artificial-intelligence

Definition

Rational Agent

An agent is called rational if, for each possible percept history, it selects an action that is expected to maximise its performance measures, given the evidence by the percept history and whatever built-in knowledge the agent has.

“Do the right thing.”

Note that:

  • Rational omniscient: percepts may lack relevant infos
  • Rational clairvoyan: action may turn out differenlty
  • Rational successful

PEAS Features

Performance Measure

Examples: safety, destination, profits, legality, comfort, …

Environment

Examples: streets/freeways, traffic, pedestrians, weather, …

Actuators

Examples: steering, accelerator, brake, horn, speaker/display, …

Sensors

Examples: video, accelerometers, gauges, engine sensors, keyboard, GPS, …