Intelligent Agents

1
Intelligent Agents
2
Learning Objectives

• Agents and environment
• Rationality
• PEAS (Performance measure, Environment,
  Actuators, Sensors)
• Environment types
• Agent types

3
Acknowledgments

• These slides have been adapted from Stuart
  Russell and Peter Norvig

4
Agents and Environments

• Russell and NorvigAn agent is anything that
  can be viewed as perceiving its environment
  through sensors and acting upon them through
  actuators.

5
Agents and Environments

• Ferber
• An agent is a physical or virtual entity
• Which is capable of acting in an environment,
• Which can communicate directly with other agents,
• Which is driven by a set of tendencies,
• Which possesses resources of its own,
• Which is capable of perceiving its environment,
• Which has only a partial representation of this
  environment,
• Which possesses skills and can offer services,
• Which may be able to reproduce itself,
• Whose behavior tends towards satisfying its
  objectives, taking account of the resources and
  skills available to it and depending on its
  perception, its representations and the
  communications it receives.

6
Agents and Environments

• Agents include humans, robots, softbots,
  thermostats, etc.
• The agent function maps from percept histories to
  actions
• The agent program runs on the physical
  architecture to produce

7
Agents and Environments

• Vacuum-Cleaner World

8
Agents and Environments

• Vacuum-Cleaner World

9
Agents and Environments

• Vacuum-Cleaner World
• What is the right function?Can it be implemented
  in a small agent program?

10
Rationality

• A rational agent does the right thing.
• What is the right thing?
• One possibility
• The action that will maximize success.
• But what is success?
• The action that maximizes the agents goals.
• Use a performance measure to evaluate agents
  success.

11
Rationality

• Fixed performance measure evaluates the
  environment sequence
• One point per square cleaned up in time T
• One point per clean square per time step, minus
  one per move?
• Penalize for more than k dirty squares?
• A rational agent chooses whichever action
  maximizes the expected value of the performance
  measure given the percept sequence to date.

12
Rationality

• Rational agent definitionFor each possible
  percept sequence, a rational agent should select
  an action that is expected to maximize its
  performance measure, given the evidence provided
  by the percept sequence and whatever built-in
  knowledge the agent has.

13
Rationality

• Rationality is not
• Omniscience
• Clairvoyance
• Success
• Rationality implies
• Exploration
• Learning
• Autonomy

14
PEAS

• Task environments are the problems to which
  rational agents are the solutions.
• PEAS Performance, Environment, Actuators,
  Sensors.
• Task environment needs to be defined first.
• Example the task of designing an automated taxi.

15
PEAS

• Performance measure?
• Environment?
• Actuators?
• Sensors?

16
PEAS

• Medical diagnosis system
• Satellite image analysis system
• Part-picking robot
• Refinery controller
• Interactive English tutor

17
PEAS

• Softbots (software robots, software agents) are
  purely software agents.
• Example Internet shopping agentscans Internet
  news sources to pick interesting items for its
  customers
• Performance measures?
• Environment?
• Actuators?
• Sensors?

18
Environment Types

• Categorization of environment tasks
• Fully/partially observableextent to which an
  agents sensors give it access to the complete
  state of the environment
• Deterministic/stochasticextent to which the next
  state of the environment is determined by the
  current state

19
Environment Types

• Categorization of environment tasks
• Episodic/sequentialextent to which the agents
  experience is divided into atomic episodes
• Static/dynamicextent to which the environment
  can change while the agent is deliberating

20
Environment Types

• Categorization of environment tasks
• Discrete/continuousextent to which state of the
  environment, time, percepts and actions of the
  agent are expressed as a set of discrete values
• Single agent/multiagent

21
Environment Types
22
Environment Types
23
Environment Types
24
Environment Types
25
Environment Types
26
Environment Types
27
Environment Types
28
Environment Types

• The environment type largely determines the agent
  design
• The real world is (of course) partially
  observable, stochastic, sequentiql, dynamic,
  continuous, multi-agent

29
Agent Types

• Four basic types in order of increasing
  generality
• Simple reflex agents
• Reflex agents with state
• Goal-based agents
• Utility-based agents
• All these can be turned into learning agents.

30
Agent Types

• Agent architecture program
• Simple reflex agent
• Reflex agent with state
• Goal-based agent
• Utility-based agent
• Learning agent

31
Agent Types

• Simple reflex agent

32
Agent Types

• Reflex agent with state

33
Agent Types

• Goal-based agent

34
Agent Types

• Utility-l-based agent

35
Agent Types

• Learning agent

36
Source Code

• Source code for the book is on the book Web-site
• In Lisp, Python
• Partially in C and Java