KUSHPAL SINGH YADAV
1. Partially Observable:
It is not possible for an agent’s sensor to see all the choices(states) because an agent does not know what other agents are going to do.
Because it is not possible for an agent to predict the outcomes, here next state cannot be determined by the current state or if it is partially observable then it will be stochastic.
An agent makes a decision which can affect future decisions also because in the game agent cannot decide which option is best.
e.g: Kicking the ball towards the opponent team.
The agent has to look at the environment whenever agent takes any decision on an action.
e.g: If a player moves from one place to other places then the field has to change.
In the field, an agent has to move continuously to hit the ball so player’s place is not fixed that’s why it is continuous.
In the task environment, there are many agents so it is multi-agent.
For this scenario, Utility-Based Agent is the best choice.
Suppose a player wants to put a ball into goal post then what problems can arise while putting the ball into the goal post and what is the possible solutions which can be had by Utility-Based Agent.
1. It might possible for a player that he/she will not be able to reach the goal because there are some other players who want to post the ball in another goal post so players will try to stop other players so that player will not be able to score a goal.
In this scenario, Utility-Based Agent will calculate the best path which one is the best where there will be no player so that player can reach as fast as possible by maximizing the value of its utility function.
2. When a player kicks the ball if the ball touches other player’s hand or a player can hit other players then a penalty is awarded to other team and a player is given a red card.
Here Utility-Based Agent tries to manage himself such a way that there will be a minimum penalty by using utility function how much I am happy or unhappy.
3. If a player collides with other players of the opposite team then it comes in fouls category.
In this situation while achieving the goal, Agent can have different utilities quicker, safer, reliable or Utility-Based Agent can also handle the situation if a player comes to the other player immediately by a utility function.
In this scenario, scoring a goal is not good enough because if a player moves towards the goal post he will have to face many problems like he has to fight with other players after he has to reach to the goal post if reaches there then there will be a goalkeeper for all of this we need a good agent who can manage all of this a utility function maps each state after each action to a real number representing how efficiently each action achieves the goal.