1: Wumpus World
· Looking somewhere in the cave is the Wumpus, a beast that eats any agent that enters its room.
· Some room contains bottomless pits that trap any agent that wanders into room.
SSEE MORE: How to install Whatsapp on Windows 7
· Occasionally, there is a heap of gold in a room.
· The goal is to collect the gold and exit the World without being eaten.
Here every block is a room. Agent will move in each block one by one. The agent always starts in the field (1, 1). Agent doesn’t know what exactly is in the next room. But agent has idea.
There are three actions in this World.
PIT: When agent goes inside the PIT, it will fell down. PIT has no floor.
Beast: When agent go there. It will eaten by beast.
In above both situations, agent will either trap or die.
Arrow: Agent has one arrow only. Agent can kill beast with arrow but he doesn’t know that there is a beast in next room or not. Also, agent doesn’t know that there is PIT in next room or not.
How can we guess there is a PIT?
We have few clues or guesses about PIT is present there or not. In the side rooms of PIT, there will be breeze. For example, PIT is present in room (1, 3) then breeze will be in rooms (1, 2), (1, 4), (2, 3).
How can we guess there is a Beast?
In this case, agent will feel some stench in the sides of the room. Stench can be noise or like smell. Through this, agent can get information about the beast is present in next room or not. Gold will glowing and agent collect the gold and come back to its original point.
· The agent Perceives
Ø A stench in the square containing the Wumpus and in the adjacent squares(not diagonally).
Ø A breeze in the square adjacent to a pit.
Ø A glitter in the square where the gold is.
Ø A bump, if it walks into wall.
Ø A woeful scream everywhere in the cave, if the Wumpus is killed.
· The Percepts will be given as a five-symbol list: if there is a stench, and a breeze, but no glitter, no bump, and no scream, the percept is
[Stench, Breeze, None, None, None]
· The agent can not perceive its own location.
Wumpus Goal:
The agent’s goal is to find the gold and bring it back to the start as quickly as possible, without getting killed.
· 1000 points reward for climbing out of the cave with the gold.
· 1 point deducted for every action taken.
· 10000 penalties for getting killed.
Characteristics of Wumpus World
Fully Observable: No- Only local perception.
Deterministic: Yes
Static: Yes- Wumpus and PITS don’t move.
Discrete: Yes
Single: Yes- Wumpus is essentially a natural feature.
2: Vacuum Cleaner
There is a robot, which is doing vacuum cleaning. In this example, there are different states but have 8 states possibilities of robot.
In the first state, there are 2 blocks and robot (vacuum cleaner) is on left block. In both blocks (right and left), there exists some dust.
Similarly, in third state, there is dust on left but not on right. In 5th state, dust is not present on left but on right. Similarly, dust is neither present on right nor left. On the other hand, In 8th state, dust is not present on right and left and robot (vacuum cleaner) is present on right. In the above example, we have 8 different states and each state is a possibility.
Actions:
We have 3 different Actions: 1st is left, 2nd is right and 3rd is suck. The robot will either on right or on left. When robot goes to right or left, it will suck the dust.
Goal: Clean up all the dust=state (7) or state (8). It means that when robot is in state 7 or 8 in above image then that will be our goal. Because in both states we don’t need to clean anything. Also, seven and eight is a goal state. Goal test is clean up all the dust.
We are taking each block as state. We have representation of states as:
States: s1, s2, s3, s4, s5, s6, s7, s8
The agent can solve the problem if it can perform sensing actions during execution. For example, starting from one of {1, 3}: first suck dirt, then move right, then suck only if there is dirt there. In this case the agent must calculate a whole tree of actions rather than a single sequence.
Measuring performance
With any intelligent agent, we want it to find a (good) solution and not spend forever doing it.
The interesting quantities are, therefore,
- the search cost--how long the agent takes to come up with the solution to the problem, and
- The path cost--how expensive the actions in the solution are.
The total cost of the solution is the sum of the above two quantities.
References:
Comments
Post a Comment