KR102560693B1 - Apparatus and method for determining return time of ship - Google Patents

Abstract

An apparatus and method for determining a ship return time point are disclosed. A method for determining a ship return time point according to an embodiment includes inputting state data and external factor data of a ship at each of a plurality of points in time as learning data, and reinforcing and learning a return time point determination model based on compensation according to whether or not the return of the ship is successful; Receiving the ship's state data and external factor data at the current time point; and determining whether to return the current ship by inputting ship state data and external factor data at the current point in time into the learned return point determination model.

Description

Apparatus and method for determining ship return time {APPARATUS AND METHOD FOR DETERMINING RETURN TIME OF SHIP}

The disclosed embodiments relate to techniques for determining a ship return point.

Recently, as the unmanned vehicle industry develops, safe and highly usable unmanned technology is in the limelight. An unmanned mobile body is a body that moves by itself by recognizing an external environment, and may include an unmanned aerial vehicle, a car, a ship, and the like. These aircraft are studied considering air, land and sea characteristics. Among them, the return point of a ship operating in the sea should be determined in particular by considering external factors that affect navigation, such as the size of waves, the direction of waves, and the strength and direction of wind. This is attributable to the fact that, compared to land where refueling is relatively easy or in the air where it is possible to return quickly, in the case of sea, there is a restriction to fly to a specific area for refueling.

In this regard, there is a technique for determining the return point of the ship using an operator operating the ship or a rule-based decision-making system. However, there is a problem in that a number of experiences of the operator are required in advance because the method for determining the return time by the operator requires expertise. On the other hand, in the case of a rule-based decision-making system, there are many environmental factors to consider, so there is a problem of low reliability because accurate values cannot be measured. Therefore, there is a need for a return point determination model that determines the return point with high reliability without human evaluation.

Republic of Korea Patent Registration No. 10-2078488 (registered on February 11, 2020)

Disclosed embodiments are to provide an apparatus and method for determining a vessel return point.

A method for determining a ship return time point according to an embodiment includes inputting state data and external factor data of a ship at each of a plurality of points in time as learning data, and reinforcing and learning a return time point determination model based on compensation according to whether the return of the ship is successful; Receiving the ship's state data and external factor data at the current time point; and determining whether to return the current ship by inputting ship state data and external factor data at the current point in time into the learned return point determination model.

The ship's state data at each of the plurality of points of time or the current point of time may include data on at least one of the position of the ship, operating speed, self-determination of defects, a point on a route to be reached by the ship, and a return point measured at corresponding points in time, and the external factor data at each point of time or the current point of time may include data on at least one of meteorological conditions measured at corresponding points in time, whether an obstacle has been detected, whether an enemy has been detected, and the amount of remaining fuel of the ship.

The return time determination model may, in a learning process, determine an action guide to be applied to the ship from among a plurality of preset action guidelines based on the learning data, determine whether the ship returns successfully by operating the ship based on the determined action guide, calculate the compensation based on whether the ship returns successfully, and in a reasoning process, determine an action guide to be applied to the ship from among the plurality of action guides based on the state data of the ship and external factor data at the current time point.

The return time determination model, in the learning process, when the ship has successfully returned as a result of operating the ship, calculates the compensation as a positive value, and when the ship fails to return, calculates the compensation as a negative value.

The plurality of action guidelines include a mission continuation guideline and a return guideline, and the return point determining model moves the position of the ship to the next point when the determined action guideline is the mission continuation guideline, and returns the position of the ship to a return point when the determined action guideline is the return guideline.

In the return point determination model, when the ship's position measured at a specific time point does not pass a turning point on a preset route, the ship is driven from the ship's position measured at the specific point in time to a preset return point passing through the turning point on the route. If the amount of fuel expected to be consumed by the ship is less than or equal to the remaining fuel amount at the specific time point, the ship can be moved to the next point based on the mission continuation guideline.

The return time determination model, when the amount of fuel expected to be consumed by the ship to navigate from the location of the ship measured at a specific point in time to a preset return point is equal to or greater than the amount of remaining fuel at the specific point in time, the ship can be returned based on the return guideline.

The return time determination model, when an external shock event by an obstacle or an enemy is detected based on the external factor state data at the current time point, the ship is operated based on the return guideline, and when the external shock event is not detected, the ship can be moved to the next point based on the mission continuation guideline.

In the receiving step, when the determined action guideline is the mission continuation guideline, the ship's state data and external factor data are re-received at a point in time when a predetermined unit time has elapsed from the current point in time, and in the determining step, whether to return the ship at the updated current point in time may be determined again by re-entering the re-received ship state data and external factor data into the return point determination model and updating the elapsed point in time to the current point in time.

A computing device according to an embodiment includes one or more processors; Memory; and one or more programs, the one or more programs being stored in the memory and configured to be executed by the one or more processors, the one or more programs inputting state data and external factor data of the ship at each of a plurality of points in time as learning data, and reinforcing and learning a return time point determination model based on compensation according to whether or not the return of the ship is successful; a command for receiving the state data of the vessel and the external factor data at the current point in time; and a command for determining whether to return the current ship by inputting ship state data and external factor data at the current point in time into the learned return point determination model.

The ship's state data at each of the plurality of points of time or the current point of time may include data on at least one of the position of the ship, operating speed, self-determination of defects, a point on a route to which the ship should reach, and a return point measured at a corresponding point of time, and the external factor data at each of the plurality of points of time or the current point of time may include data on at least one of meteorological conditions, whether an obstacle is detected, whether an enemy is detected, and the amount of remaining fuel of the ship, measured at a corresponding point in time.

The return time determination model may, in a learning process, determine an action guide to be applied to the ship from among a plurality of preset action guidelines based on the learning data, determine whether the ship returns successfully by operating the ship based on the determined action guide, calculate the compensation based on whether the ship returns successfully, and in a reasoning process, determine an action guide to be applied to the ship from among the plurality of action guides based on the state data of the ship and external factor data at the current time point.

The return time determination model, in the learning process, when the ship has successfully returned as a result of operating the ship, calculates the compensation as a positive value, and when the ship fails to return, calculates the compensation as a negative value.

The plurality of action guidelines include a mission continuation guideline and a return guideline, and the return point determining model moves the position of the ship to the next point when the determined action guideline is the mission continuation guideline, and returns the position of the ship to a return point when the determined action guideline is the return guideline.

In the return point determination model, when the ship's position measured at a specific time point does not pass a turning point on a preset route, the ship can be moved to the next point based on the mission continuation guideline if the estimated fuel consumption of the ship is less than or equal to the remaining fuel amount at the specific time point in order to navigate from the ship's position measured at the specific time point to the preset return point through the turning point on the route.

The return time determination model, when the amount of fuel expected to be consumed by the ship to navigate from the location of the ship measured at a specific point in time to a preset return point is equal to or greater than the amount of remaining fuel at the specific point in time, the ship can be returned based on the return guideline.

The return time determination model, when an external shock event by an obstacle or an enemy is detected based on the external factor state data at the current time point, the ship is operated based on the return guideline, and when the external shock event is not detected, the ship can be moved to the next point based on the mission continuation guideline.

The one or more programs may include, when the determined action guideline is the mission continuation guideline, a command for re-receiving ship state data and external factor data at a time point when a preset unit time has elapsed from the current time point; and re-inputting the re-received ship state data and external factor data into the return time determination model, updating the elapsed time point to the current time point, and determining again whether to return the ship at the updated current time point. It may further include a command.

According to the disclosed embodiments, it is possible to improve the accuracy and reliability of prediction of a return time by calculating the return time through a reinforcement-learned return time determination model without performing a calculation considering the correlation between factors influencing the return time in a complex manner.

In addition, according to the disclosed embodiments, an optimal return time point may be determined by inputting data related to a factor that fluctuates in real time.

In addition, according to the disclosed embodiments, even if a factor influencing the return point is added or changed, the return point determination model does not need to be newly designed, so that the return point determination model can be improved in compatibility.

1 is an exemplary diagram for comparing a method for determining a ship return time point according to the prior art and an embodiment
2 is a block diagram illustrating and describing a computing environment including a computing device suitable for use in example embodiments;
3 is a flowchart for explaining a method for determining a ship return time point according to an embodiment
4 is an exemplary view for explaining a ship return time determination model according to an embodiment
5 is a flowchart for explaining a method for determining a ship return time point according to an additional embodiment

Hereinafter, specific embodiments of an embodiment will be described with reference to the drawings. The detailed descriptions that follow are provided to provide a comprehensive understanding of the methods, devices and/or systems described herein. However, this is only an example and the present invention is not limited thereto.

In describing one embodiment, if it is determined that a detailed description of a known technology related to the present invention may unnecessarily obscure the subject matter of an embodiment, the detailed description will be omitted. In addition, terms to be described later are terms defined in consideration of functions in the present invention, which may vary according to the intention or custom of a user or operator. Therefore, the definition should be made based on the contents throughout this specification. Terminology used in the detailed description is for describing only one embodiment and should in no way be limiting. Unless expressly used otherwise, singular forms of expression include plural forms. In this description, expressions such as “comprising” or “comprising” are intended to indicate certain characteristics, numbers, steps, operations, elements, parts or combinations thereof, and should not be construed to exclude the existence or possibility of one or more other characteristics, numbers, steps, operations, elements, parts or combinations thereof other than those described.

1 is an exemplary diagram for comparing a ship return time point determination device according to an embodiment with the prior art.

The trajectory of a ship operating along a designated route from the departure point to the return point may be as shown in Figure (1). However, Figure (1) shows the case where the amount of fuel held by the ship is more than the amount of fuel required to navigate all the set routes.

Referring to Figure (2), the navigation trajectory of a ship that failed to return is shown even though the return point has been determined according to a conventional technique for determining the return point of a ship including an operator or a rule-based decision-making system.

Specifically, the conventional technology for determining the return time of the ship determines that the amount of remaining fuel of the ship to be consumed from point A (23) to the return point is more than the amount of remaining fuel held by the ship, and determines to continue the operation of the ship.

Then, in the conventional technology for determining the return time of the ship, it is determined that the amount of fuel to be consumed from point B 21 to the return point is less than the remaining amount of fuel, and the return of the ship is determined only when the point B 21 is reached. However, Figure (2) misjudges the amount of fuel consumed up to the point of return and fails to reach the point of return due to insufficient amount of fuel. As such, in the prior art, there are many cases in which a return point is not reached by determining an erroneous judgment due to various external factors. That is, the prior art has a problem in that it does not efficiently operate the ship, such as preventing the ship from returning to the return point by determining an inappropriate return point or returning the ship to the return point during navigation even though it can sufficiently navigate the set route.

However, as shown in Figure (3), when the ship's return point is determined using the reinforcement learning-based return point determination model, the return point is determined by learning the influence of various external factors on the ship's operation. Therefore, it is possible to determine an appropriate return point rather than the conventional ship return point determination technology.

For example, in the case of the prior art in Figure (2), the return point is determined only when the ship passes point A (23), whereas in the same situation, the method of determining the return point using the reinforcement learning-based return point determination model of Figure (3) can determine the return point as the return point when passing point B (21) even in the same situation as in Figure (2), thereby allowing the ship to return successfully. After all, if the ship's return time is determined using the reinforcement learning-based return time determination model, it has the advantage of being able to perform more requirements compared to the amount of remaining fuel.

2 is a block diagram illustrating and illustrating a computing environment 10 including a computing device suitable for use in example embodiments. In the illustrated embodiment, each component may have different functions and capabilities other than those described below, and may include additional components other than those described below.

The illustrated computing environment 10 includes a computing device 12 that determines when to return the vessel to the device. In one embodiment, computing device 12 may be a device for performing image processing according to an embodiment of the present invention.

Computing device 12 includes at least one processor 14 , a computer readable storage medium 16 and a communication bus 18 . Processor 14 may cause computing device 12 to operate according to the above-mentioned example embodiments. For example, processor 14 may execute one or more programs stored on computer readable storage medium 16 . The one or more programs may include one or more computer-executable instructions, which when executed by processor 14 may be configured to cause computing device 12 to perform operations in accordance with an example embodiment.

Computer-readable storage medium 16 is configured to store computer-executable instructions or program code, program data, and/or other suitable form of information. Program 20 stored on computer readable storage medium 16 includes a set of instructions executable by processor 14 . In one embodiment, computer-readable storage medium 16 may be memory (volatile memory such as random access memory, non-volatile memory, or a suitable combination thereof), one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, other forms of storage medium that can be accessed by computing device 12 and store desired information, or any suitable combination thereof.

Communications bus 18 interconnects various other components of computing device 12, including processor 14 and computer-readable storage medium 16.

Computing device 12 may also include one or more input/output interfaces 22 and one or more network communication interfaces 26 that provide interfaces for one or more input/output devices 24 . An input/output interface 22 and a network communication interface 26 are connected to the communication bus 18 . Input/output device 24 may be coupled to other components of computing device 12 via input/output interface 22 . Exemplary input/output devices 24 may include input devices such as pointing devices (such as a mouse or trackpad), keyboards, touch input devices (such as a touchpad or touchscreen), voice or audio input devices, sensor devices of various kinds, and/or imaging devices, and/or output devices such as display devices, printers, speakers, and/or network cards. The exemplary input/output device 24 may be included inside the computing device 12 as a component constituting the computing device 12, or as a separate device distinct from the computing device 12, and may be connected to the computing device 12 for determining the return time of the ship.

3 is a flowchart illustrating a method for determining a ship return time point according to an exemplary embodiment.

The method shown in FIG. 3 may be performed by the computing device 12 for determining when to return to the vessel. To this end, the method for determining the return time of the ship may be implemented in the form of a program or software including one or more computer executable instructions and stored on the memory.

Referring to FIG. 3 , a process of determining a return time point by the return point determination model 410 may be explained.

The computing device 12 inputs the ship's state data and external factor data at each of a plurality of points in time as learning data, and reinforces the return time point determination model 410 based on compensation according to whether or not the ship returns successfully (310).

Thereafter, the computing device 12 receives state data and external factor data of the ship at the current point in time (320).

Thereafter, the computing device 12 determines whether to return the current ship by inputting the state data and external factor data of the ship at the current point in time to the learned return point determination model 410 (330).

According to one embodiment, the state data of the ship at each of a plurality of points in time or at the current point in time may include data on at least one of the location of the ship measured at the corresponding point in time, navigation speed, self-determination of defects, points on the route to which the ship should reach, and return points.

The location of the ship may be a location that becomes a reference point for determining whether the ship is to be returned by the return time determination model 410 .

Sailing speed may mean the sailing speed of the ship recorded until the time of return is determined. The navigation speed may include at least one of an average speed, a minimum speed, and a maximum speed of ships that have previously operated until it is determined whether or not the ship returns.

The self-determination of a defect is to determine whether or not there is a defect in the ship when determining the time of return, and may include at least one of navigation capability, sensor and armament abnormality.

The route that the ship should reach may mean a preset route that the ship should navigate. At this time, the route to be reached by the ship may be set through a starting point, a return point, and a return point.

The return point may refer to a point to which the ship must finally return even if it performs or fails to perform its mission.

On the other hand, the ship's position and return point can be expressed through a coordinate system. At this time, the coordinate system may be, for example, a coordinate system such as a 1D coordinate system, a 2D coordinate system, a 3D coordinate system, a polar coordinate system, a cylindrical coordinate system, a spherical coordinate system, a homogeneous coordinate system, but is not necessarily limited to a specific coordinate system. In addition, it can be any one of various types of coordinate systems as long as it can represent a position.

According to an embodiment, the external factor data at each of a plurality of points in time or at the current point in time may include at least one of weather conditions measured at corresponding points in time, whether an obstacle has been detected, whether an enemy has been detected, and the amount of remaining fuel of the ship. It may include data.

The weather condition is a factor that may affect the operation of the ship and may include weather conditions such as wave direction, wind direction, wave height, wind speed, and weather.

The wave direction may refer to a direction of a wave in an environment to which a navigation route belongs. When the ship's operating direction is the same as the wave direction, the possible operating distance against fuel consumption may increase. On the other hand, if the sailing direction of the ship is different from the wave direction, the fuel consumption versus the cruising distance may decrease. That is, since wave direction can affect the distance that a ship can navigate relative to fuel, data on wave direction can be considered as learning data and input data.

The wind direction may mean a direction of wind in an environment to which a navigation route belongs. In the case of the wind direction, similar to the wave direction, when the sailing direction of the ship is the same as the wind direction, the distance that can be navigated compared to the fuel consumption may increase. On the other hand, if the sailing direction of the ship is different from the wave direction, the fuel consumption versus the cruising distance may decrease. That is, since wave direction can affect the distance that a ship can navigate relative to fuel, data on wave direction can be considered as learning data and input data.

Whether or not an obstacle is detected and whether an enemy is detected may be data related to whether an obstacle or an enemy is detected that interferes with the ship's navigation within a preset distance range in a preset route. That is, data on whether an obstacle has been detected or not and whether an enemy has been detected can enable safe navigation by detecting an event that is difficult to predict due to a sudden occurrence, such as encountering an obstacle or an enemy during ship navigation. That is, whether or not an obstacle is detected and whether an enemy is detected can be considered as learning data and input data because it helps to determine whether to bypass the ship.

The remaining amount of fuel of the ship may refer to the amount of fuel possessed by the ship to operate the ship.

That is, the method for determining the return time of the ship can predict the return time by learning the return time determination model 410 in consideration of factors that affect ship operation and are closely correlated with each other. Therefore, the method for determining the time to return to the ship can determine the appropriate time to return based on the learned result.

4 is an exemplary view for explaining a vessel return time determination model 410 according to an embodiment.

Referring to FIG. 4 , a process of determining an appropriate return time by using ship state data and external factor data as learning data by the ship return time determination model 410 can be explained.

In the vessel return timing determination model 410, factors influencing ship operation are defined as state information 420, and as a result of state observation in a given environment 450, the state information 420 can be recognized and an action 430 of the ship can be implemented through an artificial neural network (ANN).

At this time, the return time determination model 410 may learn to determine the action guide 430 in a direction in which more rewards can be obtained when the state information 420 affecting the operation of the ship changes.

Specifically, in FIG. 3, the state information 420 is defined as a total of 7 items, namely, the position of the ship at each point in time from s1 to s7, the return point, the amount of remaining fuel, the operating speed, the unexpected state, the self-fault state, and the sea state. However, the state information 420 is not necessarily limited to the above-mentioned matters, and other factors may be added to the state information 420 as other factors that may affect the operation of the ship.

At this time, the state information 420 may correspond to values obtained by measuring the ship's position, return point, remaining fuel amount, sailing speed, sudden state, self-fault state, and sea state information at any point in time by any one of the ship's internal sensor or external sensor and equipment.

In other words, the value of the state information 420 input to s1 to s7 may mean a value measured at an arbitrary point in time. In addition, the return time point determination model 410 may receive state information measured at a plurality of points of time as values of s1 to s7.

Meanwhile, a plurality of points of time may mean points of time at which a unit time preset by a user has elapsed from a reference point of time. For example, when the user sets a preset unit time to 10 seconds, the plurality of points of time may mean points of time after 10 seconds, 20 seconds, 30 seconds, ??, 10*n seconds have elapsed from the reference point of time. In this case, the reference point in time may mean the earliest point in time at which the ship state data and external factor data used as learning data were measured.

In addition, in the learning process after receiving the state information 420 as learning data, the return time determination model 410 determines the action guideline 430 to be applied to the ship among the plurality of preset action guideline 430 based on the learning data, and operates the ship based on the determined action guideline 430, thereby confirming whether the return of the ship is successful or not, and calculating a reward based on whether the return of the ship is successful or not.

For example, at time t, the return time determination model 410 determines an action guideline to be applied to the ship based on the learning data, and determines an appropriate return time point according to the action guideline 430 determined at time t. According to the action guideline 430, the state information 420 for s1 to s7 measured at time t+1 in the environment 450 recognized at time t+1 is input to the model again. Then, the return time determination model 410 is repeatedly learned to determine the action guide 430 so that the reward applied to the return time determination model 410 is maximized.

In this case, the plurality of action guidelines 430 may include mission continuation guidelines and return guidelines. Specifically, the plurality of action guidelines 430 may include mission continuation guidelines and return guidelines stored in the action guide 430 database (Activity DB) 440 . The return time determination model 410 may determine the ship's action guideline 430 as either a mission continuation guideline or a return guideline based on the learning data.

The return point determination model 410 moves the ship's position to the next point when the determined action guideline 430 is the mission continuation guideline, and returns the ship's position to the return point when the determined action guideline 430 is the return guideline.

Specifically, the return point determination model 410 may determine a mission continuation guideline so that the ship continues to navigate to the next point in the following cases.

According to an embodiment, the return point determination model 410 may move the ship to the next point based on the mission continuation guideline, when the ship's position measured at a specific point in time does not pass the return point on the preset route, if the amount of fuel expected to be consumed by the ship to navigate from the ship's position measured at the specific point in time to the preset return point through the return point on the route is equal to or less than the remaining fuel amount at the specific point in time.

That is, in the above situation, if the amount of fuel expected to be consumed is less than or equal to the amount of remaining fuel, even if the ship is further operated, the ship can consume the remaining amount of fuel and return to the return point. Therefore, for efficient ship operation, the return time determination model 410 determines the mission continuation guideline as the action guideline 430 for the ship, and further operates the ship.

According to an embodiment, the return time determination model 410 may return the ship based on the return guideline when the amount of fuel expected to be consumed by the ship to navigate from the location of the ship measured at a specific point in time to the preset return point is equal to or greater than the remaining fuel amount at the specific point in time.

However, if the expected amount of fuel to be consumed is less than or equal to the remaining fuel amount, the ship cannot return to the return point using the remaining amount of fuel when the ship is further operated. Therefore, in order to successfully return the ship, the return time determination model 410 can return the ship at the specific location described above.

According to an embodiment, the return point determination model 410 may return the ship based on a return guideline when an external shock event by an obstacle or an enemy is detected based on external factor state data at a current point in time, and may move the ship to the next point based on a mission continuation guideline when an external shock event is not detected.

The return time determination model 410 may detect an external impact event when it is determined that the ship is externally impacted by an obstacle or an enemy based on external factor state data. At this time, since there is a high possibility that the return of the ship will fail in the case of continued operation, the return time determination model 410 may determine a return guideline for the return of the ship in the above situation. However, these action guidelines 430 may be set as mission continuation guidelines according to a user's decision, and are not necessarily set as return guidelines.

The return time determination model 410 needs to be trained in a direction in which compensation is maximized in order to be learned to determine an appropriate return time point. Therefore, since the goal to be achieved through the return time determination model 410 is the return of the ship, in order to maximize the probability of the ship returning successfully, the return time determination model 410 may calculate the compensation as a positive value if the ship has successfully returned as a result of operating the ship in the learning process, and may calculate the compensation as a negative value if the ship has failed to return.

The return time point determination model 410 learned through the above-described process may determine the action guideline 430 to be applied to the ship among the plurality of action guideline 430 based on the state data of the ship and external factor data at the current time point in the reasoning process.

5 is a flowchart illustrating a method for determining a ship return time point according to an additional embodiment. Specifically, FIG. 5 may specifically explain a process in which the return time determination model 410 determines an appropriate return time point.

The method shown in FIG. 5 may be performed by the computing device 12 shown in FIG. 2 .

The computing device 12 inputs the ship's state data and external factor data at each of a plurality of points in time as learning data, and reinforces the return time point determination model 410 based on compensation according to whether or not the ship returns successfully (510).

Thereafter, the computing device 12 receives state data and external factor data of the ship at the current point in time (520).

Thereafter, the computing device 12 determines whether to return the current ship by inputting the state data and external factor data of the ship at the current point in time to the learned return point determination model 410 (530).

Thereafter, the computing device 12 determines whether the current time point is a return time point based on the ship's state data and the external factor data (540), and if the current time point is the return time point, determines the return of the ship at the current time point (560).

On the other hand, if the return time point determining model 410 determines that the current time point is not the return time point, the point at which the unit time preset by the user has elapsed from the current time point is updated as a new current time point (550).

Thereafter, the computing device 12 repeatedly performs steps 530 to 560 until the updated current time point is determined as the return time point.

On the other hand, in FIGS. 3 and 5, the method is divided into a plurality of steps, but at least some steps are performed in reverse order, combined with other steps, performed together, omitted, divided into detailed steps, or one or more steps not shown may be added and performed.

10: Computing environment
12: Computing device that determines when to return to the vessel
14: Processor
16: computer readable storage medium
18: communication bus
20: program
22: I/O interface
24: I/O device
26: network communication interface
410: return point determination model
420: status information
430: Action Guidelines
440: Action Guidelines Database
450: environment

Claims (18)

inputting ship state data and external factor data at each of a plurality of points in time as learning data, and reinforcing and learning a return time point determination model based on compensation according to whether or not the return of the ship succeeds;
Receiving the ship's state data and external factor data at the current time point; and
Determining whether to return the current ship by inputting state data and external factor data of the ship at the current point in time to the learned return point determination model;
The return point determination model,
In the learning process, determining an action guide to be applied to the ship from among a plurality of preset actions based on the learning data, and operating the ship based on the determined action guide to determine whether the return of the ship is successful, Calculate the reward based on whether the return of the ship is successful,
In the inference process, determining an action guide to be applied to the ship from among the plurality of action guides based on the state data of the ship and external factor data at the current time point.
The method of claim 1,
The state data of the ship at each of the plurality of points in time or the current point in time includes data on at least one of the position of the ship measured at the corresponding point in time, operating speed, self-defect determination, a point on a route to which the ship should reach, and a return point,
The external factor data at each of the plurality of points in time or the current point in time includes data on at least one of weather conditions measured at corresponding points in time, whether or not an obstacle has been detected, whether or not an enemy has been detected, and the amount of remaining fuel of the ship. How to determine ship return time.
delete The method of claim 1,
The return point determination model,
In the learning process,
When the ship is successful in returning as a result of operating the ship, the compensation is calculated as a positive value,
If the ship fails to return, calculating the compensation as a negative value, the ship return time determination method.
The method of claim 1,
The plurality of action guidelines,
Includes mission continuation guidelines and return guidelines;
The return point determination model,
When the determined action guideline is the mission continuation guideline, the position of the ship is moved to the next point,
When the determined action guideline is the return guideline, returning the position of the ship to the return point.
The method of claim 5,
The return point determination model,
When the position of the ship measured at a specific point in time does not pass the turning point on a preset route, the vessel returns to the next point based on the mission continuation guideline, if the amount of fuel expected to be consumed by the ship to navigate from the position of the ship measured at the specific point in time to the preset return point through the turning point on the route is equal to or less than the amount of remaining fuel at the specific point in time.
The method of claim 5,
The return point determination model,
When the amount of fuel expected to be consumed by the ship to navigate from the location of the ship measured at a specific point in time to a preset return point is greater than or equal to the amount of remaining fuel at the specific point in time, returning the ship based on the return guideline. Ship return time determination method.
The method of claim 5,
The return point determination model,
When an external impact event caused by an obstacle or an enemy is detected based on the external factor state data at the current time point, the ship is operated based on the return guideline;
Wherein, when the external impact event is not detected, the ship is moved to a next point based on the mission continuation guideline.
The method of claim 5,
In the receiving step,
When the determined action guideline is the mission continuation guideline, re-receiving ship state data and external factor data at a time point when a preset unit time has elapsed from the current time point;
The determining step is
Re-inputting the re-received ship state data and external factor data into the return time determination model and updating the elapsed time to the current time to determine again whether or not to return the ship at the updated current time point. Ship return time determination method.
one or more processors;
Memory; and
contains one or more programs;
the one or more programs are stored in the memory and configured to be executed by the one or more processors;
The one or more programs,
A command for inputting ship state data and external factor data at each of a plurality of points in time as learning data, and reinforcing and learning a return point determination model based on a reward according to whether the return point of the ship is successful or not;
a command for receiving the state data of the vessel and the external factor data at the current point in time; and
A command for determining whether to return the current ship by inputting state data and external factor data of the ship at the current point in time to the learned return point determination model;
The return point determination model,
In the learning process, determining an action guide to be applied to the ship from among a plurality of preset actions based on the learning data, and operating the ship based on the determined action guide to determine whether the return of the ship is successful, Calculate the reward based on whether the return of the ship is successful,
In the reasoning process, determining an action guide to be applied to the ship from among the plurality of action guides based on the state data of the ship and external factor data at the current time point.
The method of claim 10,
The state data of the ship at each of the plurality of points in time or the current point in time includes at least one of the position of the ship measured at the corresponding point in time, operating speed, self-defect determination, a point on a route to which the ship should reach, and a return point,
The external factor data at each of the plurality of points in time or the current point in time includes data on at least one of weather conditions measured at corresponding points in time, whether an obstacle is detected, whether an enemy is detected, and the amount of remaining fuel of the ship. Computing device.
delete The method of claim 10,
The return point determination model,
In the learning process,
When the ship is successful in returning as a result of operating the ship, the compensation is calculated as a positive value,
Calculating the compensation as a negative value if the ship fails to return.
The method of claim 10,
The plurality of action guidelines,
Includes mission continuation guidelines and return guidelines;
The return point determination model,
When the determined action guideline is the mission continuation guideline, the position of the ship is moved to the next point,
and if the determined action guideline is the return guideline, returning the ship's position to a return point.
The method of claim 14,
The return point determination model,
If the ship's position measured at a specific point in time does not pass the turning point on the preset route, the ship's position measured at the specific point in time passes through the turning point on the route to the preset return point, so that the ship can consume and moving the ship to a next point based on the mission continuation guideline if the expected fuel amount is less than or equal to the remaining fuel amount at the specific point in time.
The method of claim 14,
The return point determination model,
When the amount of fuel expected to be consumed by the ship to navigate from the location of the ship measured at a specific point in time to a preset return point is equal to or greater than the amount of remaining fuel at the specific point in time, returning the ship based on the return guideline. Computing device.
The method of claim 14,
The return point determination model,
When an external impact event caused by an obstacle or an enemy is detected based on the external factor state data at the current time point, the ship is operated based on the return guideline;
and moving the ship to a next point based on the mission continuation guideline when the external shock event is not detected.
The method of claim 14,
The one or more programs,
When the determined action guideline is the mission continuation guideline, a command for re-receiving ship state data and external factor data at a time point when a predetermined unit time has elapsed from the current time point; and
Further comprising a command for re-entering the re-received ship state data and external factor data into the return time determination model and updating the elapsed time point to a current time point to determine again whether to return the ship at the updated current time point. Computing device further comprising a command.