To generate the relaxed plan, FF starts with the three goal propositions In(B), closed, Polished For each goal it selects a noop or action from action layer 2 that achieves it, preferring no-ops over actions. The selected actions/noops are highlighted in bold. In (B)is achieved by action Move(A, B), Closed is carried forward by a noop, and Polished is achieved using the Polish action. The corresponding subgoals for fact layer 2 are collected and consist of In(A), Closed, Open Next, actions and noops from action layer l are selected to achieve these subgoals. In(A) and Closed are achieved by noops, while Open is achieved with the Close action. The corresponding subgoals are all propositions in the initial state relaxed plan, shown in bold, consists of three actions--Open, Move(A, B)and polish, a The heuristic value of Node 1 is the number of actions in the relaxed plan. The comple lence. the heuristic value of node 1 is 3 FF expands Node 1 by considering the application of each action in the first layer of the relaxed plan. Note that FF only uses an action in this first layer if the effects of the action are used by an action in layer 2. ff stops evaluating actions as soon as it finds an action whose target state has a better heuristic value. If the heuristic value does not improve, it performs a breadth first search For Node 1, only one action, Closed, appears in the first layer of the relaxed plan, hence it is the only action expanded. The expanded graph is shown below, with Node 2 added Node 1 In(A) Estimate 3 Closed Node 2 In(A)Estimate? Opened lext we generate the relaxed plan graph and plan for Node 2, as shown belowTo generate the relaxed plan, FF starts with the three goal propositions: In(B), Closed, Polished. For each goal it selects a noop or action from action layer 2 that achieves it, preferring no-ops over actions. The selected actions/noops are highlighted in bold. In(B) is achieved by action Move(A,B), Closed is carried forward by a noop, and Polished is achieved using the Polish action. The corresponding subgoals for fact layer 2 are collected, and consist of: In(A), Closed, Open Next, actions and noops from action layer 1 are selected to achieve these subgoals. In(A) and Closed are achieved by noops, while Open is achieved with the Close action. The corresponding subgoals are all propositions in the initial state. The heuristic value of Node 1 is the number of actions in the relaxed plan. The complete relaxed plan, shown in bold, consists of three actions -- Open, Move(A,B) and Polish – hence, the heuristic value of Node 1 is 3. FF expands Node 1 by considering the application of each action in the first layer of the relaxed plan. Note that FF only uses an action in this first layer if the effects of the action are used by an action in layer 2. FF stops evaluating actions as soon as it finds an action whose target state has a better heuristic value. If the heuristic value does not improve, it performs a breadth first search. For Node 1, only one action, Closed, appears in the first layer of the relaxed plan, hence it is the only action expanded. The expanded graph is shown below, with Node 2 added: Next we generate the relaxed plan graph and plan for Node 2, as shown below: Node 1 In(A) Estimate 3 Closed Estimate ? Open Node 2 In(A) Opened