Algorithm 3.2.

Expand subroutine of RTBSS.

1:	Function Expand(b, d)
	Inputs: b: The belief node we want to expand.
	d: The depth of expansion under b.
	Static: T: An AND-OR tree representing the current search tree.
	L: A lower bound on V *.
	U: An upper bound on V *.
2:	if d = 0 then
3:	LT (b) ← L(b)
4:	else
5:	Sort actions {a1, a2, …, a|A|} such that U(b, ai) ≥ U(b, aj) if i ≤ j
6:	i ← 1
7:	LT (b) ← −∞
8:	while i ≤ |A| and U(b, ai) > LT (b) do
9:	LT (b, ai) ← RB(b, ai) + γ Σz ∈ Z Pr(z|b, ai)Expand(τ(b, ai, z), d − 1)
10:	LT (b) ← max{LT (b), LT (b, ai)}
11:	i ← i + 1
12:	end while
13:	end if
14:	return LT (b)