2. Artificial Intelligence Approach
Agents and Environments:
AI approach is called intelligent agent
An agent is anything that can perceive its environment through
sensors and acts upon that environment through effectors.
A function of an agent that maps sensors to effectors is called
Control-Policy of the agent.
4. An agent is anything that can be viewed as perceiving its
environment through sensors and acting upon that environment
through effectors
A rational agent is one that does the right action. The right action is
the one that will cause the agent to be the most successful.
Performance measure is the criteria that determine how successful
an agent is e.g. % accuracy achieved, amount of work done, energy
consumed, time in seconds etc.
Intelligent Agent
5. An ideal rational agent:
For each possible percept sequence, an ideal rational agent should
do whatever action is expected to maximize its performance
measure, on the basis of the evidence(information) provided by the
percept sequence and whatever built-in knowledge the agent has.
Intelligent Agent
6. Mapping:
An agent’s behaviour depends only on its percept sequence
We can describe any particular agent by making a table of the action
it takes in response to each possible percept sequence.
Intelligent Agent
P1
P2
P3
A1
A2
A3
Percept
Sequence
Action
Mapping from percept sequnce to actions
7. Autonomy:
A system is autonomous, if it’s behaviour is determined by its own
experience and learning
If the agent’s actions are based completely on built-in knowledge,
then we say that the agent lacks autonomy.
It would be reasonable to provide an artificial agent with some built-
in knowledge as well as ability to learn.
A truly autonomous intelligent agent should be able to operate
successfully in a wide variety of environments, given sufficient time
to adapt
Intelligent Agent
8. Agent program: A function that implements the agent mapping from
precepts to actions.
Computing device on which program will run, we call architecture
The architecture makes the percepts from the sensors available to
the program, run the program, and feeds the program’s action to the
effectors.
Agent = architecture + program
Agent Program
9. Designing agent program, we need to keep in mind
Possible percepts and actions
Performance measures (Goals)
What sort of environment it will operate
Agent Program
Agent type Percepts Actions Goals Enviroments
Taxi driver
(self driving car)
Cameras,
speedometer,
GPS, sonar,
microphone etc
Steer,
accelarate,
brake, talk to
passenger
Safe, fast,
legal,
comfortable
trip,
maximize
profits
Roads, traffic,
pedestrians,
customers
11. Fully observable vs. partially observable:
If an agent's sensors give it access to the complete state of the
environment at each point in time then we say that the environment
is fully observable/accessible to that agent e.g.
Game of chess (fully observable)
Poker, self driving car (partially observable)
Properties of Environments
12. Deterministic vs. nondeterministic:
If the next state of the environment is completely determined by the
current state and the actions selected by the agent, then we say
environment is deterministic e.g.
Chess (deterministic)
Dice, poker, taxi driver (non deterministic)
If environment is fully observable and deterministic then agent need
not to worry about uncertainty
Properties of Environments
13. Episodic vs. non episodic:
The agent's experience is divided into atomic "episodes" (each
episode consists of the agent perceiving and then performing a single
action), and the quality of action in each episode depends only on
the episode itself.
Subsequent episodes do not depend on what actions occur in
previous episodes e.g.
Part-picking robot (Episodic)
Chess, poker, Taxi driving (nonepisodic)
Properties of Environments
14. Static vs. dynamic:
The environment can change while an agent is deliberating then we
say the environment is dynamic for that agent otherwise it is static
e.g.
Chess (static)
Taxi driving (dynamic)
Properties of Environments
15. Discrete (vs. continuous):
A limited number of distinct, clearly defined percepts and actions ,
we say that environment is discrete e.g.
Chess (discrete)… there are fixed number of possible moves on each
turn
Driving (continuous)
Properties of Environments
16. Different environment types require different agent programs to
deal with them effectively.
Environment that is partially observable, non episodic, dynamic
and continuous is hardest for artificial intelligent agent to deal
with.
Properties of Environments
17. Types of agents are
1. Simple reflex agents
2. Model-based reflex agents
3. Goal-based agents
4. Utility-based agents
5. Learning agents
Agent Program
18. Simple Reflex Agents
They choose actions only based on the current percept.
They are rational only if a correct decision is made only on the basis
of current precept.
Their environment is completely observable.
Condition-Action Rule − It is a rule that maps a state (condition) to
an action.
19. Simple Reflex Agents
This agent function only succeeds when the environment is fully
observable
e.g. If the car-in-front is braking then initiate braking.
If hand is in fire then pull away hand
If there is a rock then pick it up (Mars lander)
20. Model-based Reflex Agents
Model-based reflex agents are made to deal with partial
accessibility
They use a model of the world to choose their actions. They
maintain an internal state.
Model − The knowledge about how the things happen in the world.
Internal State − It is a representation of unobserved aspects of
current state depending on percept history.
Updating the state requires the information about −
1. How the world evolves.
2. How the agent’s actions affect the world.
21. Model-based Reflex Agents
e.g. This time out mars Lander after picking up its first sample, it
stores this in the internal state of the world around it so when it come
across the second same sample it passes it by and saves space for
other samples.
22. Goal-based Reflex Agents
Goal-based agents further expand on the capabilities of the model-
based agents, by using "goal" information.
Goal information describes situations that are desirable.
This allows the agent a way to choose among multiple
possibilities/actions, selecting the one which reaches a goal state.
Search and planning are the subfields of artificial intelligence
devoted to finding action sequences that achieve the agent's goals.
24. Utility-based Reflex Agents
Just having goals isn’t good enough because often we may have
several actions which all satisfy our goal so we need some way of
working out the most efficient one.
A utility function maps each state after each action to a real number
representing how efficiently each action achieves the goal.
This is useful when we either have many actions all solving the same
goal
27. Learning Agents
A learning agent can be divided into four conceptual components,
Learning element, which is responsible for making improvements
Performance element select actions. The performance element is what we have
previously considered to be the entire agent: it takes in percepts and decides on
actions.
The learning element uses CRITIC feedback from the critic on how the agent is doing
and determines how the performance element should be modified to do better in
the future
Problem generator is responsible for suggesting actions that will lead to new and
informative experiences
Editor's Notes
#2:Effector: something that respond to signal/stimulus/information
#3:Effector: something that respond to signal/stimulus/information
