JP7804271B2 - Steering pattern identification device, control device, steering pattern identification method and program - Google Patents

Description

This disclosure relates to a steering pattern identification device, a control device, a steering pattern identification method, and a program.

Patent Document 1 describes the following ship-steering assistance device. Specifically, the ship-steering assistance device described in this document acquires actual values for ship speed, approach angle, propulsive force, thrust, and steering angle each time a ship approaches or leaves a berth, and stores these values in a database. Furthermore, during the docking or leaving process, the ship-steering assistance device determines the actual value that has the greatest correlation with the propulsive force, thrust, and steering angle values set in response to the operation of the control unit by the ship operator (hereinafter referred to as "operation-dependent values"), and calculates the deviation between the operation-dependent values and the actual value. The ship-steering assistance device then corrects the propulsive force, thrust, and steering angle set in response to the operation of the control unit in accordance with the deviation.

Japanese Patent Application Laid-Open No. 2020-26208

In the ship steering assistance device described in the above prior art document, the propulsion force, thrust, and steering angle set in response to the operator's operation of the control unit are corrected according to deviations from actual values, and then the ship's propulsion system (propulsion source, thruster, and rudder) is controlled. As a result, there is a possibility that the deviations set by the operator's operation of the control unit may fall outside the range of deviations set. In this case, there is a problem in that it may become impossible to keep the ship's state space within the range set by the control unit.

The present disclosure aims to provide a maneuvering pattern identification device, control device, maneuvering pattern identification method, and program that can limit the ship's behavioral space.

To solve the above problem, the maneuvering pattern identification device disclosed herein includes a storage unit that stores a tracking control program that associates port entry/departure route data, including target hull states for the port entry/departure route, and a maneuvering pattern of a steering device; an acquisition unit that acquires the navigation hull state, including the hull state of the hull while sailing; and an identification unit that identifies the maneuvering pattern associated with the navigation hull state based on the tracking control program.

The maneuvering pattern identification method disclosed herein stores a tracking control program that associates port entry/departure route data, including target hull states for the port entry/departure route, with the maneuvering pattern of the steering device; acquires the navigation hull state, including the hull state of the hull while sailing; and identifies the maneuvering pattern associated with the navigation hull state based on the tracking control program.

The program disclosed herein causes a computer to store a tracking control program that associates entry/departure route data, including target hull states for entry/departure routes, with steering patterns for steering devices; acquire the navigation hull state, including the hull state of the hull while sailing; and identify the steering pattern associated with the navigation hull state based on the tracking control program.

The maneuvering pattern identification device, control device, maneuvering pattern identification method, and program disclosed herein can reduce the ship's operating space.

FIG. 1 is a block diagram illustrating an example configuration of a ship according to an embodiment of the present disclosure. FIG. 10 is a plan view schematically illustrating an example of an operation when a vessel according to an embodiment of the present disclosure is docked. FIG. 10 is a plan view schematically illustrating an example of an operation of a ship according to an embodiment of the present disclosure when the ship leaves the quay. FIG. 2 is a schematic diagram illustrating an example of the configuration of port entry/departure route data according to an embodiment of the present disclosure. FIG. 1 is a block diagram for explaining a learning example of a learning model according to an embodiment of the present disclosure. FIG. 10 is a schematic diagram for explaining a learning example of a learning model according to an embodiment of the present disclosure. FIG. 10 is a schematic diagram for explaining a learning example of a learning model according to an embodiment of the present disclosure. FIG. 2 is a schematic diagram for explaining a steering pattern identification unit according to an embodiment of the present disclosure. 10 is a flowchart illustrating an example of the operation of a control device according to an embodiment of the present disclosure. FIG. 1 is a schematic block diagram illustrating a configuration of a computer according to an embodiment of the present disclosure.

First Embodiment
(Configuration examples of ship, control device, and maneuvering pattern identification unit)
A maneuvering pattern identification device, a control device, a maneuvering pattern identification method, and a program according to embodiments of the present disclosure will be described below with reference to FIGS. 1 to 10 . Note that the same or corresponding components in each drawing will be designated by the same reference numerals, and descriptions thereof will be omitted as appropriate. FIG. 1 is a block diagram illustrating an example configuration of a ship 100 according to an embodiment of the present disclosure. FIG. 2 is a plan view schematically illustrating an example operation of a ship 100 according to an embodiment of the present disclosure when docking. FIG. 3 is a plan view schematically illustrating an example operation of a ship 100 according to an embodiment of the present disclosure when leaving docking. FIG. 4 is a schematic diagram illustrating an example configuration of port entry/departure route data 81 according to an embodiment of the present disclosure. FIG. 5 is a block diagram for explaining a learning example of a learning model 83u according to an embodiment of the present disclosure. FIG. 6 is a schematic diagram for explaining a learning example of a learning model 83u according to an embodiment of the present disclosure. FIG. 7 is a schematic diagram for explaining a learning example of a learning model 83u according to an embodiment of the present disclosure. FIG. 8 is a schematic diagram for explaining a maneuvering pattern identification unit 2 according to an embodiment of the present disclosure. FIG. 9 is a flowchart illustrating an example operation of a control device 1 according to an embodiment of the present disclosure. FIG. 10 is a schematic block diagram illustrating the configuration of a computer according to an embodiment of the present disclosure.

The vessel 100 shown in FIG. 1 includes a control device 1, a measurement device 4, a steering device 5, and a propulsion device 6. The control device 1 includes a steering pattern identification unit (steering pattern identification device) 2 and a control unit 3 as functional components formed by a combination of hardware, such as a computer and peripheral devices for the computer, and software, such as a program, executed by the computer. The steering pattern identification unit 2 also includes a storage unit 21, an acquisition unit 22, an identification unit 23, a return route setting unit 24, and a deviation calculation unit 25 as functional components.

The measurement device 4 includes a satellite positioning device 41 and a gyrocompass 42. The satellite positioning device 41 measures the position of the ship 100 using signals emitted from artificial satellites and outputs information indicating the measured position. The gyrocompass 42 measures the heading of the ship 100 and outputs information indicating the measured heading.

The control device 5 includes a joystick 51 and a turning dial 52. The joystick 51 is an input device that has a lever that can be tilted in all directions (360 degrees), including forward, backward, left, and right, and outputs signals corresponding to the tilt direction and angle of the lever. The operator can specify the direction and magnitude of the thrust of the vessel 100 by adjusting the tilt direction and angle of the joystick 51. The turning dial 52 is an input device that has a dial and outputs signals corresponding to the rotation direction and angle of the dial. The operator can specify the turning direction of the vessel 100 by adjusting the rotation direction and angle of the turning dial 52. The control device 5 outputs signals corresponding to command values for the direction and magnitude of thrust of the vessel 100 and the turning moment command value, depending on the operation state of the joystick 51 and the turning dial 52 by the operator. In this embodiment, the information (or signals) determined depending on the operation state of the joystick 51 and the turning dial 52 is referred to as a control pattern. Furthermore, a steering pattern determined by the steering device 5 in response to the actual operation of the operator is referred to as a manual steering pattern, and a steering pattern automatically generated by the steering pattern identification unit 2 separately from the manual steering pattern, as described below, is referred to as an automatic steering pattern.

The propulsion device 6 includes a propeller 61, a thruster 62, and a rudder 63. The propeller 61 converts the power of an engine (not shown) equipped on the vessel 100 into thrust in the longitudinal direction of the vessel 100. In this embodiment, the propeller 61 is a controllable pitch propeller (CPP), and the forward and backward speed (thrust) of the vessel 100 can be adjusted by controlling the propeller blade angle. The thruster 62 converts the power of an engine (not shown) equipped on the vessel 100 into thrust in the lateral direction of the vessel 100. In this embodiment, the thruster 62 is a controllable pitch propeller, and the lateral speed (thrust) of the vessel 100 can be controlled by adjusting the thruster blade angle. The rudder 63 is a marine device used to control the direction of thrust generated by the propeller 61, and the direction of travel of the vessel 100 can be controlled by adjusting the rudder angle.

Meanwhile, in the maneuvering pattern identification unit 2 within the control device 1, the storage unit 21 stores a tracking control program 8. The tracking control program 8 is software for automatically controlling the navigation of the ship 100 when it leaves and arrives at a berth. The tracking control program 8 includes multiple pieces of port entry/departure route data 81, a trained model 83, and a program that the identification unit 23 uses as a reference when executing processing. Furthermore, each piece of port entry/departure route data 81 includes multiple target hull states (data representing them) 82.

Entry/departure route data 81 may be, for example, data indicating the entry route 300 that the ship 100 targets when entering port, as shown in FIG. 2, or data indicating the departure route 400 that the ship 100 targets when departing port, as shown in FIG. 3. Entry/departure route data 81 may be created, for example, under the supervision of a skilled captain or other sailor with operational experience of the ship 100, or may be created based on actual performance data during navigation. The entry route 300 shown in FIG. 2 is a route connecting multiple target points 301. Target point 301s is the starting point of the entry route 300, and is set, for example, to correspond to the entry position within the port 201. Target point 301e is the end point of the entry route 300, and is set, for example, to correspond to the docking position at the quay 200. The departure route 400 shown in FIG. 3 is a route connecting multiple target points 401. Target point 401s is the starting point of the departure route 400, and is set, for example, to correspond to the departure position from quay 200. Target point 401e is the ending point of the departure route 400, and is set, for example, to correspond to the departure position within port 201.

Figure 4 shows an example of the configuration of port entry/departure route data 81. The port entry/departure route data 81 shown in Figure 4 is, for example, data representing the port entry route 300 shown in Figure 2. In this case, the port entry/departure route data 81 shown in Figure 4 includes a time series of target values (referred to as target hull state 82) of the hull state at each target point 301 from target point 301s (number = 0; start point) to target point 301e (number = M; end point). The hull state is information representing the state of the ship 100 (hull), and includes, for example, information representing the hull position and hull heading (heading). Alternatively, the hull state includes information representing the hull position and hull heading, and information representing the hull speed. However, the information representing the hull speed may be replaced with information for calculating the hull speed, such as the travel distance, turning angle, travel time, and time required for turning. Furthermore, the information representing the hull speed included in the hull state may be composed of multiple speed components such as the forward speed, lateral drift speed, and turning angular velocity of the ship 100. Furthermore, the hull state during navigation, which will be described later, may include information representing the deviation of the hull position and hull heading during navigation from the target hull position and hull heading, instead of information representing the hull position and hull heading.

In the example shown in Figure 4, the hull state at each target point 301 is represented by the position (x coordinate value, y coordinate value) of the ship 100 at each target point 301, the orientation of the ship 100 (angle ψ value), and the elapsed time T from the passage of the starting point (target point 301s) corresponding to the passage of the target point 301. In this case, the speed at the target point 301 can be calculated, for example, from the difference in the distance and orientation angle to the previous and next target points 301 and the time difference between the passage times. Note that in the example shown in Figure 4, for example, the time of target point 301s numbered 0 is 0 seconds, the position is x0 and y0, and the orientation is ψ0. Also, for example, the time of target point 301e numbered M is TM seconds, the position is xM and yM, and the orientation is ψM. Note that the coordinate values x and y can be, for example, coordinate values in an XY coordinate system such as a plane rectangular coordinate system, latitude and longitude in a latitude-longitude coordinate system, coordinate values in another local coordinate system, etc.

The trained model 83 is a trained machine learning model that, in response to input hull conditions, outputs a maneuvering pattern related to the input hull conditions. The hull conditions are information corresponding to the hull conditions during actual navigation. In the configuration example shown in Figure 1, the trained model 83 inputs information indicating the deviation between the position and heading of the ship and a target, information indicating the forward speed, lateral drift speed, and turning angular rate included in the hull conditions, and information indicating the operation variables of the propulsion unit 6 (propeller blade angle, thruster blade angle, and rudder angle), and outputs information indicating a maneuvering pattern. The maneuvering pattern is the autopilot pattern described above, and is information simulating information determined based on the operation status of the joystick 51 and the turning dial 52. The trained model 83 is, for example, a trained model that uses a neural network as an element. The weighting coefficients between neurons in each layer of the neural network are optimized by machine learning so that the desired solution is output for a large amount of input data. The trained model 83 is, for example, composed of a program that performs calculations from input to output and a combination of weighting coefficients (parameters) used in the calculations.

The trained model 83 can be trained, for example, using a deep reinforcement learning algorithm. Figure 5 shows the flow of the learning process for the learning model 83u, which is the state of the trained model 83 before (or during) learning. The learning model 83u includes a program that optimizes weighting coefficients related to action values and action strategies in response to externally provided rewards in reinforcement learning. Since reinforcement learning using an actual ship is not practical, a virtual ship 503 is used, which is a model with maneuvering motion characteristics equivalent to those of the ship 100 (control target). The virtual ship 503, for example, inputs and outputs data at each step.

The virtual ship 503 inputs the operation variables calculated by the operation variable calculation unit 502, and calculates and outputs the forward speed u, lateral drift speed v, turning angular rate r, position, and orientation of the ship 100. The operation variables include the propeller blade angle, thruster blade angle, and rudder angle shown in FIG. 1. The forward speed u, lateral drift speed v, and turning angular rate r output by the virtual ship 503 are input to the learning model 83u along with the propeller blade angle, thruster blade angle, and rudder angle. The position and orientation output by the virtual ship 503 are input to the deviation calculation unit 504.

The deviation calculation unit 504 inputs the target position and target heading contained in the port entry/departure route data created, for example, by random maneuvering, and the position and heading output by the virtual ship 503, calculates the deviation (Δx, Δy, Δψ) between the current step and the position and heading for n steps in the future (n is a natural number), and outputs this to the learning model 83u.

The learning model 83u inputs, for example, the forward speed u, lateral drift speed v, turning angular rate r of the virtual ship 503, the control variables (propeller blade angle, thruster blade angle, rudder angle), and the current and future n-step values of the position deviation and azimuth deviation between the ship and the target, and outputs an autopilot pattern.
For example, the input may include historical or track history, which can improve the accuracy of predictions of actual movements.
Note that the autopilot pattern output by the learning model 83u may be limited to K representative cases (K is a natural number) to restrict the behavioral space. As shown in FIG. 6 , for example, if the operation state of the joystick 51 is limited to two tilting states in eight directions: stop A0, forward left diagonal A1, forward A2, forward right diagonal A3, left A4, right A5, rear left diagonal A6, rear A7, and rear right diagonal A8, and the operation state of the turning dial 52 is limited to two operation states: left A11 and right A12, the operation state of the joystick 51 will be 17 cases and the operation state of the turning dial 52 will be four cases. The learning model 83u selects one of these operation states and outputs it as the autopilot pattern. The number of cases, K, is set appropriately for each ship or port.

The operation amount calculation unit 502 inputs the autopilot pattern output by the learning model 83u, calculates the operation amount, and outputs it to the virtual ship 503.

On the other hand, the reward calculation unit 501 calculates a reward based on, for example, the deviation between the position and orientation of the current step, and outputs it to the learning model 83u. For example, as shown in FIG. 7, the reward calculation unit 501 outputs a reward RE1 when the position deviation is smaller than PD1 and smaller than ±DD1, outputs a reward RE2 when the position deviation is smaller than PD2 and smaller than ±DD2, and outputs a reward RE3 when the position deviation is smaller than PD3 and smaller than ±DD3. Note that PD1<PD2<PD3, DD1<DD2<DD3, RE1>RE2>RE3.

In the reinforcement learning of learning model 83u, the conditions for ending an episode (trial) are set to (1) to (3) below in order to improve calculation efficiency.

(1) The position deviation is equal to or greater than a predetermined first threshold value.
(2) The absolute value of the azimuth deviation is equal to or greater than a predetermined second threshold value.
(3) The number of steps is equal to or greater than a predetermined third threshold.

When the termination conditions (1) and (2) are set, if the position deviation is equal to or greater than the first threshold value or the absolute value of the orientation deviation is equal to or greater than the second threshold value, the learning model 83u (and the trained model 83) will be given an input that deviates from the learning range.

Returning to Figure 1, the acquisition unit 22 acquires information representing the navigating hull state, including the hull state, of the ship 100 (hull) while sailing. The acquisition unit 22 acquires information indicating the position output by the satellite positioning device 41 and information indicating the orientation measured by the gyrocompass 42 as information representing the navigating hull state. Then, based on the acquired information indicating the position and orientation, the acquisition unit 22 outputs the position and orientation included in the navigating hull state to the deviation calculation unit 25, and calculates the forward speed, lateral drift speed, and turning angular velocity included in the navigating hull state, and outputs them to the trained model 83.

The deviation calculation unit 25 refers to the port entry/departure route data 81, calculates the deviation between the position and heading included in the navigating hull state and the target position and target heading included in the port entry/departure route data 81, and outputs this to the trained model 83. In doing so, the deviation calculation unit 25 calculates the deviation between the target position and target heading from the present to n steps into the future in the port entry/departure route data 81 from the position and heading of the ship 100 currently sailing, and outputs this to the trained model 83. In addition, when the return route setting unit 24 sets a return route as described below, the deviation calculation unit 25 calculates the deviation between the current and future positions and headings using the set return route as the target position and target heading, and outputs this to the trained model 83.

If there is a possibility that the navigating hull state will deviate from the port entry/departure route data 81, the return route setting unit 24 sets a route to return to within the port entry/departure route data 81. When the learned model 83 is trained under the termination conditions (1) and (2) described above, the conditions for the learning range are that the position deviation is less than the first threshold and the absolute value of the heading deviation is less than the second threshold. Therefore, if there is a possibility that the navigating hull state (position and heading) will deviate from the port entry/departure route data 81 outside the learning range, the return route setting unit 24 sets a return route to return to the port entry/departure route data 81, sets a target position and target heading based on the set return route, and prevents the navigating hull state from deviating outside the learning range. The return route setting unit 24 obtains, for example, information indicating the deviation between position and heading and the port entry/departure route data 81 from the deviation calculation unit 25, and if the position deviation becomes equal to or greater than a predetermined first L threshold value that is smaller than the first threshold value, or if the absolute value of the heading deviation becomes equal to or greater than a predetermined second L threshold value that is smaller than the second threshold value, sets a route starting from the current position and heading to return to the port entry/departure route data 81 so that the deviation between position and heading is reduced.

Figure 8 schematically shows an example of setting a return route. The port entry route 300 shown in Figure 8 includes a target value for the hull condition at target point 301a and a target value for the hull condition at target point 301b. Note that target point 301b is a target point that has already been passed, and target point 301a is a target point that is about to be passed. One or more other target points may be set between target point 301b and target point 301a. When the position deviation (distance) pd1 from the current port entry route 300 to the center of gravity (reference point) 101 of the ship 100 becomes equal to or greater than the first L threshold, the return route setting unit 24 calculates a route that returns to target point 301a from the current center of gravity (reference point) 101 as the starting point, and sets it as the return route wr1. In this case, the return route setting unit 24 may set the return route wr1 so that the speed on the return route wr1 is slower than the speed set for the port entry route 300 (the passage time may be adjusted). In addition, the return route setting unit 24 may set the return route wr1 so that the turning angular velocity is slower than the angular velocity set for the port entry route 300 when the turning angle is restored.

The identification unit 23 shown in Figure 1 also identifies an autopilot pattern, which is a maneuvering pattern related to the state of the navigating hull, based on the tracking control program 8. The identification unit 23 inputs information indicating the deviation between position and heading, information indicating the state of the navigating hull (forward speed, lateral drift speed, and turning angular rate), and information indicating the propeller blade angle, thruster blade angle, and rudder angle to the trained model 83, and identifies the maneuvering pattern output from the trained model 83 as the autopilot pattern to be input to the control unit 3.

The control unit 3 shown in FIG. 1 also controls the propulsion device 6 of the ship 100 (hull) according to the manual steering pattern output by the steering device 5 or the autopilot pattern identified by the steering pattern identification unit 2. The control unit 3 calculates and outputs the propeller blade angle, thruster blade angle, and rudder angle, for example, according to the manual steering pattern when sailing outside the port, and according to the autopilot pattern when an instruction to execute autopilot is given when docking or leaving the berth. Information indicating the propeller blade angle, thruster blade angle, and rudder angle output by the control unit 3 is input to the propulsion device 6 and the trained model 83.

(Example of operation of the ship, the control device, and the maneuvering pattern identification unit)
Next, an example of the operation of the control device 1 and the maneuvering pattern identification unit 2 will be described with reference to Fig. 9. For example, when an operator performs a predetermined operation on an input/output device (not shown) of the control device 1 to instruct the start of automatic maneuvering, the control device 1 repeatedly executes the process shown in Fig. 9 at a predetermined cycle until the end point of the port entry/departure route data 81 is reached.

In the process shown in Figure 9, first, the acquisition unit 22 acquires the state of the navigating vessel (step S1). Next, the deviation calculation unit 25 calculates the deviation of the position and heading from the port entry/departure route data 81 (step S2). Next, the return route setting unit 24 determines whether the position deviation or heading deviation is outside a predetermined range (step S3).

If there is a deviation (step S3: YES), the return route setting unit 24 sets a route to return to within the port entry/departure route data 81 (step S4). Next, the deviation calculation unit 25 calculates the deviation of the position and heading from the return route and the port entry/departure route data 81 (step S5).

If there is no deviation (step S3: NO), or after processing in step S5, the identification unit 23 inputs the state of the navigating hull, the operation amount of the propulsion device 6, and the deviations in position and heading into the trained model 83, and identifies the operation pattern of the steering device 5 based on the output of the trained model 83 (step S6). Next, the control unit 3 determines the operation amount of the propulsion device 6 based on the identified operation pattern, and controls the propulsion device 6 based on the determined operation amount of the propulsion device 6 (step S7).

(Action and effect)
The maneuvering pattern identification device, control device, maneuvering pattern identification method, and program of this embodiment store a tracking control program 8 in which port entry/departure route data 81 including a target hull state for the port entry/departure route and the maneuvering pattern of the steering device 5 are associated, acquire a navigation hull state including the hull state of the hull of the ship 100 while sailing, and identify a maneuvering pattern related to the navigation hull state based on the tracking control program 8. Therefore, according to the maneuvering pattern identification device, control device, maneuvering pattern identification method, and program of this embodiment, it is possible to restrict the behavioral space by identifying the maneuvering pattern of the steering device 5.

In addition, the maneuvering pattern identification device of this embodiment further includes a return route setting unit 24 that sets a route to return to within the port entry/departure route data 81 when there is a possibility that the navigational state of the vessel may deviate from the port entry/departure route data 81. By returning to the route, it becomes possible to maneuver the vessel within the learned state space, for example.

In recent years, there has been a growing need for autonomous navigation on ships, driven by factors such as improved safety, reduced crew burdens, and a declining number of experienced crew members. While a wide range of technologies are required for autonomous navigation, accurate route tracking is an important technology, along with collision avoidance. In particular, maneuvering ships in narrow spaces and harbors, and maneuvering ships while docking and leaving the berth, requires a highly accurate and robust route tracking system that is resistant to external disturbances.

In this embodiment, the above is achieved by applying AI (Artificial Intelligence) technology based on deep reinforcement learning to port entry/departure route tracking, and has the following features:

[1] The AI automatically controls the propulsion system by selecting the control pattern of the joystick and steering dial.

[2] It has a course correction function that keeps the ship's position and heading within the AI's learning range.

In this embodiment, for example, 1) port entry/departure route data is created in advance by an experienced captain or the like, consisting of the ship's position, heading, and speed at the time of approaching and leaving the quay when entering or leaving port. It may also be created using other methods.

2) Ships are also equipped with steering devices such as a propeller propulsion unit, a rudder, and thrusters for lateral movement, as well as a steering device that controls these devices in an integrated manner. These are common devices.

3) In addition to the above, we will create a tracking control program that uses AI (trained model) created using deep reinforcement learning techniques to select control patterns for the joystick and turning dial.

4) In addition, the ship's propulsion device configured in 2) is operated along the entry and exit route created in 1) by activating the AI tracking control program created in 3), thereby automatically maneuvering the ship when entering and leaving port.

5) Furthermore, with this system, the route itself is created by an experienced captain and using reliable route planning methods, ensuring a safe and practical route. By using AI to automate the detailed actuator control along that route, it is possible to provide powerful, redundant control against disturbances such as wind and waves.

6) Furthermore, for fine control, an AI program is used to issue operating values to joysticks and other devices that operate the steering devices in an integrated manner, and to steer the ship in a way that gradually eliminates deviations from the set route. If, for some reason, the ship's deviation from its route deviates from the AI's learning range, the AI program is equipped with a mechanism that forcibly rewrites part of the route to keep it within the learning range. In other words, with a mechanism for returning route deviations to within the learning range (route resetting), for example, if the controlled object deviates from the learning range due to an unexpected factor, the state at the time of deviation is set as the initial value, and a route that smoothly connects to the original route is reset, thereby keeping the ship's position and heading within the AI's learning range.

(Other embodiments)
The above describes in detail the embodiments of the present disclosure with reference to the drawings, but the specific configuration is not limited to this embodiment, and design changes and the like are also included within the scope that does not deviate from the gist of the present disclosure.

For example, in the above embodiment, the propulsion device 6 is equipped with one set of propeller 61, thruster 62, and rudder 63, but this combination is not limited. For example, the propulsion device 6 may be equipped with two propellers, two bow thrusters, two stern thrusters, two rudders, etc. Furthermore, in the example described with reference to FIG. 4, the target hull state 82 included in the port entry/departure route data 81 includes information representing the hull's position, heading, and elapsed time from the starting point (the time of passing the target point), but this is not limited to this. For example, the target hull state 82 may include information representing velocities such as forward speed, drift speed, and turning angular velocity in addition to the position, heading, and time of passing at the target point. Furthermore, the steering device 5 may be configured with, for example, a single joystick with a lever equipped with a dial function. Alternatively, the steering device 5 may be equipped with one or more dials, sliders, etc. instead of a joystick. Furthermore, the steering device 5 may not be equipped with a joystick or dial for manual steering.

In one example of this embodiment, a route return algorithm is added, such that the return route setting unit 24 returns to the entry/departure route data 81, but as a variant, a time adjustment algorithm may be added, such as adjusting the set time for reaching the target point.
For example, the time elapsed at the future target point may simply be advanced or delayed.
For example, if the ship 100 is delayed relative to the elapsed time of the future target point, the elapsed time of the target point may be delayed.
For example, if the vessel 100 is moving forward relative to the elapsed time of the future target point, the elapsed time of the target point may be advanced.
According to this modified example, the elapsed time is originally set as information for each arrival point on the set route, but if the deviation from the elapsed time becomes too large, it becomes possible to deal with the case where the control range is exceeded "in terms of time."

<Computer Configuration>
FIG. 10 is a schematic block diagram illustrating the configuration of a computer according to at least one embodiment.
The computer 90 includes a processor 91 , a main memory 92 , a storage 93 , and an interface 94 .
The above-described control device 1 (the steering pattern identification unit 2 and the control unit 3) is implemented in a computer 90. The operations of the above-described processing units are stored in the storage 93 in the form of a program. The processor 91 reads the program from the storage 93, loads it into the main memory 92, and executes the above-described processing in accordance with the program. The processor 91 also allocates storage areas in the main memory 92 corresponding to the above-described storage units in accordance with the program.

The program may be for realizing some of the functions to be performed by the computer 90. For example, the program may be combined with other programs already stored in storage or implemented in other devices to achieve the functions. In other embodiments, the computer may include a custom LSI (Large Scale Integrated Circuit) such as a PLD (Programmable Logic Device) in addition to or instead of the above configuration. Examples of PLDs include PAL (Programmable Array Logic), GAL (Generic Array Logic), CPLD (Complex Programmable Logic Device), and FPGA (Field Programmable Gate Array). In this case, some or all of the functions realized by the processor may be realized by the integrated circuit.

Examples of storage 93 include HDD (Hard Disk Drive), SSD (Solid State Drive), magnetic disk, magneto-optical disk, CD-ROM (Compact Disc Read Only Memory), DVD-ROM (Digital Versatile Disc Read Only Memory), semiconductor memory, etc. Storage 93 may be internal media directly connected to the bus of computer 90, or external media connected to computer 90 via interface 94 or a communication line. Furthermore, if this program is distributed to computer 90 via a communication line, the computer 90 that receives the program may load the program into main memory 92 and execute the above-mentioned processing. In at least one embodiment, storage 93 is a non-transitory, tangible storage medium.

<Additional Notes>
The steering pattern identification device (steering pattern identification unit 2) described in the above embodiment can be understood, for example, as follows.

(1) The maneuvering pattern identification device (maneuvering pattern identification unit 2) according to the first aspect includes a storage unit 21 that stores a tracking control program 8 associated with port entry/departure route data 81 including target hull states for the port entry/departure route and the maneuvering pattern of the steering device 5, an acquisition unit 22 that acquires a navigation hull state including the hull state of the hull (ship 100) currently sailing, and an identification unit 23 that identifies the maneuvering pattern associated with the navigation hull state based on the tracking control program 8. According to this aspect and each of the following aspects, the maneuvering pattern identification device (maneuvering pattern identification unit 2) can restrict the behavioral space by identifying the maneuvering pattern of the steering device 5.

(2) A steering pattern identification device (steering pattern identification unit 2) according to a second aspect is the steering pattern identification device (steering pattern identification unit 2) of (1), in which the steering device 5 includes a joystick 51 or a dial (a steering dial 52).

(3) A steering pattern identification device (steering pattern identification unit 2) according to a third aspect is the steering pattern identification device (steering pattern identification unit 2) of (1) or (2), in which the tracking control program 8 includes a trained model 83 that, in response to an input of the navigating hull state, outputs the steering pattern associated with the input navigating hull state.

(4) A maneuvering pattern identification device (maneuvering pattern identification unit 2) according to a fourth aspect is a maneuvering pattern identification device (maneuvering pattern identification unit 2) according to any one of (1) to (3), and further includes a return route setting unit 24 that sets a route to return to within the port entry/departure route data 81 when there is a possibility that the navigation hull state will deviate from the port entry/departure route data 81. According to this aspect, returning to the route makes it possible to maneuver the ship within the learned state space, for example.

(5) A steering pattern identification device (steering pattern identification unit 2) according to a fifth aspect is a steering pattern identification device (steering pattern identification unit 2) according to any one of (1) to (4), in which the navigating hull state includes the hull position, the bow heading, and the hull speed.

(6) A control device 1 according to a sixth aspect includes a maneuvering pattern identification device (maneuvering pattern identification unit 2) according to any one of (1) to (5), and a control unit 3 that controls the propulsion device 6 of the hull (ship 100) in accordance with the identified maneuvering pattern. According to this aspect, the control device 1 can restrict the behavioral space by identifying the maneuvering pattern of the maneuvering device 5.

(7) A maneuvering pattern identification method according to a seventh aspect stores a tracking control program 8 in which port entry/departure route data 81 including target hull states for the port entry/departure route and a maneuvering pattern of a steering device 5 are associated, acquires a navigation hull state including the hull state of the hull (ship 100) currently sailing, and identifies the maneuvering pattern associated with the navigation hull state based on the tracking control program 8. According to this aspect, the maneuvering pattern identification method can restrict the behavioral space by identifying the maneuvering pattern of the steering device 5.

(8) The program according to the eighth aspect stores a tracking control program 8 that associates port entry/departure route data 81, including target hull states for the port entry/departure route, with the steering pattern of the steering device 5, and acquires the navigation hull state, including the hull state of the hull (ship 100) currently sailing, and identifies the steering pattern associated with the navigation hull state based on the tracking control program 8. According to this aspect, the program can restrict the behavioral space by identifying the steering pattern of the steering device 5.

1... control device 2... operation pattern identification unit (operation pattern identification device)
3...Control unit 4...Measuring device 5...Steering device 6...Propulsion device 8...Tracking control program 21...Storage unit 22...Acquisition unit 23...Identification unit 24...Return route setting unit 25...Deviation calculation unit 41...Satellite positioning device 42...Gyrocompass 51...Joystick 52...Turning dial (dial)
61... Propeller 62... Thruster 63... Rudder 81... Entry/Departure route data 82... Target hull state 83... Learned model 83u... Learning model 90... Computer 91... Processor 92... Main memory 93... Storage 94... Interface 100... Ship (hull)
101... Center of gravity (reference point)
200...quay 201...port 300...entry route 301...target point 301a...target point 301b...target point 301e...target point 301s...target point 400...departure route 401...target point 401e...target point 401s...target point 501...reward calculation unit 502...operation amount calculation unit 503...virtual ship 504...deviation calculation unit A0...stop A1...left diagonally forward A2...forward A3...right diagonally forward A4...left A5...right A6...left diagonally backward A7...back A8...right diagonally backward A11...left direction A12...right direction pd1...position deviation (distance)
WR1...Return route

Claims (7)

a storage unit for storing a tracking control program in which port entry/departure route data including a target hull state for the port entry/departure route and a steering pattern of a steering device are associated with each other;
an acquisition unit for acquiring a hull state during navigation, including a hull state of the hull during navigation;
an identification unit that identifies the maneuvering pattern associated with the state of the navigating vessel based on the tracking control program;
Equipped with
The tracking control program includes a trained model that outputs, in response to an input of the hull state, the steering pattern associated with the input hull state.
Maneuvering pattern identification device. The steering pattern identification device according to claim 1 , wherein the steering device includes a joystick or a dial. The maneuvering pattern identification device of claim 1 or 2 further comprises a return route setting unit that sets a route to return to within the port entry/departure route data when there is a possibility that the navigating vessel state will deviate from the port entry/departure route data. The maneuvering pattern identification device according to any one of claims 1 to 3, wherein the navigation hull state includes a hull position, a hull heading, and a hull speed. The steering pattern identification device according to any one of claims 1 to 4,
a control unit that controls a propulsion device of the hull in accordance with the identified maneuvering pattern;
A control device comprising: a step of storing a tracking control program in which port entry/departure route data including a target hull state for the port entry/departure route and a steering pattern of a steering device are associated with each other;
acquiring a sailing hull state including a hull state of the hull during sailing;
Identifying the steering pattern associated with the navigating vessel state based on the tracking control program;
A method for identifying a steering pattern, comprising:
the tracking control program stored in the step of storing the tracking control program includes a trained model that outputs, in response to an input of the hull state, the steering pattern associated with the input hull state;
A method for identifying maneuvering patterns. On the computer,
a step of storing a tracking control program in which port entry/departure route data including a target hull state for the port entry/departure route and a steering pattern of a steering device are associated with each other;
acquiring a sailing hull state including a hull state of the hull during sailing;
Identifying the steering pattern associated with the navigating vessel state based on the tracking control program;
A program for executing
the tracking control program stored in the step of storing the tracking control program includes a trained model that outputs, in response to an input of the hull state, the steering pattern associated with the input hull state;
program.