Intelligent Agents Guide

Intelligent Agents
1AI, Subash Chandra Pakhrin

Instructional Objective
• Define an agent
• Define an intelligent agent
• Define a Rational agent
• Explain Bounded rationality
• Discuss different types of environment
• Explain different agent architectures

On completion of this lesson the student will be able to
• Understand what an agent is and how an agent
interacts with the environment.
• Given a problem situation, the student should be able
to
– Identify the percepts available to the agent and
– the actions that the agent can execute.
• Understand the performance measures used to
evaluate an agent
• Understand the definition of a rational agent
• Understand the concept of bounded rationality

• On completion of this lesson the student will
– Be familiar with
• Different agent architectures
• Stimulus response agents
• State based agents
• Deliberative / goal – directed agents
• Utility based agents
• Learning agents
– Be able to analyze a problem situation and be able to
• Identify the characteristics of the environment
• Recommend the architecture of the desired agent

Agent and Environment
Percepts: Hear, see / inputs
Actions: actuators / effectors

Agents
• Operate in an environment
• Perceives its environment through sensors
• Acts upon its environment through
actuators/effectors
• Have goals

Sensors and effectors
• An agent perceives its environment through
sensors
– The complete set of inputs at a given time is called a
percept
– The current percept, or a sequence of percepts can
influence the actions of an agent
• It can change the environment through effectors
– An operation involving an actuator is called an action
– Actions can be grouped into action sequences

Agents
• Have sensors, actuators
• Have goals
• Implement mapping from
percept sequence to
actions
• Performance measure to
evaluate agents
• Autonomous agent decide
autonomously which
action to take in the
current situation to
maximize progress towards
its goals.

Performance
• Behavior and performance of IAs in terms of
agent function
– Perception history (sequence) to Action Mapping
– Ideal Mapping: specifies which actions an agent to
take at any point in time
• Performance measure: a subjective measure
to characterize how successful an agent is
(e.g., speed, power usage, accuracy, money,
etc)

Examples of Agent
• Humans
– Eyes, ears, skin, taste buds, etc. for sensors
• Robots
– Camera, infrared, bumper, etc. for sensors
– Grippers, wheels, lights, speakers, etc. for
actuators
• Software agent (soft bots)
– Functions as sensors
– Functions as actuators

Types of Agents: Robots
http://www.ai.mit.edu/projects/humanoid-robotics-group/cog/overview.html
Cog (MIT) 11AI, Subash Chandra Pakhrin

Types of Agents: Robots
https://www.youtube.com/watch?v=8t8fyiiQVZ0
The AIBO Entertainment
Robot is a totally new
kind of robot-
autonomous, sensitive to
his environment, and able
to learn and mature like a
living creature. Since each
AIBO experiences his
world differently, each
develops his own unique
personality – different
from any other AIBO in
the world!
Aibo (SONY)

Types of Agents
• Soft bots
– Askjeeves.com
• Expert System
– Cardiologist
• Autonomous spacecraft
• Intelligent buildings

Agents
• Fundamental faculties of intelligence
– Acting
– Sensing
– Understanding, reasoning, learning
• In order to act you must sense. Blind actions is
not a characterization of intelligence.
• Robotics: sensing and acting, understanding
not necessary.
• Sensing needs understanding to be useful.

Intelligent Agents
• Intelligent Agents
– Must sense
– Must act
– Must be autonomous (to some extent),
– Must be rational.

Rational Agent
• AI is about building rational agents
• An agent is something that perceives and acts.
• A rational agent always does the right thing.
– What are the functionalities (goals) ?
– What are the components ?
– How do we build them ?

Rationality
• Perfect Rationality
– Assumes that the rational agent knows all and will
take the action that maximizes his/her utility.
– Equivalent to demanding that the agent is Omniscient.
– Human beings do not satisfy this definition of
rationality.
• Bounded Rationality Herbert Simon, 1972 (CMU)
– Because of the limitations of the human kind, humans
must use approximate methods to handle many tasks.

Rationality
• Rational Action: The action that maximizes the
expected value of the performance measure
given the percept sequence to date
– Rational = Best ?
• Yes, to the best of its knowledge
– Rational = Optimal ?
• Yes, to the best of its abilities
• And its constraints

Omniscience
• A rational agent is not omniscient
– It doesn’t know the actual outcome of its actions
– It may not know certain aspects of its
environment.
• Rationality must take into account the
limitations of the agent
– Percept sequence, background knowledge,
feasible actions
– Deal with the expected outcome of actions

Bounded Rationality
• Evolution did not give rise to optimal agents,
but to agents which are in some senses locally
optimal at best.
• In 1957, Simon proposed the notion of
Bounded Rationality:
that property of an agent that behaves in a
manner that is nearly optimal with respect to
its goals as its resources will allow.

Agent Environment
• Environments in which agents operate can be
defined in different ways.
It is helpful to view the following definitions as
referring to the way the environment appears
from the point of view of the agent itself.

Environment: Observability
• Fully observable
– All of the environment relevant to the action being
considered is observable
– Such environments are convenient, since the agent is
freed from the task of keeping track of the change in
the environment.
• Partially observable
– The relevant features of the environment are only
partially observable
• Example:
– Fully obs: Chess; Partially obs: Poker

Environment: Determinism
• Deterministic: The next state of the environment
is completely described by the current state and
the agent’s action. Image analysis
• Stochastic: If an element of interference or
uncertainty occurs then the environment is
stochastic. Note that a deterministic yet partially
observable environment will appear to be
stochastic to the agent. Ludo
• Strategic: environment state wholly determined
by the preceding state and the actions of multiple
agents is called strategic. Chess

Environment: Episodicity
• Episodic / Sequential
– An episodic environment means that subsequent
episodes do not depend on what actions occurred
in pervious episodes.
– In a sequential environment, the agent engages in
a series of connected episodes.

Environment: Dynamism
• Static Environment: does not change from one
state to the next while the agent is considering its
course of action. The only changes to the
environment as those caused by the agent itself
• Dynamic Environment: Changes over time
independent of the actions of the agent – and
thus if an agent does not respond in a timely
manner, this counts as a choice to do nothing
– Interactive tutor

Environments: Continuity
• Discrete / Continuous
– If the number of distinct percepts and actions is
limited, the environment is discrete, otherwise it
is continuous.

Environments: other agents
• Single agent/ Multi-agent
– If the environment contains other intelligent
agents, the agent needs to be concerned about
strategic, game – theoretic aspects of the
environment (for either cooperative or
competitive agents)
– Most engineering environments don’t have multi-
agent properties, whereas most social and
economic systems get their complexity from the
interactions of (more or less) rational agents.

Complex Environments
• Complexity of the environment includes
– Knowledge rich: enormous amount of information that the
environment contains and
– Input rich: the enormous amount of input the
environment can send to an agent. T
• The agent must have a way of manage this complexity.
Often such considerations lead to the development of
– Sensing strategies and
– Attentional mechanisms
• So that the agent may more readily focus its efforts in
such rich environments.

Table based agent
• Information comes from sensors – percepts
• Look it up !
• Triggers actions through the effectors
• In table based agent the mapping from percepts to
actions is stored in the form of table
Reactive agents
No notion of history, the current state is the sensors see it
right now.
Percepts Actions

Table based agent
• a table is simple way to specify a mapping from
percepts to actions
– Tables may become very large
– All work done by the designer
– No autonomy, all actions are predetermined
– Learning might take a very long time
• Mapping is implicitly defined by a program
– Rule based
– Neural networks
– algorithms

Percept based agent
• Information comes from sensors – percepts
• Changes the agents current state of the world
• Triggers action through the effectors
Reactive agents / Stimulus – response agents
Agent takes some action without any deliberation
No notion of history, the current state is as the
sensors see it right now.

Subsumption Architecture
• Rodney Brooks, 1986
• Sensory inputs – action (lower animals)
• Brooks – follow the evolutionary path and build
simple agents for complex worlds.
• Features
– No explicit knowledge representation
– Distributed behavior (not centralized)
– Response to stimuli is reflexive
– Bottom up design – complex behaviors fashioned from
the combination of simpler underlying ones.
– Inexpensive individual agents

Subsumption Architecture
• Subsumption Architecture built in layers.
• Time scale of evolution – 5 billion years (cells)
– First humans – 2.5 million years
– Symbols – 5000 years
• Different layers of behavior
• Higher layers can override lower layers.
• Each activity consists of a Finite State Machine
(FSM).

Mobile Robot Example
• Layer 0: Avoid Obstacles
– Sonar: generate sonar scan
– Collide: send HALT message
to forward
– Feel force: signal sent to
run-away, turn
• Layer 1: Wander Behavior
– Generates a random
heading
– Avoid reads repulsive force,
generates new heading,
feeds to turn and forward

Mobile Robot Example
• Layer 2: Exploration
behavior
– Whenlook notices idle
time and looks for an
interesting place.
– Pathplan sends new
direction to avoid.
– Integrate monitors
path and sends them
to the path plan.

Percept based Agent
• Efficient
• No internal representation for reasoning,
interference.
• No strategic planning, learning.
• Percept-based agents are not good for
multiple, opposing, goals.

Simple Reflex Agent (E.g. Chess)
• They choose actions only based on the current
percept.
• Their environment is completely observable.
• Condition-Action Rule – It is a rule that maps a
state (condition) to an action
Vacuum cleaner: Condition
if dirty then clean
if it is clean then stop
AI, Subash Chandra Pakhrin 39

Model Based Reflex Agents
• It doesn’t know about the whole world
• It has the knowledge of the partial world
• It keeps a small model in which the agent can
predict and control the world / perform the
action.
• There is a predefined model which we keep on
the basis of history and we predict what
action we should take now.

Updating the state requires the
information about -
• How the world evolves.
• How the agent’s actions affect the world.

State based Agent
• Information comes from sensors-percepts
• Based on state of the world and knowledge
(memory), it triggers actions through the
effectors
• In order to this the agent does some
deliberation

Goal-based Agent
• Information comes from sensors-percepts
• Based on state of the world and knowledge (memory)
and goals/intentions, it chooses actions and does them
through the effectors.
• Agent’s actions will depend upon its goal.
• Goal formulation based on the current situation is a
way of solving many problems and search is a universal
problem solving mechanism in AI.
• The sequence of steps required to solve a problem is
not known a priori and must be determined by a
systematic exploration of the alternatives.

Goal-based Agent
• Knowing something about current state of
environment is not enough to decide what to
do
• Agent also needs some sort of goal
information

Goal-based Agent

Utility based agent
• A more general framework
• Different preferences for different goals
• A utility function maps a state or a sequence
of states to a real valued utility.
• The agent acts so as to maximize expected
utility.

Learning Agent
• Learning allows an agent to operate in initially
unknown environments
• The learning element modifies the
performance element
• Learning is required for true autonomy

Learning Agent

PEAS Representation
• Task Environment
– Problems to which rational agent are solutions
– In task environment there are problems and they
are solved by rational agent

To specify Task environment, we need
• P – Performance measures (what output are
you expecting from an agent)
• E – Environment
• A – Actuators
• S - Sensors

Automated Taxi Driver Agent
• Performance Measures:
– Getting to correct
Destination
– Less cost
– High safety
• Environment:
– Variety of roads
– Traffic
– Different Types of passenger
• Actuators:
– Accelerators
– Steering & Brakes
• Sensors:
– Camera
– GPS
– IR Sensors

Medical Diagnoses System
• Performance measures
– Healthy patients
– Minimize cost
• Environment
– Patients
– Hospital
– Staff
• Actuators:
• Screen Display
(Question test,
treatment)
• Sensors
• Keyboard
(Entry of Symptoms)

Summary
• An agent perceives and acts in an
environment, has an architecture, and is
implemented by an agent program.
• An ideal agent always chooses the action
which maximizes its expected performance,
given its percepts sequence so far.
• An autonomous agent uses its own experience
rather than built-in knowledge of the
environment by the designer.

Summary
• An agent program maps from percept to action and
updates its internal state.
– Reflex agents respond immediately to percepts.
– Goal-based agents act in order to achieve their goal(s).
– Utility-based agents maximize their own utility function.
• Representing Knowledge is important for successful
agent design.
• The most challenging environments are partially
observable, stochastic, sequential, dynamic, and
continuous, and contain multiple intelligent agents.

Questions
1. Define an agent.
2. What is a rational agent ?
3. What is bounded rationality ?
4. What is an autonomous agent ?
5. Describe the salient features of an agent.

Questions
6. Find out about the Mars rover.
a. What are the percepts for this agent ?
b. Characterize the operating environment.
c. What are the actions the agent an take ?
d. How can one evaluate the performance of the agent
?
e. What sort of agent architecture do you think is most
suitable for this agent ?
7. Answer the same questions as above for an
Internet shopping agent.

Intelligent Agents Guide

Recommended

Recommended

More Related Content

Similar to Intelligent Agents Guide

Similar to Intelligent Agents Guide (20)

More from Subash Chandra Pakhrin

More from Subash Chandra Pakhrin (20)

Recently uploaded

Recently uploaded (20)

Intelligent Agents Guide

Editor's Notes