Professional Documents
Culture Documents
2. Intelligent Agents
Dr. M. Tounsi
PSU
Definition
Is anything:
perceiving its environment through sensors
acting upon its environment through effectors
PSU
PSU
Rational Agent
One that does the right thing (most successful!)
Issue: how and when to evaluate the agents success ?
Performance Measure
Percept Sequence
PSU
Rational agents
PSU
PEAS
PSU
PEAS
PSU
PEAS
PSU
PEAS
PSU
PSU
Table-lookup agent
Function Table-Agent(percept) returns an Action
Static: percepts // sequence, initially empty
Table // table of actions, indexed by
percept, initially fully specified
Begin
Append perept to the end of the percepts
Action := LOOKUP(percepts, Table)
return Action
End
Drawbacks:
Huge table
Take a long time to build the table
No autonomy
Even with learning, need a long time to learn
the table entries
PSU
Program Agent
The design of the agent depends on the percepts, actions, goals
and environment
Function SKELETON-AGENT (percept)
Static: memory;
memory <- Update-Mem (memory, percept)
action <- Choose-Best-Action (Memory)
Memory <- Update-Memory(memory, action)
Return action
PSU
Agent types
Four basic types in order of increasing generality:
1.
2.
3.
4.
Utility-based agents
try to maximize their own happiness
PSU
PSU
Autonomy
PSU