Characteristics of Autonomous Agents

Environments

• An environment with a limited number of distinct, clearly defined percepts and actions is characterized as having discrete state and action spaces.
• The reward function determines if an environment is considered hostile or friendly.

Single-Agent Environments

• A single-agent environment involves a single agent solving a problem individually, like an agent solving a crossword puzzle.

Agent Types

• A simple reflex agent acts only on the basis of the current percept and ignores the rest of the percept history.

Environment Types

• An environment with a known outcome based on the current state and action taken is called a fully observable environment.
• Competitive environments, where agents compete to maximize their performance and minimize their opponents' performance, are exemplified by chess.
• Sequential environments, where infinite loops are often unavoidable for simple reflex agents, require search and planning for agents.
• Multi-agent systems involve multiple agents interacting with each other.

Agent Characteristics

• Goal-based agents base their decisions on their distance from the goal.
• Model-based reflex agents are more flexible because they maintain an internal model of the environment.

Environment Characteristics

• An environment is deterministic if the next state is completely determined by the current state and action.
• An environment is friendly if the reward function is designed to encourage the agent to learn.
• In an episodic environment, the agent's experience is divided into separate, independent episodes.
• A non-episodic environment has a continuous sequence of actions and states.