Sort by information used:
Uninformed methods (DFS, BFS) know only the graph shape. They cannot judge which state looks closer to the goal.
Binary search is not even a state-space technique; it halves a sorted list.
Informed methods add a heuristic $h(n)$ that guesses the remaining cost to the goal. A* combines it with the known path cost using $f(n) = g(n) + h(n)$ and always expands the smallest $f$ node.
The phrase 'estimate the cost from the current state to the goal state' is the plain meaning of the heuristic term $h(n)$, which only A* uses here.
\[\boxed{\text{(D) A* Search}}\]