Please enable JavaScript.
Coggle requires JavaScript to display documents.
Reflex and goal-based agents - making decisions - Coggle Diagram
Reflex and goal-based agents - making decisions
Decision Making in Agents
Requires knowledge representation
Agents evaluate current state and environment
Must choose next action to meet a goal or follow rules
References
https://www.ibm.com/think/topics/ai-agent-types
https://utdes.com/2024/03/15/react-vs-plan-understanding-reflex-agents-and-goal-based-agents/
https://smythos.com/ai-agents/ai-agent-development/types-of-intelligent-agents/
https://www.simform.com/blog/types-of-ai-agents/
Types of Agents
Simple Reflex Agent
Uses condition-action rules (if...then)
Acts only on current percept
Model-Based Reflex Agent
Maintains internal state
More accurate in partially observable environments
Goal-Based Agent
Makes decisions based on defined goals
Uses search and planning
Utility-Based Agent
Chooses actions based on a utility function
Aims for the most beneficial outcome
Finite State Machine (FSM)
Definition
FSM is a model that defines a system's behavior using states and transitions based on inputs.
Why FSM is Useful
Can simulate memory by storing previous state
Intuitive and visual
Easy to debug and manage
Simple to implement
Low computational cost (rule-based)
Embed in Classes (State Design Pattern)
Each state = a class
ex: PatrolState, FightState, FlightState
Handles its own entry, behavior, and exit
Makes transitions by calling methods
Benefits:
Modular and scalable
Easy to debug
Clean and reusable architecture
Types of FSM
Deterministic
Predictable behavior (eg: pedestrian crossing)
Example: Pedestrian crossing light
Non-Deterministic
Multiple possible outcomes
Good for game AI
Flexible and adaptive
Can be event-driven, fuzzy, or randomized
States & Transitions
State: Represents a specific behavior or condition
Transition: Movement from one state to another
Triggered by: Input events (conditions)
Pac-Man ghost states example:
Waiting → Chasing → Evading