The algorithm will pass each immediate next state to a function which will return some value. The action/ future state which gets the max value is selected as the next move.

Thus the goal is to maximize some function.

Things that would influence agent's movement

• Score
• Distance from food
  • [x] lesser the distance from food, higher is the will to move in that direction
  • [x] if len(food) is getting reduced, more reward
• Distance from ghosts (implemented even though not required)
  • [x] The farther away from ghosts the better
  • [ ] Punish for being too close to ghosts ?
• Distance from power pellet (not implemented as not required)
  • [ ] Reward for moving close to the power pellet
• Scared time of ghosts (not implemented as not required)
  • [ ] Don't consider Distance from ghosts if this is not null