=== p1 ===
Introduction to 
Artificial Intelligence
Chapter 2
Intelligent Agents
Wei-Ta Chu (朱威達)
1

=== p2 ===
• An agent is anything that can be viewed as perceiving its environment
through sensors and acting upon that environment through actuators (促動
器). 
Agents and Environments
2

=== p3 ===
• The term percept refers to the agent’s perceptual inputs at any given instant. 
• An agent’s behavior is described by the agent function that maps any given 
percept sequence to an action. 
• A very simple example—the vacuum-cleaner world
Agents and Environments
3

=== p4 ===
• It can choose to move left, move right, suck up the dirt, or do nothing. One 
very simple agent function: if the current square is dirty, then suck; otherwise, 
move to the other square. 
• What is the right way to fill out the table? What makes an agent good or bad, 
intelligent or stupid?
Agents and Environments
4

=== p5 ===
• A rational agent is one that does the right thing. 
• Right thing: The agent’s actions causes the environment to go through a 
sequence of states. If the sequence is desirable, then the agent has performed 
well. This notion of desirability is captured by a performance measure that 
evaluates any given sequence of environment states.
• It is better to design performance measures according to what one actually 
wants in the environment, rather than according to how one thinks the agent 
should behave.
The Concept of Rationality
5

=== p6 ===
• Definition of a rational agent
• For each possible percept sequence, a rational agent should select an 
action that is expected to maximize its performance measure, given the 
evidence provided by the percept sequence and whatever built-in 
knowledge the agent has.
• Consider the simple vacuum-cleaner agent. Is this a rational agent? That 
depends! 
• Performance measure: the amount of dirt being cleaned up
• Performance measure: a clean floor
Rationality
6

=== p7 ===
• Rationality maximizes expected performance, while perfection maximizes 
actual performance. 
• Our definition of rationality does not require omniscience (全知), because the 
rational choice depends only on the percept sequence to date. 
• Our definition requires a rational agent not only to gather information but 
also to learn as much as possible from what it perceives. 
Omniscience, Learning, and Autonomy
7

=== p8 ===
• A rational agent should be autonomous—it should learn what it can to 
compensate for partial or incorrect prior knowledge.
• A vacuum-cleaning agent that learns to foresee where and when additional 
dirt will appear will do better than one that does not.
• The incorporation of learning allows one to design a single rational agent that 
will succeed in a vast variety of environments.
Omniscience, Learning, and Autonomy
8

=== p9 ===
• Task environment: we have to specify the performance measure, the 
environment, and the agent’s actuators and sensors.
• PEAS (Performance, Environment, Actuators, Sensors) 
Task Environments
9

=== p10 ===
Task Environments
10
[IMAGE-ONLY]

=== p11 ===
• Fully observable vs. partially observable
• Fully observable: if the sensors detect all aspects that are relevant to the 
choice of action
• Single agent vs. multiagent
• The key distinction is whether B’s behavior is best described as 
maximizing a performance measure whose value depends on agent A’s 
behavior. 
• Chess is a competitive multiagent environment
Properties of Task Environments
11

=== p12 ===
• Deterministic vs. stochastic
• If the next state of the environment is completely determined by the 
current state and the action executed by the agent, then we say the 
environment is deterministic; otherwise, it is stochastic.
• Taxi driving is clearly stochastic
• “Stochastic” generally implies that uncertainty about outcomes is 
quantified in terms of probabilities
Properties of Task Environments
12

=== p13 ===
• Episodic (情節不連貫的) vs. sequential
• Episodic: the next episode does not depend on the actions taken in 
previous episodes.
• Chess and taxi driving are sequential
• Static vs. dynamic
• If the environment can change while an agent is deliberating (思考), then 
we say the environment is dynamic for that agent. 
• Taxi driving is clearly dynamic: the other cars and the taxi itself keep 
moving while the driving algorithm dithers (猶豫) about what to do next. 
Properties of Task Environments
13

=== p14 ===
• Discrete vs. continuous
• The chess environment has a finite number of distinct states. Chess also 
has a discrete set of percepts and actions. Taxi driving is a continuous-
state and continuous-time problem. 
• Known vs. unknown
• The agent’s (or designer’s) state of knowledge about the “laws of physics” 
of the environment.
Properties of Task Environments
14

=== p15 ===
Properties of Task Environments
15
[IMAGE-ONLY]

=== p16 ===
• The job of AI is to design an agent program that implements the agent 
function— the mapping from percepts to actions.
• We assume this program will run on some sort of computing device with 
physical sensors and actuators—we call this the architecture.
• agent = architecture + program 
The Structure of Agents
16

=== p17 ===
• A trivial agent program
• The table-driven approach to agent construction is doomed to failure – tables 
could be too huge
Agent Programs
17

=== p18 ===
• Select actions on the basis of the current percept, ignoring the rest of the 
percept history.
Simple Reflex Agents
18

=== p19 ===
Simple Reflex Agents
19
• Condition-action rule
• Work only if the correct decision can 
be made on the basis of only the 
current percept—that is, only if the 
environment is fully observable.

=== p20 ===
• Handle partial observability: the agent keeps track of the part of the world it 
can’t see now. The agent should maintain some sort of internal state that 
depends on the percept history. 
Model-based Reflex Agents
20

=== p21 ===
Model-based Reflex Agents
21
[IMAGE-ONLY]

=== p22 ===
• Knowing the current state of environment is not always enough to decide what 
to do. The agent needs some sort of goal information that describes situations 
that are desirable. 
• The goal-based agent’s behavior can easily be changed to go to a different 
destination, simply by specifying that destination as the goal. 
• Decision making different from condition-action rules
• Consideration of the future – both “What will happen if I do such-and-
such?” and “Will that make me happy?”
Goal-based Reflex Agents
22

=== p23 ===
Goal-based Reflex Agents
23
[IMAGE-ONLY]

=== p24 ===
• Goals alone are not enough to generate high-quality behavior in most 
environments. Sometimes, goals are inadequate but a utility-based agent can 
still make rational decisions.
• When there are conflicting goals, only some of which can be achieved
• When there are several goals that the agent can aim for
• A rational utility-based agent chooses the action that maximizes the expected 
utility of the action outcomes. 
Utility-based Reflex Agents
24

=== p25 ===
• Utility-based agent programs handle the uncertainty inherent in stochastic or 
partially observable environments.
Utility-based Reflex Agents
25

=== p26 ===
• Learning element is responsible for making improvements, and the 
performance element is responsible for selecting external actions.
• The learning element uses feedback from the critic on how the agent is doing 
and determines how the performance component should be modified to do 
better in the future. 
Learning Agents
26