,LECTURE I
Al &
Intelligent Agents
intelligent Agent Paradigm .
An viewed
agent is everything that can be as
perceiving as environment
through sensors and
acting upon that environment through effectors
Internally ,
an
agent is :
agent
=
architecture +
program
core business of Al :
designing agent architectures &
programs
Natural agents
Animals :
can perceive using senses
-
can act
by their muscle movements
-
cognitive skills : perception attention , memory language and problem
-
, , ,
solving
Humans :
-
same as animals
versatile
lability to adapt) survive environment
-
: can in
any
advanced skills
cognitive
'
Plants
can also
perceive and act
'
no
cognitive skills needed
-
Artificial Agents
Robotic Agents
their
'
-
can perceive environment using their cameras , microphones , touch
sensors . . . .
-
can act upon their environment by their motor and sound systems ,
displays .
.
. .
currently not versatile can function in specific contexts
-
:
currently do not have advanced cognitive skills limited perception
'
:
,
detention language learning and problem solving skills
,
memory , , ,
But , robots can be
programmed
,Intelligent Agents
-
Reactive : ability to receive information and respond
to take
Pro active ability initiative
'
-
:
Social and
ability communicate cooperate
'
:
to
-
Autonomous :
agents control their own processes
cognitive C- Affective) Agents
An agent is anything that can be
carefully)
viewed as a
system that has beliefs desires
. .
goals ,
intentions , plans , expectations .
nope , fear , joy ,
- . . ( intentional Stance ,
Dennett
1987)
intentional systems
First -
order
bee ( p) the
agent that believes that p
-
:
the
goal Cp) agent that has a
goal p
-
:
second -
order
!
bee ca ; bee ( bjp) a believes that b believes that p
'
:
bee Ca ; goal Cb p)) a believes that b wants that p
-
:
Ca ; bell b ; p)) wants that b believes that
goal
:
a p
-
goal ( ai goal ( b p)) that
:
; a wants p
-
" "
our notion of cognitive agent
Agents with
following basic capacities
the :
and
I . Event processing process events like percepts :
messages
2 .
Knowledge representation :
process events like percepts and messages .
It allows to maintain a model of the environment and other
agents (e g -
.
use propositional FOL ,
or
Prolog)
3. Decision is able action based
making :
agent to select an on its
-
beliefs ,
knowledge ,
and
goals
cognitive state the : internal state of a
cognitive agent
Typically includes :
Event component
-
percepts
-
messages
-
Informational component
'
:
( static)
Knowledge
-
-
beliefs (dynamic)
Motivational component
-
goals
-
, Environments
Agents are situated in an environment
percepts :
Agents to usually not see the real state of the environment
but only receive percepts
Designer : the
designer has to process percepts and
possibly store them
within the agent's memory
Environment properties : such it is important know the
as
very to
-
characteristics of an environment , before designing an
agent
properties of Environments
I partially if environment is
Fully observable the not
completely
-
:
observable the
agent will need internal States
Deterministic 1 stochastic : deterministic if completely determined
by
'
's
agent action .
If the environment is only partially observable ,
then it stochastic
may appear
static dynamic environment unite
agent is
'
the can
:
change an
deliberating
Discreet continuous if there is limited number of percepts and
-
:
a
actions the environment is discrete
single / multi agents is there just one
agent or are there several
-
:
interacting with each other
cognitive Agents & Environments
Environment offers controllable entities that are connected to cognitive
agents
Possible view :
-
cognitive agent is the mind
controllable entitity is the
body
-
An action specification defines
union actions are available to an
agent and when , defined by the
environment
Interacting with Environment &
Agents
Observation
passive perceives results observations without
the
taking
-
:
of any
initiative or control to observe
Active which Observations
actively initiates controls
-
and
the
agent
:
it wants to perform
Al &
Intelligent Agents
intelligent Agent Paradigm .
An viewed
agent is everything that can be as
perceiving as environment
through sensors and
acting upon that environment through effectors
Internally ,
an
agent is :
agent
=
architecture +
program
core business of Al :
designing agent architectures &
programs
Natural agents
Animals :
can perceive using senses
-
can act
by their muscle movements
-
cognitive skills : perception attention , memory language and problem
-
, , ,
solving
Humans :
-
same as animals
versatile
lability to adapt) survive environment
-
: can in
any
advanced skills
cognitive
'
Plants
can also
perceive and act
'
no
cognitive skills needed
-
Artificial Agents
Robotic Agents
their
'
-
can perceive environment using their cameras , microphones , touch
sensors . . . .
-
can act upon their environment by their motor and sound systems ,
displays .
.
. .
currently not versatile can function in specific contexts
-
:
currently do not have advanced cognitive skills limited perception
'
:
,
detention language learning and problem solving skills
,
memory , , ,
But , robots can be
programmed
,Intelligent Agents
-
Reactive : ability to receive information and respond
to take
Pro active ability initiative
'
-
:
Social and
ability communicate cooperate
'
:
to
-
Autonomous :
agents control their own processes
cognitive C- Affective) Agents
An agent is anything that can be
carefully)
viewed as a
system that has beliefs desires
. .
goals ,
intentions , plans , expectations .
nope , fear , joy ,
- . . ( intentional Stance ,
Dennett
1987)
intentional systems
First -
order
bee ( p) the
agent that believes that p
-
:
the
goal Cp) agent that has a
goal p
-
:
second -
order
!
bee ca ; bee ( bjp) a believes that b believes that p
'
:
bee Ca ; goal Cb p)) a believes that b wants that p
-
:
Ca ; bell b ; p)) wants that b believes that
goal
:
a p
-
goal ( ai goal ( b p)) that
:
; a wants p
-
" "
our notion of cognitive agent
Agents with
following basic capacities
the :
and
I . Event processing process events like percepts :
messages
2 .
Knowledge representation :
process events like percepts and messages .
It allows to maintain a model of the environment and other
agents (e g -
.
use propositional FOL ,
or
Prolog)
3. Decision is able action based
making :
agent to select an on its
-
beliefs ,
knowledge ,
and
goals
cognitive state the : internal state of a
cognitive agent
Typically includes :
Event component
-
percepts
-
messages
-
Informational component
'
:
( static)
Knowledge
-
-
beliefs (dynamic)
Motivational component
-
goals
-
, Environments
Agents are situated in an environment
percepts :
Agents to usually not see the real state of the environment
but only receive percepts
Designer : the
designer has to process percepts and
possibly store them
within the agent's memory
Environment properties : such it is important know the
as
very to
-
characteristics of an environment , before designing an
agent
properties of Environments
I partially if environment is
Fully observable the not
completely
-
:
observable the
agent will need internal States
Deterministic 1 stochastic : deterministic if completely determined
by
'
's
agent action .
If the environment is only partially observable ,
then it stochastic
may appear
static dynamic environment unite
agent is
'
the can
:
change an
deliberating
Discreet continuous if there is limited number of percepts and
-
:
a
actions the environment is discrete
single / multi agents is there just one
agent or are there several
-
:
interacting with each other
cognitive Agents & Environments
Environment offers controllable entities that are connected to cognitive
agents
Possible view :
-
cognitive agent is the mind
controllable entitity is the
body
-
An action specification defines
union actions are available to an
agent and when , defined by the
environment
Interacting with Environment &
Agents
Observation
passive perceives results observations without
the
taking
-
:
of any
initiative or control to observe
Active which Observations
actively initiates controls
-
and
the
agent
:
it wants to perform
