Abstract: This paper presents a parallel simulation based predictive control algorithm for automatic
berthing. The proposed scheme uses many number of control sequences for on-line simulation of
future behaviour of the ship movement. Each control sequence used in simulation generates a predicted
sequence of the ships future course. The predicted courses of the ship are evaluated based on some cost
function suitable for control objective and then, the optimal control signal is applied to the ship. To
simulate the future behaviour of the ship, it is used a non-linear ships dynamic model (called MMG
model) and also the massively parallel computational hardware (called GPU: Graphical Processing Unit).
Application of GPU for simulation (GPGPU) permits directly use of the nonlinear model of the ship for
parallel simulation and makes the control algorithm with highly computational demand feasible.
In order to investigate the effectiveness of the proposed algorithm, computer simulations and actual sea
tests are carried out using a training ship Shioji Maru.
The experimental results show that the proposed method is successfully applied to the approaching
control for automatic berthing.
Key Words: ship control, predictive control, simulation based control, parallel simulation, GPGPU
Based on the above mentioned algorithm and implementation Moreover, Mizuno et al. [17] have extended the above
for modern GPU, we develop a feasible approaching control method applicable to minimum-time approaching control for
system for automatic berthing. automatic berthing under some disturbances. The extended
It is very important for a ship's master to draw up a system is composed of a neural network based optimal
ship-handling plan before approaching a berth, leaving it, solution generator and a nonlinear model predictive
altering the heading and so on. In order to cope with these compensator. The neural network generates the optimal
problems, they are required to derive the maximum solution for real situation by interpolating pre-computed
maneuverability from the ship. The minimum-time control minimum-time berthing solutions for typical control
technique is one of the rational and effective maneuvering conditions. The optimal solutions for the various minimum
methods for such tedious problems if the mathematical model time berthing are numerically computed based on the
representing a ship's dynamics is available. However, it should sophisticated non-linear dynamical model of the ship (MMG
be noted that the model becomes highly non-linear, especially model) and are learned off-line by the neural network for
in the case of low speed and large maneuvering motion. In this interpolation. Moreover, the same nonlinear model, which is
case, the solutions of minimum-time maneuvering problem used for the computation of the optimal solutions, is used to
can be obtained by solving non-linear two-point boundary simulate the ships future position and heading. Based on the
value problems (TPBVP). receding horizon cost function of the predicted control error
In order to take enough account of the non-linearity, Shoji caused by some disturbances, the control inputs for minimum
and Ohtsu [3] have formulated these problems as a non-linear, time approaching are modified.
two-point boundary-value problem (TPBVP) in the calculus of
variations. The problems have been solved, using the In this paper, we extend the above non-linear model
numerical method called the sequential conjugate gradient predictive compensator to achieve more accurate tracking
restoration (SCGR) method [4], [5] under various situations: under some disturbances by parallel simulation based
1) The minimum-time course-alteration problem, approach. Based on the parallel simulation, the control input
2) The minimum-time stopping problem, which achieve the minimum time maneuvering and
3) The minimum-time parallel deviation problem [6], disturbance rejection under realistic sea conditions. This is a
4) The minimum-time maneuvering problem with wind new feature of this system.
disturbances [7],
5) The minimum-time berthing problems [3], [8]. First, the basic idea of parallel simulation based predictive
However, the solutions, unfortunately, do not yield on-line control is briefly reviewed. Next, the berthing problems and
control laws, because, it took long computational time to the mathematical model of the ships dynamics for parallel
obtain the optimal solution even by using a high-speed simulation of ship behaviour are explained. In the third part of
workstation. Moreover, the geometrical and other sea this paper, an approaching control system for automatic
conditions of maneuvering problem facing to the real ship berthing is introduced. Finally, computer simulations and
handling scenes are various. Thus, it is impossible to prepare on-line experiments are carried out for a training ship Shioji
whole minimum time maneuvering solutions to such various Maru (425 gross tonnage). The experimental results show that
conditions. the proposed method is successfully applied.
For these problems, Okazaki et al., [9] have investigated the
minimum time maneuvering with neural network. In the
proposed method, an off-line trained neural network 2. BASIC IDEA OF PARALLEL SIMULATION BASED
interpolates the real-time maneuvering solution from the CONTROL
available solutions, which has been already solved, but the 2.1 Formulation of Simulation Based Control
control performance varies depend on the disturbances in real
sea conditions. To overcome this problem, Mizuno et al., [10] The purpose of the controller is typically to force the output to
have proposed a ships minimum-time maneuvering system follow the reference signal. If the dynamics of the system is
with two types of neural networks. This system contains the linear, the control input can be solved using the linear model
on-line trained neural network, which compensates the of the process.
difference between the real ships dynamics and the
mathematical model, and the other errors caused by some However, in case of non-linear system, the input cannot be
disturbances etc. From the experimental results, the method solved in general case.
has some advantages compared with the previous studies. For this situation, the simple idea of the solving algorithm is
There are some papers in which the neural networks are used to compute the following quantities for every sample period:
in order to compensate the non-linear dynamics of the ship 1) the behaviour of system output over a finite horizon (N);
during the tracking or berthing phase [11]-[14]. Unfortunately 2) the cost of the objective function for all (prescribed) control
the on-line learning speed of the neural network is rather slow. sequences:
This means that it is difficult to obtain the good transient
performance for tracking or berthing in real situations. u(.) = {u(t), u(t + 1),.., u(t + N )} (1)
From this point of view, Mizuno et al., [15],[16] have
proposed a new design method for ships minimum time 3) to choose the first element of control sequence which gives
maneuvering system with neural network and non-linear the best cost.
model based compensator.
20
IFAC MCMC 2012
September 19-21, 2012. Arenzano, Italy
t0 t1 t2 tf
Simulator Simulator Simulator
Fig.1 Candidate of input sequences
21
IFAC MCMC 2012
September 19-21, 2012. Arenzano, Italy
v r
O y
22
IFAC MCMC 2012
September 19-21, 2012. Arenzano, Italy
Before implementing the proposed parallel simulation based 6.2 Real Time Control System
predictive control system, computer simulations of proposed
scheme are performed for nonlinear dynamical model of the The Shioji Maru has her own on-board control system with
Shioji Maru. Figure 8 shows the typical simulation results for GPS position measurement as shown in Fig. 10. However, to
approaching control for automatic berthing with wind implement the parallel simulation function, we use the
disturbances by using proposed scheme. notebook type computer with GPU for real-time controller.
The conditions of parallel simulations are also indicated.
23
IFAC MCMC 2012
September 19-21, 2012. Arenzano, Italy
6.3 Experimental results The authors wish to thank the crewmembers of the Shioji
Maru, for their helpful support in the actual sea tests.
Figure 11 shows the typical experimental result using This work is supported in part by Grant-in-Aid for Scientific
proposed control scheme. Research (c) from the Ministry of Education, Culture, Sports,
Science and Technology of Japan.
References
24