Auto steering operation refers to controlling the vessel’s movement automatically without pilot’s involvement, and is being accomplished by controlling the rudder angle based on the vessel’s heading angle. Referring to the statistics [1], each year the demand to possess recreational marine crafts is dramatically increasing, which has made this industry to be a stable, profitable and attractive market considered by governments worldwide. According to this, South Korea, as world’s second largest merchant ship developer country [2] has decided to enter to this market and help Korean small and medium-size businesses compete with companies around the globe. In recent years, the dramatic improvements in technology have had profound impacts on maritime industry which has resulted in development of intelligent systems for vessels. As the use of tablets and smartphones has become rife, more companies are interested in introducing their product converged with this new technology. For sure Android operating system, which is the most popular operating system on smartphones [3], is being considered as a tool for developers to implement their ideas. In our proposed method, the mariner can navigate the leisure vessel using tablet, smartphone, PDA or any other types of embedded systems which have android operating system installed. Also in this paper, besides having auto-steering application on the android system, several navigational applications such as ECDIS, AIS and Radar can be installed on the same platform which results in cut back on expenses by not buying separate monitoring systems. In large vessels, the rudder controller is a continuous system

while in small boats including leisure crafts, it is an on/off controller due to the power saving purpose. The rudder system in a ship is usually based on hydraulic systems. The idea of auto-steering is that the controller receives the desired heading angle and forces the hydraulic actuators to perform in a way which the desired track being fulfilled and keep the vessel on the course [4][5]. Since the controllers’ performance depends on sea environment, the vessel may have different responses under different circumstances. In order to prove the stability of the proposed controllers under such circumstances, an accomplishment of simulations and tests of the whole system in computer is of great importance. In the paper at first, the whole idea is implemented and tested in Simulink/Matlab software. In the second stage a similar hydraulic rudder system is designed, connected to a NMEA 2000 network with a heading sensor and examined by the proposed method on it. Third, the designed system is installed on a leisure boat and the effectiveness of our proposed method is proved. At last, the results of onboard test is discussed.

NMEA 2000 is a plug and play means of communication between the electrical instruments in a vessel. Today, almost all the pleasure vessels worldwide are using this network as a way to connect vessels’ instruments together. In this study, we have used both NMEA 0183 and NMEA 2000 network. As shown in Figure 1, the Android platform sends and receives the data in NMEA 0183 format and NMEA 2000 stack board translates them into NMEA 2000 format.

Figure 1: 
Data exchange between Android and NMEA network

2.1 NMEA 0183

NMEA 0183 is a combined electrical and data specification, which through specific sentence formats makes it possible for marine electronic instruments to exchange data between each other [7]. In NMEA 0183, only printable ASCII characters are allowed and all settings and orders should be expressed by this format. Figure 2 shows an example of a NMEA 0183 sentence. It should be noted that the length of sentences varies based on different types of packets.

Figure 2: 
NMEA 0183 sentence example for GPS sentence

The sentences used in this study are as follows:

HTC: Heading Track Control (HTC) is the order to the heading controller unit to set values, and control the heading of the vessel.

HTD: Heading Track Control Data (HTD) is the response from heading controller unit containing information about values, modes and current status of the system.

THS: True Heading and Status (THS) provides information about actual vessel heading in degrees.

ROR: Rudder Order Status (ROR) is the order to the rudder controller to control the rudder system of the vessel.

RSA: Rudder Sensor Angle Sentence (RSA) indicates the rudder angle, measured by rudder angle sensor.

Figure 3 illustrates the idea of whole auto-steering system. The orders will be collected from Android platform in NMEA 0183 format and will be translated to NMEA 2000 PGN numbers by NMEA 2000 Stack board.

Figure 3: 
Auto-steering system configuration

The characteristics of each instruments and steps which should be taken in order to perform an auto-steering scenario has briefed as follows:

• 1. Heading Track control commands (HTC) are being produced by Android device in NMEA 0183 sentence format and sent to NMEA 2000 Stack board by serial communication.
• 2. NMEA 2000 Stack board translates the NMEA 0183 Command to appropriate PGN number and sends the PGN packet through NMEA 2000 network.
• 3. The heading commands are being delivered to the heading controller unit by NMEA 2000 Network with PGN 127237.
• 4. The heading controller receives information from heading sensor unit with PGN 127250 and makes rudder angle commands.
• 5. The NMEA 2000 stack board also receives PGN 127250 and interprets them to THS sentences for android system.
• 6. The rudder controller receives rudder commands from heading controller and urges the hydraulic unit to fulfill the orders.
• 7. NMEA 2000 Stack board receives reported data from heading controller and interprets them to NMEA 0183 HTD data to be sent to android platform through serial communication.
• 8. Rudder angle sensor is connected to the rudder controller and rudder angle is converted by ADC and being reported to NMEA 2000 Network with PGN 127245.
• 9. NMEA 2000 Stack board receives the rudder angle sensor reports from NMEA 2000 network, and interprets them to NMEA 0183 RSA Sentences.
• 10. In manual mode, the orders should be sent directly to the rudder controller by android platform. In that case heading controller is bypassed. Therefore, the android platform sends ROR sentences to the NMEA 2000 stack board and that unit translates them to PGN 127245 for NMEA 2000 network.
• 11. The rudder controller receives PGN 127245 commands and urges the hydraulic unit to fulfill the orders.

Basically, Rudder is a vertical blade located at the vessel’s stern and is the main part responsible to control yaw movement (Heading angle) of that vessel [6]. Nowadays the movement of this part is accomplished by a hydraulic actuator connected to the rudder.

For this reason, as mentioned earlier, in order to test the effectiveness of our method, besides computer simulation we developed a hydraulic steering model as it is being shown in Figure 4.

Figure 4: 
Designed vessel hydraulic steering model

Most of the hydraulic cylinders used for vessel steering system are double-acting cylinders. In hydraulic system when orders come to the unit, the pump streams the oil from reservoir to the control valve section. Based on the directional orders from rudder controller to the hydraulic controller unit and because in small boat the hydraulic unit is an on/off system, the right or left side solenoid unit will be energized, and spool inside control valve directs the oil flow to the cylinder. Qiu, et al. [11] described a hydraulic system from mathematical point of view. In that paper it has been stated that the whole steering system can be represented as a third order transfer function for velocity control:

and a fourth order transfer function for position control

where Ω is the rudder angular velocity, φ is the rudder angle, U is input signal, T is time constant for the steering linkage subsystem, ς is damping ratio for the valve subsystem, ω_n is natural frequency for the valve subsystem and K is overall gain. The aforementioned equations for further simplification can be written as a first-order equation for velocity control and a second-order system for position control as follows:

As said before, in small pleasure boat, hydraulic system is on/off by rudder controller, so when the system is off and turns on again, there is a delay time to start set up and pumping again which means each time there will be a delay time which should be considered. In this case the Equation (3) can be rewritten by:

where τ_r is the time delay for the rudder system.

Basically, a vessel has 6 degrees of freedom (6 DOF), including three translational motions (surge, sway and heave) and the other three rotational motions (roll, pitch and yaw) [8][9]. Figure 5 shows the 6 degrees of freedom in a vessel motion.

Figure 5: 
Vessel’s degree of freedom

As Tzeng et al. described the Nomoto model [4], in vessel’s maneuvering problem, only the horizontal motion is considered, which results in considering only surge, sway and heave movements. Therefore we have

Where X,Y and N are surge, sway and yaw motions, m is the mass of the ship, I_z is the moment of inertia to the Z axis, u, v and r are the surge, sway and yaw speed, and u̇,v̇,ṙ are surge, sway and yaw acceleration. In considering only the steering motion, surge motion is separated from steering motion and especially, by neglecting the sway motion in Equation (7) and (8) and by taking the Laplace Transform of yaw and surge with considering a constant speed, a second order system is modeled as

After further simplifying the equation, a 1^st order system is obtained with a time delay due to the process time inside the system which should be considered.

Where τ_h is the time delay of vessel’s heading.

Usually, the heading and rudder controller on vessels are PID controllers due to their simplicity and reliability [5]. In this case in our study also we have used PID controllers with deadband zone. Using the dead-band is to increase error tolerance of the controllers which results in overcoming with chattering phenomena both in rudder actuation system and whole ship in presence of minor errors.

6.2 Controlling inside dead-band Zone

The dead-band or dead-zone is a block which generates zero output within a specific area.

Figure 6: 
Dead-band zone

As mentioned previously, the aim of using dead-band is to decrease the sensitivity of the controllers and avoid chattering phenomena which may occur due to the minor errors between desired orders to the controller and the system’s response. Thus by using dead-band block, the controller will consider a specified range as no error and it will take no action. In spite of advantages which can be achieved by applying the dead-band block, steady state error is the main disadvantage. In autosteering mode, and in long term travel by vessel, due to the steady state error inside the controller, the vessel may change its rout and the controllers consider it as no error.

Figure 7: 
The ship changes its rout due to steady state error

Thus, in order to prevent such problems, a separate controller must be implemented as a supervisory control to control the Vessel’s rudder and heading’s behavior in the deadband Zone. Supervisory controller or Ramadge-Wonham framework is a method in which the behavior of the system is being restricted by a controller. In this work, we implemented a separate integral controller to do the job. This idea has been shown in Figure 8.

Figure 8: 
PID controller with supervisory controller in DB

6.3 Controlling in Manual Steering Mode

In this mode the vessel’s control is performed by mariner’s involvement. In another word, the rudder angle command is ordered by the mariner and rudder control unit operates to fulfill the tasks. From control aspects of view, the idea could be illustrated as Figure 9.

Figure 9: 
Block diagram of manual steering mode

6.4 Controlling in Automatic Steering Mode.

In this mode, the heading controller and rudder controller cooperate together to fulfill the heading orders. The heading degree commands are specified by the mariner. The heading controller unit receives the heading commands and based on the current heading angle, manipulates rudder angle commands automatically to decrease the error between desired heading angle and current heading angle. Heading sensor and rudder degree sensor also send their information to the controllers to indicate the current rudder and heading position. This idea has been represented in Figure 10.

Figure 10: 
Block diagram of automatic steering mode

In order to connect the android platform to the network, we proposed and developed that the android system generate NMEA 0183 sentences and by a translator (NMEA 2000 Stack board) to exchange data between Android platform and NMEA 2000 network. In order to bring the idea to the reality and examine the real performance of the system, before installing it on a real vessel, we developed a model of our work on a hydraulic steering system model and also to test the effectiveness of the proposed controller, we did computer simulation. The proposed controllers were two PID controllers, which one of them is controlling the rudder position and another is controlling the heading angle of that vessel. After computer simulation of the controller, we concluded that in an environment such as water ways, in order to eliminate the fluctuations in rudder system, the error tolerance of the controllers must be increased. To do so, we decided to implement a dead-band block inside the controllers. Nevertheless, the PID controller with dead-band block was not a complete solution, as it had a huge steady state error inside. Thus, the final decision for the controller was 2 PID controllers, each with a dead-band zone and supervisory integral controllers which control the system’s behavior inside the dead-band and reject errors as much as possible. Due to the power saving issues, it was decided that the rudder controller to be an on/off controller. After successful results in computer simulation, we implemented the proposed controller on an embedded system and we did testing on a steering system model described in earlier chapters and after the test on real steering model was successful, the system was installed on a real pleasure boat. The navigation data also were gathered during the test navigation. By comparing sent orders with data collected during the navigation, we concluded that the designed Autosteering system is working properly both in manual and automatic steering modes and it is able to be commercialized.