lOMoAR cPSD|5906219 0
Chapter 2 1
Chapter 2: Intelligent Agents
Reminders:
Assignment 0 (lisp refresher) due 1/28
Lisp/emacs/AIMA tutorial: 11-1 today and Monday, 271 Soda
Outline:
Agents and environments
♦ Ra onality
♦ PEAS (Performance measure, Environment, Actuators, Sensors)
♦ Environment types
Agent types
Chapter 2
2
lOMoAR cPSD|59062190
Agents and environments
Agents include humans, robots, so bots, thermostats, etc.
The agent func on maps from percept histories to ac ons:
f : P
→A
The agent program runs on the physical architecture to produce f
Chapter 2
Chapter 2 3
lOMoAR cPSD|59062190
Vacuum-cleaner world
A
B
Percepts: loca on and contents, e.g., [A,Dirty]
Ac ons: Left, Right, Suck, NoOp
A vacuum-cleaner agent
Percept sequence Ac on
4
lOMoAR cPSD|59062190
[A,Clean] Right
Chapter 2 5
lOMoAR cPSD|59062190
What is the right func on?
Can it be implemented in a small agent program?
Chapter 2 6
lOMoAR cPSD|59062190
Rationality
Fixed performance measure evaluates the environment
sequence – one point per square cleaned up in me T? – one
point per clean square per me step, minus one per move? –
penalize for > k dirty squares?
A ra onal agent chooses whichever ac on maximizes the expected value
of the performance measure given the percept sequence to date
Ra onal =6 omniscient
percepts may not supply all relevant informa on
Ra onal =6 clairvoyant
ac on outcomes may not be as expectedHence, ra onal
=6 successful
Chapter 2 7
lOMoAR cPSD|59062190
Ra onal explora on, learning, autonomy
PEAS
To design a ra onal agent, we must specify the task environment
Consider, e.g., the task of designing an automated taxi:
Performance measure??
Environment??
Actuators??
Sensors??
Chapter 2 8
lOMoAR cPSD|59062190
PEAS
To design a ra onal agent, we must specify the task environment
Consider, e.g., the task of designing an automated taxi:
Performance measure?? safety, des na on, profits, legality, comfort, ...
Environment?? US streets/freeways, traffic, pedestrians, weather, ...
Actuators?? steering, accelerator, brake, horn, speaker/display, ...
Sensors?? video, accelerometers, gauges, engine sensors, keyboard, GPS,
...
Chapter 2 9
lOMoAR cPSD|59062190
Internet shopping agent
Performance measure??
Environment??
Actuators??
Sensors??
lOMoAR cPSD|5906219 0
Chapter 2 10
Internet shopping agent
Performance measure?? price, quality, appropriateness,
efficiency
Environment?? current and future WWW sites, vendors,
shippers
Chapter 2 11
lOMoAR cPSD|59062190
Environment types
Actuators?? display to user, follow URL, fill in form
Sensors?? HTML pages (text, graphics, scripts)
lOMoAR cPSD|5906219 0
Chapter 2 12
Solitaire Backgammon Internet shopping Taxi
Observable??
Determinis c??
Episodic??
Sta c??
Discrete??
Chapter 2 13
lOMoAR cPSD|59062190
Environment types
Single-agent??
lOMoAR cPSD|5906219 0
Chapter 2 14
Environment types
Solitaire Backgammon Internet shopping Taxi
Observable??
Determinis c??
Episodic??
Sta c??
Discrete??
Single-agent??
Yes Yes No No
Chapter 2 15
lOMoAR cPSD|59062190
Chapter 2 16
lOMoAR cPSD|59062190
Environment types
Solitaire Backgammon Internet shopping Taxi
Observable?? Yes Yes No No
Chapter 2 17
lOMoAR cPSD|59062190
Determinis c??
Episodic??
Sta c??
Discrete??
Single-agent??
Yes No Partly No
Environment types
Solitaire Backgammon Internet shopping Taxi
Chapter 2 18
lOMoAR cPSD|59062190
Observable?? Yes Yes No No
Determinis c?? Yes No Partly No
Episodic??
Sta c??
Discrete??
Single-agent??
No No No No
Environment types
Chapter 2 19
lOMoAR cPSD|59062190
Solitaire Backgammon Internet shopping Taxi
Observable?? Yes Yes No No
Determinis c?? Yes No Partly No
Episodic?? No No No No
Sta c??
Discrete??
Single-agent??
Yes Semi Semi No
Chapter 2 20
lOMoAR cPSD|59062190
Environment types
Solitaire Backgammon Internet shopping Taxi
Observable?? Yes Yes No No
Determinis c?? Yes No Partly No
Episodic?? No No No No
Sta c?? Yes Semi Semi No

Preview text:

lOMoAR cPSD|5906219 0 Chapter 2: Intelligent Agents Reminders:
Assignment 0 (lisp refresher) due 1/28
Lisp/emacs/AIMA tutorial: 11-1 today and Monday, 271 Soda Outline: ♦ Agents and environments ♦ Ra onality
♦ PEAS (Performance measure, Environment, Actuators, Sensors) ♦ Environment types ♦ Agent types Chapter 2 Chapter 2 1 lOMoAR cPSD|590621 90 Agents and environments
Agents include humans, robots, so bots, thermostats, etc.
The agent func on maps from percept histories to ac ons: f : P∗ →A
The agent program runs on the physical architecture to produce f Chapter 2 2 lOMoAR cPSD|590621 90 Vacuum-cleaner world B A
Percepts: loca on and contents, e.g., [A,Dirty]
Ac ons: Left, Right, Suck, NoOp A vacuum-cleaner agent Percept sequence Ac on Chapter 2 3 lOMoAR cPSD|590621 90 [A,Clean] Right 4 lOMoAR cPSD|590621 90 What is the right func on?
Can it be implemented in a small agent program? Chapter 2 5 lOMoAR cPSD|590621 90 Rationality
Fixed performance measure evaluates the environment
sequence – one point per square cleaned up in me T? – one
point per clean square per me step, minus one per move? –
penalize for > k dirty squares?
A ra onal agent chooses whichever ac on maximizes the expected value
of the performance measure given the percept sequence to date Ra onal =6 omniscient
– percepts may not supply all relevant informa on Ra onal =6 clairvoyant
– ac on outcomes may not be as expectedHence, ra onal =6 successful Chapter 2 6 lOMoAR cPSD|590621 90
Ra onal ⇒ explora on, learning, autonomy PEAS
To design a ra onal agent, we must specify the task environment
Consider, e.g., the task of designing an automated taxi: Performance measure?? Environment?? Actuators?? Sensors?? Chapter 2 7 lOMoAR cPSD|590621 90 PEAS
To design a ra onal agent, we must specify the task environment
Consider, e.g., the task of designing an automated taxi:
Performance measure?? safety, des na on, profits, legality, comfort, ...
Environment?? US streets/freeways, traffic, pedestrians, weather, ...
Actuators?? steering, accelerator, brake, horn, speaker/display, ...
Sensors?? video, accelerometers, gauges, engine sensors, keyboard, GPS, ... Chapter 2 8 lOMoAR cPSD|590621 90 Internet shopping agent Performance measure?? Environment?? Actuators?? Sensors?? Chapter 2 9 lOMoAR cPSD|5906219 0 Internet shopping agent
Performance measure?? price, quality, appropriateness, efficiency
Environment?? current and future WWW sites, vendors, shippers Chapter 2 10 lOMoAR cPSD|590621 90 Environment types
Actuators?? display to user, follow URL, fill in form
Sensors?? HTML pages (text, graphics, scripts) Chapter 2 11 lOMoAR cPSD|5906219 0
Solitaire Backgammon Internet shopping Taxi Observable?? Determinis c?? Episodic?? Sta c?? Discrete?? Chapter 2 12 lOMoAR cPSD|590621 90 Environment types Single-agent?? Chapter 2 13 lOMoAR cPSD|5906219 0 Environment types
Solitaire Backgammon Internet shopping Taxi Observable?? Yes Yes No No Determinis c?? Episodic?? Sta c?? Discrete?? Single-agent?? Chapter 2 14 lOMoAR cPSD|590621 90 Chapter 2 15 lOMoAR cPSD|590621 90 Environment types
Solitaire Backgammon Internet shopping Taxi Observable?? Yes Yes No No Chapter 2 16 lOMoAR cPSD|590621 90 Determinis c?? Yes No Partly No Episodic?? Sta c?? Discrete?? Single-agent?? Environment types
Solitaire Backgammon Internet shopping Taxi Chapter 2 17 lOMoAR cPSD|590621 90 Observable?? Yes Yes No No Determinis c?? Yes No Partly No Episodic?? No No No No Sta c?? Discrete?? Single-agent?? Environment types Chapter 2 18 lOMoAR cPSD|590621 90
Solitaire Backgammon Internet shopping Taxi Observable?? Yes Yes No No Determinis c?? Yes No Partly No Episodic?? No No No No Sta c?? Yes Semi Semi No Discrete?? Single-agent?? Chapter 2 19 lOMoAR cPSD|590621 90 Environment types
Solitaire Backgammon Internet shopping Taxi Observable?? Yes Yes No No Determinis c?? Yes No Partly No Episodic?? No No No No Sta c?? Yes Semi Semi No Chapter 2 20