1. Agents and Environments

• Definition of an Agent: An agent is anything that perceives its environment through sensors and acts upon that environment through actuators.
  
  
• Key Components:
  • Percept: The agent's perceptual inputs at any specific instant.
  • Percept Sequence: The complete history of everything the agent has ever perceived.
  • Agent Function: A mathematical mapping that links any given percept sequence to an action ($f:P^{\ast }\to A$).
  • Agent Program: The internal software that runs on the physical architecture (hardware) to produce the agent function.
    
    
• Examples of Agents:
  • Human Agent: Uses eyes and ears as sensors; uses hands, legs, and mouth as actuators.
  • Robotic Agent: Uses cameras and infrared range finders as sensors; uses motors as actuators.

2. Rationality

• Definition: A rational agent is one that acts to achieve the best outcome (or best expected outcome under uncertainty). It is described as "doing the right thing".
  
  
• Factors of Rationality: Rationality depends on four "pillars":
  1. The performance measure defining success.
  2. The agent's prior knowledge of the environment.
  3. The actions the agent can perform.
  4. The agent's percept sequence to date.
     
     
• Distinctions:
  • Rationality vs. Omniscience: Rationality is based on what is perceived, not necessarily all possible information (which may be unavailable).
    
    
  • Rationality vs. Perfection: Rationality maximizes the expected outcome, while perfection maximizes the actual outcome.
    
    
• Autonomy: An agent is autonomous if it can learn from its perceptions and act without human assistance. An ideal agent should have some autonomy and increase it through experience.

3. PEAS: Specifying the Task Environment

To design an agent, one must specify its "PEAS" description:

• P - Performance Measure: The objective criterion for the agent's success (e.g., safety, speed, or profit).
  
  
• E - Environment: The external world or context the agent operates in (e.g., roads, traffic, or a set of students).
  
  
• A - Actuators: The tools the agent uses to change the environment (e.g., steering wheel, screen display, or robotic arm).
  
  
• S - Sensors: The tools the agent uses to perceive the environment (e.g., cameras, GPS, or keyboard).

4. Task Environment Characteristics

Task environments are classified by several dimensions:

• Observability: Fully observable (sensors see everything needed to choose an action) vs. partially observable.
  
  
• Certainty: Deterministic (next state is determined only by the current state and action) vs. stochastic (uncertainty exists).
  
  
• Episodic vs. Sequential: In episodic environments, the current action does not affect future actions; in sequential environments, agents must plan ahead.
  
  
• Static vs. Dynamic: Static environments do not change while the agent is "thinking"; dynamic environments change over time.
  
  
• Discrete vs. Continuous: Discrete environments have a limited number of distinct percepts and actions (like Chess); continuous environments use a range of values (like Taxi driving).
  
  
• Single agent vs. Multiagent: Whether the agent works alone or interacts with others.