/.R. Marx et al.
» Overtake on her port side until she is finally past and clear (COL-
REG 13).
» Evade to the starboard side in a head-on situation in a fairway (COL-
REG 14).
Give the right of way in a crossing situation
(COLREG 9).
Testing automatic evasive maneuvers in port is a major safety chal-
‚enge, especially when a real ship under the command of a captain is
involved. Therefore, a compromise must be found between the colli-
sion risks associated with manually operated ships and the design for
automatic avoidance. Automatic evasion must be able to be safely re-
solved at any time by manual intervention. This generally applies to
conventional port operations as well, but due to the limited maneuver-
ng space and the constant risk of collision, the responsible navigators
nust also maintain the highest level of concentration in order to be able
oO assess the situation proactively. Nevertheless, accidents do happen,
causing damage to people, machines, and the environment, which must
ve completely ruled out for tests conducted in automatic mode.
In addition to these traffic rules, the internal and external conditions
ander which automatic maneuvering is applicable must be clearly de-
fined. Nowadays, every experienced nautical officer knows for his own
ship when the installed autopilot should no longer be used, from which
weather or traffic situation or for which maneuvers and speeds it is not
zuitable. External conditions include wind, current, sea conditions, wa-
zer depth and visibility. Internal conditions describe all properties of
che ship that influence her motion behavior, the condition of the ship’s
aull and propulsion units, maneuverability, loading and the functions
of sensors and actuators. The large number of parameters and variables
zlarifies the complexity of a ship maneuver, even if it is a single ship.
The degree of complexity increases when several ships maneuver coop-
eratively in confined waters in port.
In order to transfer such a maneuver from manual to automatic con-
:rol mode, control engineers reduce the complexity, firstly in simulation
aut also later on for real-world experiments. The simulation works with
vehicle motion models that are assumed to be disturbance-free. The in-
Iuence of the environment is initially neglected. First experiments with
‚eal ships take place in designated or separate areas that are closed to
public traffic and in moderate weather and sea conditions.
Ocean Engineering 343 (2026) 123388
The maneuver scenarios and their safe, collision-free solution are de-
signed conservatively so that there is sufficient maneuvering space even
ın the event of an error, the maneuverability of the individual vehicles is
:aken into account and the skipper of each vessel can assume his respon-
ibility for the crew and vessel during the entire automatic maneuver.
This integrates the practice of good seamanship into the optimization
aılgorithm by following the principle of not unnecessarily increasing the
complexity of encounter situations. The analysis not only assesses the
collision risk, but also determines which evasive maneuver should be
aısed in accordance with the COLREGs.
2.3. Defined safety aspects
The nautical officers of DENEB, in particular the captain, and ISSIMS
GmbH supported the development of the automatic maneuvers and the
definition of the safety aspects with their maritime expertise. Due to the
very different sizes of the vehicles, DENEB, as the largest vessel, is used
as a benchmark for the safety distances. The main focus of DENEB’s nav-
'gators is on starboard ahead, for which a vessel’s length is defined as
‘he safety distance, i.e. 50 m. For the sides and stern, a distance of 25m
rom other ships or fixed objects is sufficient. The maximum allowed
wind speed is initially set at 8 m/s. The DENEB should not be maneu-
vered automatically in higher wind forces and especially if the wind
s coming from behind and supports the forward motion. In DP mode,
‘he wind plays less of a role as there is hardly any motion in the ship.
Mandatory for navigators, each vehicle must have a responsible skipper
or chief mate at all times who has no other tasks. This operator must
»e able to resume manual control at any time if the automatic scenario
s aborted. When comparing the target and actual trajectories, the fol-
owing values are used as abort criteria, 5 deg deviation in course angle,
10 m cross track error, and 1 m/s velocity difference. These values were
determined empirically by the captain, taking into account the size of
he maneuvering space in the naval port, especially at the point of col-
ision risk, and the maneuverability of the DENEB. The combination of
che values mentioned is particularly critical.
2.4. Initial encounter situation
The finally planned encounter situation is shown in Fig. 5. It is
nılanned for the Hohe Düne naval port near Rostock, which is closed
4.18
54 178
54 176
- 54.174
2
3 54.172
ij
SA 47
5A 468
54 16€
4 164 -
” 4105
7 41
12.115 12.17
Lonaitude [*] East
4172 175
17
7ig. 5. Initial trajectories of an encounter situation in a port for the shins DENEB, BELA and MESSIN that would lead to multiple collisions.
