Algorithm 1: Beam Search
1:	CT: List of current trajectories, where T is one trajectory in the list (CT is initially empty)
2:	FT: Set of final trajectories (FT is initially empty)
3:	OT: Own trajectory.
4:	asT: A current processed segment of T trajectory.
5:	maxN: Maximal number of processed solutions.
6:	Add OT to CT
7:	while CT is not empty do
8:		Remove T (the first trajectory of CT) from the start of list CT
9:		for each Target(i) ship’s trajectory do
10:			if the DCPA of the present ship is violated in section asT by Target(i)’s trajectory then
11:				Generate two anti-collision maneuvers (port and starboard) and segments back to the waypoint
12:				Based on the two maneuvers and T, create two new trajectories (T1 and T2)
13:			else
14:				Generate an artificial collision course TA of the present ship with Target(i)’s trajectory.
15:				For the artificial course TA, generate two anti-collision maneuvers (port and starboard) and segments back to the waypoint
16:				Based on the two maneuvers and TA, create two new trajectories (T1 and T2)
17:			end if
18:			for each trajectory from {T1, T2} set do
19:				if the first of two newly created segments of Ti trajectory is DCPA safe with respect to all remaining ships then
20:					if the segments are last segments in Ti and both are DCPA safe then
21:						Add Ti to FT
22:					else
23:						Add Ti to the end of list CT
24:					end if
25:				end if
26:			end for
27:		end for
28:		if size(CT) > maxN then
29:			Remove size(CT) from maxN worst trajectories (in terms of expected length) from CT
30:		end if
31:	end while
32:	Choose the best trajectory (or trajectories) from FT.