Intelligent Agent

INTELLIGENT AGENT:

Agent            =  perceive+act

·                                  Thinking

·                                  Reasonig

·                                  Planning

Agent: entity in a progra m or environment capable of generating action.

An agent uses perceptio n of the environment to make decisions about actions to take. The perception capability is usually called a sensor.

The actions can depend on the most recent perception or on the entir e history (percept sequence).

An agent is anything that can be viewed as perceiving its environment through s ensors and acting upon the environment through ac tuators.

Ex:  Robotic agent

Human agent

Agents interact with environment through sensors and actuators.

Percept sequence  action

[A, clean]    right

[A, dirt]      suck

[B, clean]    left

[B, dirty]     suck

[A, clean], [A, clean]      right

[A, clean], [A, dirty]      suck

Fig: practical tabulation of a simple agent function for the vacuum cleaner world

Agent Function

1.The agent function is a mathematical function that maps a sequence of perceptions into action.

2.                            The function is imple mented as the agent program.

3.                          The part of the agent t aking an action is called an actuator.

4.                            Environment ®sensors ®agent function ®actuators ®environment

RATIONAL AGENT:

A rational agent is one that can take the right decision in every situation.

Performance measure: a set of criteria/test bed for the success of the age nt's behavior.

The performance measu res should be based on the desired effect of the agent on the environment.

Rationality:

The agent's rational beha vior depends on:

1.the performance me asure that defines success

2. the agent's knowledge of the environment

3.the action that it is capable of performing

4 .The current sequen ce of perceptions.

Definition: for every po ssible percept sequence, the agent is expected to take an action that will maximize its performance measure.

Agent Autonomy:

An agent is omniscient i f it knows the actual outcome of its actions. Not possible in practice. An environment can sometimes be completely known in advanc e.

Exploration: sometimes an agent must perform an action to gather information (to increase perception).

Autonomy: the capacity to compen sate for partial or incorrect prior knowledge (usually by learning).

NATURE OF ENVIRONMEN TS:

Task environment – the problem that the agent is a solution to. Includes

Performance measure

Environment

Actuator

Sensors

Properties of Task Environment:

•                            Fully Observable (vs. P artly Observable)

–  Agent sensors give complete state of the environment at each poi nt in time

–  Sensors detect all the aspect that is relevant to the choice of actio n.

– An environment might be partially observable because of noisy a nd inaccurate sensor s or apart of the state are simply missing from the sensor data.

•                            Deterministic (vs. Stochastic)

–  Next state of the e nvironment is completely determined by the current state  and the action executed by the agent

– Strategic environment (if the environment is deterministic except for the actions of other agent.)

•                      Episodic (vs. Sequential)

– Agent’s experience can be divided into episodes, each episode with what an agent perceive and what is the action

•                                                                   Next episode does not depend on the previous episode

–  Current decision will affect all future sates in sequential environment

•                      Static (vs. Dynamic)

–  Environment doesn’t change as the agent is deliberating

–  Semi dynamic

•                      Discrete (vs. Continuous)

–  Depends the way time is handled in describing state, percept, actions

•                           Chess game : discrete

•                           Taxi driving : continuous

•                      Single Agent (vs. Multi Agent)

–  Competitive, cooperative multi-agent environments

–  Communication is a key issue in multi agent environments.

Partially Observable:

Ex: Automated taxi cannot see what other devices are thinking. Stochastic:

Ex: taxi driving is clearly stochastic in this sense, because one can never predict the behaviorof the traffic exactly.

Semi dynamic:

If the environment does not change for some time, then it changes due to agent’s performance is called semi dynamic environment.

Single Agent Vs multi agent:

An agent solving a cross word puzzle by itself is clearly in a single agent environment.

An agent playing chess is in a two agent environment.

Example of Task Environments and Their Classes

Four types of agents:

1.                           Simple reflex agent

2.                          Model based reflex agent

3.                          goal-based agent

4.                          utility-

base agent

Simple reflex agent

Definition:

SRA works 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.

Characteristics

– no plan, no goal

–  do not know what they want to achieve

–  do not know what they are doing

Condition-action rule

–  If condition then action

Ex: medical diagnosis system.

Algorithm Explanation:

Interpret – Input:

Function generates an abstracte d description of the current state from the percept.

RULE- MATCH:

Function returns the first rule in the set of rules that matches the given state description.

RULE - ACTION:

The selected rule is executed as action of the given percept.

Model-Based Reflex Agents:

Definition:

An agent which combines the current percept with the old internal state to generate updated description of the current state.

If the world is not fully o bservable, the agent must remember observations about the parts of the environment it cannot currently observe.

This usually requires an internal representation of the world (or internal state).

Since this representation is a model of the world, we call this model-bas ed agent.

Ex: Braking problem

characteristics

1.Reflex agent with internal state

2.Sensor does not provide the complete state of the world.

3. must keep its internal state

Updating the internal wo rld

requires two kinds of knowledge

1.                                                               How world evolves

2.                                                               How agent’s action affect the world

Algorithm Explanation:

UPDATE-INPUT: This is responsible for creating the new internal stated description.

Goal-based agents:

The agent has a purpose and the action to be taken depends on the curre nt state and on what it tries to acc omplish (the goal).

In some cases the goal is easy to achieve. In others it involves planning, sifting through a search space for possible solutions, developing a strategy.

Characteriscs

–  Action depends on the goal. (consideration of future)

–        e.g. path finding

–  Fundamentally different from the condition-action rule.

–  Search and Planning

–  Solving “car-braking” problem?

–  Yes, possible … but not likely natural.

•                      Appears less efficient.

Utility-based agents

If one state is preferred over the other, then it has higher utility for the agent

Utility-Function (state) = real number (degree of happiness)

The agent is aware of a utility function that estimates how close the current state is to the agent's goal.

•                      Characteristics

–  to generate high-quality behavior

– Map the internal states to real numbers. (e.g., game playing)

Looking for higher utility value utility function

Learning Agents

Agents cap able of acquiring new competence through obse rvations and actions. Learning agent has the following components

Learning eleme nt

Suggests modification to the existing rule to the critic

Performance ele ment

Collectio n of knowledge and procedures for selecting the driving actions

Choice de pends on Learning element

Critic

Observes the world and passes information to the learnin g element

Problem generator

Identifies certain areas of behavior needs improvement and suggest experiments

Agent Example

A file manager agent.

Sensors: commands like ls, du, pwd.

Actuators: commands lik e tar, gzip, cd, rm, cp, etc.

Purpose: compress and archive files that have not been used in a while.

Environment: fully obse rvable (but partially observed), deterministic (st rategic), episodic, dynamic, discrete.

Problem Formulation

Problem formulation is the process of deciding what actions and states to consider, given a goal

PROBLEMS

Four components of problem definition

–  Initial state – that the agent starts in

–  Possible Actions

•                                                                 Uses a Successor Function

–  Returns <action, successor>

pair

•                                                                 State Space – the state space forms a graph in which the nodes are states and arcs between nodes are actions.

•                                                                 Path

–  Goal Test – which determine whether a given state is goal state

–  Path cost – function that assigns a numeric cost to each path.

SOME  REAL-WORLD PROBLEMS

•                   Route finding

•                   Touring (traveling salesman)

•                   Logistics

•                   VLSI layout

•                   Robot navigation

•                   Learning

TOY PROBLEM

Example-1 : Vacuum World

Problem Formulation

•                   States

–  2 x 2² = 8 states

–  Formula n2ⁿ states

•                   Initial State

–  Any one of 8 states

•         Successor Function

–  Legal states that result from three actions (Left, Right, Suck)

•         Goal Test

–  All squares are clean

•         Path Cost

–  Number of steps (each step costs a value of 1)

State Space for the Vacuum World.

Labels on Arcs denote                                L: Left, R: Right, S: Suck