Forward Chaining

• Forward chaining goes from the current state to the goal state.
• You can do this by simulating it, or, with for example a robot, doing it.
• If you simulate, it's a plan, if you make the move, it's just search.
• The initial state is: On(B, Ground), On(A, Ground), On(C,A), Clear(B) and Clear(C).
• The goal state is: On(C, Ground), On(B, C), and On(A,B).
• How do you get from the initial state to the goal state?
• Forward chaining looks at which moves are currently possible.
• Move(C,A,Ground) or Move(B,Ground,C)
• You can do a depth first search.
• Move(B,Ground,C) is wrong despite it reducing the differences in the states.
• You want to Move(C,A,Ground), then Move(B,Ground,C) and finally Move(A,Ground,B).