Description:
- In realistics games, it is hard to compute all Terminal State
- We only explore at depth n, the nodes at that layer is said to be terminal state
- Optimality is goneUse iterative deepening for an anytime algorithm
- Danger of replanning:
- example:
- Both bottom left and the bottom right has the same evaluation score
- So going left or right is the same for a first move
- Say he goes left, then looking at future 2 choices, it is exactly the symmetric
- Then he goes right, without making any move, then infinitely turn left and right