai-agents/
https://www.geeksforgeeks.org/goal-based-ai-agents/?utm_source=chatgpt.com
https://how.dev/answers/what-is-a-goal-based-agent
https://smythos.com/ai-integrations/tool-usage/goal-based-ai-agents/
Graphs and Representation
Nodes = states
Edges = actions or transitions
Directed Graphs
Edges point from one node to another
Undirected Graphs
Edges connect nodes both ways
Weighted Edges
Cost included (distance, time, etc.)
Examples
Website structure
Tile-based games
Circuit networks
Social media
Genome sequences
Problem Solving Process
Goal Formulation
Based on current state and performance measure
Search
Finds a sequence of actions to reach the goal
Well-Defined Problem Components
Initial State – Starting position (e.g., in London)
Actions – Options available at each step (e.g., go to Norwich)
Transition Model – Result of an action
Goal Test – Check if the goal is achieved (e.g., am I in Edinburgh?)
Path Cost Function – Measures cost (e.g., distance, time, effort)
Examples & Applications
Pac-Man ghosts (chase, evade)
Navigation systems (Google Maps)
Delivery drones
Robotics path planning
Tile/grid games
