阅读说明：本技术 船舶用物标检测系统及方法、可信度推定装置、以及程序 (Ship object detection system, ship object detection method, reliability estimation device, and program ) 是由 原裕一 坂本雅树 于 2021-06-17 设计创作，主要内容包括：本发明提供一种使推定物标的存在可信度变得容易的船舶用物标检测系统。船舶用物标检测系统具有：多个候补数据生成部,分别生成物标候补数据,该物标候补数据包含在船舶的周围存在的物标候补的位置数据；同一候补选出部,基于位置数据,从由多个所述候补数据生成部分别生成的物标候补数据,选出表示同一物标候补的多个物标候补数据；以及存在可信度计算部,基于多个候补数据生成部中的、生成了选出的多个物标候补数据的多个候补数据生成部的属性,计算同一物标候补的存在可信度。(The invention provides a marine object detection system which facilitates the reliability of the existence of an estimated object. The marine object detection system comprises: a plurality of candidate data generating units each generating target candidate data including position data of target candidates present around the ship; a same candidate selecting unit that selects a plurality of target candidate data indicating the same target candidate from the target candidate data generated by the plurality of candidate data generating units, respectively, based on the position data; and a presence reliability calculation unit that calculates the presence reliability of the same target candidate based on the attributes of the plurality of candidate data generation units that generate the selected plurality of target candidate data.)

1. A marine object detection system comprising:

a plurality of candidate data generating units each generating target candidate data including position data of target candidates present around a ship;

a same candidate selecting unit that selects a plurality of target candidate data indicating the same target candidate from the target candidate data generated by the plurality of candidate data generating units, respectively, based on the position data; and

and an existence reliability calculation unit that calculates existence reliability of the same target candidate based on attributes of a plurality of candidate data generation units that generate the selected plurality of target candidate data among the plurality of candidate data generation units.

2. The marine object detection system of claim 1,

the target candidate data further includes category data of the target candidates,

the ship object detection system further includes a category reliability calculation unit that calculates category reliability of the same object candidate based on the category data included in each of the selected plurality of object candidate data and attributes of a plurality of candidate data generation units that generate the selected plurality of object candidate data.

3. The marine object detection system of claim 1 or 2, wherein,

one of the candidate data generating units is any one of a radar, a communication device that receives information on another ship, a sonar, a combination of an image sensor that captures images of the outside of the ship and a recognition unit that recognizes target object candidates in an image, and an instruction receiving unit that receives an instruction of a user who receives target object candidates in an image.

4. The marine object detection system of any one of claims 1 to 3,

the plurality of candidate data generating units include two or more of a radar, a communication device that receives information of another ship, a sonar, a combination of an image sensor that captures images of the outside of the ship and a recognition unit that recognizes target object candidates in an image, and an instruction receiving unit that receives an instruction of a user who receives target object candidates in the image.

5. The marine object detection system of any one of claims 1 to 4,

the candidate data generating units include at least radar.

6. The marine object detection system of any one of claims 1 to 5,

the candidate data generating units include a plurality of radars having different use frequency bands.

7. The marine object detection system of claim 2, wherein,

one of the candidate data generating units includes:

a detection unit that outputs detection data including radar, an image sensor, or sonar; and

and a recognition unit configured to estimate a category of the target candidate represented by the detection data output from the detection unit using detection data as input data and a training model generated in advance by machine learning using a category of a target as teaching data, and generate the category data.

8. The marine object detection system of claim 7, wherein,

the recognition unit estimates the size or type of the ship as the type of the target object candidate.

9. The marine object detection system of claim 7 or 8, wherein,

the recognition unit estimates the type of the floating object on the water as the type of the target object candidate.

10. The marine object detection system of any one of claims 1 to 9,

the presence reliability calculation unit calculates the presence reliability based on one or more of size data indicating a size of the target object candidate, speed data indicating a moving speed, track data indicating a track, time data indicating an elapsed time from the detection, and intensity data indicating an intensity of the detection signal, in addition to the attribute.

11. The marine object detection system of any one of claims 1 to 10,

the presence reliability calculation unit calculates the presence reliability when the target object candidate is located on a predetermined course higher than the presence reliability when the target object candidate is not located on the predetermined course.

12. The marine object detection system of any one of claims 1 to 11,

the presence reliability calculating unit calculates the presence reliability based on one or more of sea chart data, tide data, and weather data in addition to the attributes.

13. The marine object detection system of any one of claims 1 to 12,

further comprising: and a avoidance calculation unit that calculates a avoidance planned route based on the target candidate data of the target candidate and the presence reliability.

14. The marine object detection system of claim 13,

further comprising: and a ship control part for controlling the ship according to the evasion planning route.

15. A method for detecting a marine object, wherein,

the candidate data generating units respectively generate target candidate data including position data of target candidates existing around the ship,

selecting a plurality of target candidate data indicating the same target candidate from the target candidate data generated by the plurality of candidate data generating units based on the position data,

the present invention is directed to a target object detection device that detects a target object of a target object, and a target object detection method that detects a target object of the target object.

16. A reliability estimation device is provided with:

a candidate data acquisition unit that acquires target candidate data that is generated by each of the plurality of candidate data generation units and that includes position data of target candidates present around the ship;

a same candidate selecting unit that selects a plurality of target candidate data indicating the same target candidate from the target candidate data generated by the plurality of candidate data generating units, respectively, based on the position data; and

and an existence reliability calculation unit that calculates existence reliability of the same target candidate based on attributes of a plurality of candidate data generation units that generate the selected plurality of target candidate data among the plurality of candidate data generation units.

17. A program for causing a computer to execute:

acquiring target candidate data including position data of target candidates present around the ship, the position data being generated by each of the plurality of candidate data generating sections;

selecting a plurality of target candidate data indicating the same target candidate from the target candidate data generated by the plurality of candidate data generating units, respectively, based on the position data; and

the present invention is directed to a target object detection device that detects a target object of a target object, and a target object detection method that detects a target object of the target object.

Technical Field

The present invention relates to a marine target detection system, a marine target detection method, a reliability estimation device, and a program.

Background

Patent document 1 discloses the following: an ARPA radar apparatus and an AIS apparatus of a ship are connected to the ARPA radar apparatus via an interface, and compare the received dynamic information of a nearby ship with the radar symbol information acquired from the ARPA radar apparatus at any time, and use the radar symbol information without the dynamic information determined as a predetermined match as transmission stop warning information, and use the dynamic information without the radar symbol information determined as a predetermined match as non-entity notification warning information, and display the warning information by a display device to warn a user, and convert the warning information into a notification to be transmitted to the nearby ship.

Patent document 1: japanese patent No. 3794641

However, for example, the radar detects a ship, a buoy, land, an iceberg, a floating container, and the like existing around the ship as a target object, but may detect an object unsuitable for the target object, such as noise.

Disclosure of Invention

The present invention has been made in view of the above problems, and a main object thereof is to provide a marine object detection system, a marine object detection method, a reliability estimation device, and a program, which make it easy to estimate the reliability of the presence of a marine object.

In order to solve the above problem, a marine object detection system according to an aspect of the present invention includes: a plurality of candidate data generating units each generating target candidate data including position data of target candidates present around the ship; a same candidate selecting unit that selects a plurality of target candidate data indicating the same target candidate from the target candidate data generated by the plurality of candidate data generating units, respectively, based on the position data; and a presence reliability calculation unit that calculates a presence reliability of the same target candidate based on attributes of a plurality of candidate data generation units that generate the selected plurality of target candidate data among the plurality of candidate data generation units.

In the method for detecting a marine vessel object according to the present invention, the plurality of candidate data generating units generate object candidate data including position data of object candidates existing around the marine vessel, the object candidate data generated by the plurality of candidate data generating units selects a plurality of object candidate data indicating the same object candidate based on the position data, and the existence reliability of the same object candidate is calculated based on the attributes of the plurality of candidate data generating units that generate the selected plurality of object candidate data.

In addition, a reliability estimation device according to another aspect of the present invention includes: a candidate data acquisition unit that acquires target candidate data that is generated by each of the plurality of candidate data generation units and that includes position data of target candidates present around the ship; a same candidate selecting unit that selects a plurality of target candidate data indicating the same target candidate from the target candidate data generated by the plurality of candidate data generating units, respectively, based on the position data; and a presence reliability calculation unit that calculates a presence reliability of the same target candidate based on attributes of a plurality of candidate data generation units that generate the selected plurality of target candidate data among the plurality of candidate data generation units.

A program according to another aspect of the present invention causes a computer to execute: acquiring target candidate data including position data of target candidates present around the ship, the position data being generated by each of the plurality of candidate data generating sections; selecting a plurality of target candidate data indicating the same target candidate from the target candidate data generated by the plurality of candidate data generating units, respectively, based on the position data; and calculating the existence reliability of the same target candidate based on the attributes of a plurality of candidate data generating units that generate the selected plurality of target candidate data among the plurality of candidate data generating units.

According to the present invention, the reliability of the presence of the estimation object is facilitated.

Drawings

Fig. 1 is a diagram showing a configuration example of a marine object detection system according to an embodiment.

Fig. 2 is a diagram showing a configuration example of the reliability estimating device according to the embodiment.

Fig. 3 is a diagram showing an example of the candidate management DB for radar.

Fig. 4 is a diagram showing an example of the AIS candidate management DB.

Fig. 5 is a diagram showing an example of a sonar candidate management DB.

Fig. 6 is a diagram showing an example of the image candidate management DB.

Fig. 7 is a diagram showing an example of the common candidate management DB.

Fig. 8 is a diagram showing an example of the steps of the marine object detection method according to the embodiment.

Description of the reference numerals

1: reliability estimation device, 2: radar, 3: radar, 4: AIS, 5: sonar, 6: image sensor, 7: display device, 8: GNSS receiver, 9: gyro compass, 10: ECDIS, 11: wireless communication unit, 12: automatic steering device, 21: candidate data acquisition unit, 22: the same candidate selection unit, 23: presence reliability calculation unit, 24: category reliability calculation unit, 25: target object recognition unit, 26: instruction receiving unit, 27: avoidance calculation unit, 28: ship control unit, 31: radar candidate management DB, 32: candidate management DB for AIS, 33: sonar candidate management DB, 34: image candidate management DB, 35: common candidate management DB, 100: object detection system for ship

Detailed Description

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

Fig. 1 is a block diagram showing a configuration example of a marine object detection system 100. The marine target detection system 100 is mounted on a marine vessel and is a system for detecting a target object (target) existing around the marine vessel.

The marine vessel object detection system 100 includes a reliability estimation device 1, a plurality of radars 2, 3, AIS4, sonar 5, an image sensor 6, a display device 7, a GNSS receiver 8, a gyro compass 9, an ECDIS10, a wireless communication unit 11, and an automatic steering device 12. These apparatuses are connected to a network N such as a LAN and can perform network communication with each other.

The reliability estimation device 1 is a computer including a CPU, a RAM, a ROM, a nonvolatile memory, an input/output interface, and the like. The CPU of the reliability estimation device 1 executes information processing in accordance with a program loaded from the ROM or the nonvolatile memory to the RAM.

The program may be supplied via an information storage medium such as an optical disk or a memory card, or may be supplied via a communication network such as the internet or a LAN.

In the present embodiment, the reliability estimation device 1 is an independent device, but is not limited thereto, and may be integrated with another device such as ECDIS 10. That is, the function of the reliability estimation device 1 may be incorporated in another device such as the ECDIS 10.

Fig. 2 is a block diagram showing a configuration example of the reliability estimation device 1. The reliability estimation device 1 includes a candidate data acquisition unit 21, a same candidate selection unit 22, a presence reliability calculation unit 23, a category reliability calculation unit 24, a target object recognition unit 25, an instruction reception unit 26, a navigation avoidance calculation unit 27, and a ship control unit 28. These functional units are realized by the CPU of the reliability estimation device 1 executing information processing in accordance with a program.

The reliability estimation device 1 further includes a radar candidate management DB31, an AIS candidate management DB32, a sonar candidate management DB33, an image candidate management DB34, and a common candidate management DB 35. These databases are provided in a nonvolatile memory of the reliability estimation device 1. However, the database may be provided externally and accessed via a communication network.

The description returns to fig. 1. The radars 2, 3 emit microwaves from the antenna and receive reflected waves thereof, and generate echo images based on the received signals. The radar 2, 3 is used in X-band (8-12 GHz), S-band (2-4 GHz), or Ka-band (26-40 GHz). The radars 2 and 3 detect target object candidates expressed in the echo image, and generate target object tracking data indicating positions and velocity vectors of the target object candidates.

The AIS (Automatic Identification System) 4 receives AIS data from other ships existing around the own ship. AIS4 is an example of a communication device that receives information from other vessels. But not limited to AIS, VDES (VHF Data Exchange System: Data Exchange System) may also be used. The AIS data includes, for example, an identification code, a ship name, a location, a heading, a ship speed, and a destination.

Sonar 5 emits ultrasonic waves in the circumferential direction from an ultrasonic transducer provided on the bottom of the ship, receives the reflected waves, and generates an echo image based on the received signals. Sonar 5 detects target object candidates expressed in the echo image, and generates target object tracking data indicating the positions of the target object candidates. This function may be realized by the reliability estimation device 1.

The image sensor 6 is a camera such as a visible light camera or an infrared camera, and generates an image by taking an image of the outside of the ship. The image sensor 6 is provided on the bridge deck of the ship, for example, so as to be oriented toward the bow of the ship. The image sensor 6 may be configured to perform panning, tilting, or zooming in accordance with an operation input by the user.

The display device 7 is a display device with a touch sensor. The touch sensor detects a pointing position of a finger or the like on the screen. But not limited to, a touch sensor, and the pointing position may be input by a trackball or the like. An image captured by the image sensor 6, an echo image generated by the radars 2 and 3, or the sonar 5, and the like are displayed on the display device 7.

The GNSS receiver 8 detects the position of the own ship based on radio waves received from a GNSS (Global Navigation Satellite System). The gyro compass 9 detects the bow orientation of the ship. But not limited to, a gyroscopic compass, a GPS compass or a magnetic compass may also be used.

The ECDIS (Electronic Chart Display and Information System) 10 acquires the position of the ship from the GNSS receiver 8, and displays the position of the ship on the Electronic Chart. In addition, the ECDIS10 also displays a predetermined route set in such a manner that a plurality of waypoints are sequentially tracked on the electronic chart. But is not limited to ECDIS, GNSS plotters may also be used.

The wireless communication unit 11 includes various wireless devices that realize communication with other ship or land base stations, such as ultra-short, medium-short, and short-wavelength wireless devices. The wireless Communication unit 11 accesses, for example, a MICS (Maritime Information and Communication System) and the like to acquire data such as weather and tide.

The automatic steering device 12 calculates a target rudder angle for directing the bow to a predetermined heading based on the azimuth of the bow acquired from the gyro compass 9 and the predetermined heading acquired from the ECDIS10, and drives the steering machine so that the rudder angle of the steering machine approaches the target rudder angle. The automatic steering device 12 may control the engine of the ship.

However, the radars 2 and 3 detect a ship, a buoy, a land, an iceberg, a floating container, and the like existing around the ship as target object candidates, but may detect an object such as noise that is not suitable for detection as a target object. Since it is meaningless to avoid a flight based on such an inappropriate target object, it is important to accurately grasp the presence of the target object. Further, if the presence of the target object and the type of the target object can be grasped, more effective navigation avoidance can be performed.

Therefore, in the present embodiment, the presence reliability and the category reliability of the target candidate existing around the own ship are calculated by a plurality of candidate data generating units such as radars 2, 3, AIS4, and sonar 5.

Radar 2, 3, AIS4, sonar 5, and the like are examples of candidate data generation units that generate target candidate data. The target candidate data includes position data of the target candidate. Further, the target candidate data may include category data of the target candidate.

Specifically, the radars 2 and 3 generate target tracking data indicating the positions of the target candidates and the like as target candidate data. AIS4 generates AIS data indicating the location and type of another ship as target object candidate data. Sonar 5 generates target tracking data indicating the positions of the target candidates and the like as target candidate data.

A combination of the image sensor 6 and the target recognition unit 25 is also an example of the candidate data generation unit. The target recognition unit 25 recognizes target candidates in the image captured by the image sensor 6, thereby generating target candidate data. The positions of the target candidates are calculated based on the imaging direction of the image sensor 6 and the positions of the target candidates in the image. The target recognition unit 25 also estimates the type of the target candidate.

The target object recognition unit 25 estimates the range and the type of the target object candidate in the image captured by the image sensor 6 using a training model generated in advance by machine learning, with the range and the type of the target object in the image as input data. The training model is, for example, an object detection model such as yolo (young Only Look one) or ssd (single Shot multi box detector).

The instruction accepting unit 26 is also an example of the candidate data generating unit, and the instruction accepting unit 26 accepts an instruction by the user of a target object candidate in the image displayed on the display device 7. The target candidate data is generated by a user instructing a target candidate in an image captured by the image sensor 6 and displayed on the display device 7. Further, the target candidate data may be generated by a user instructing target candidates in the echo image generated by the radar 2, 3 or sonar 5 displayed on the display device 7.

The reliability estimation device 1 may acquire target object candidate data generated by radar or the like in another ship from another ship via the wireless communication unit 11 and use the data for calculating the existence reliability or the like. In this case, the wireless communication unit 11 is an example of a communication device that receives the other ship information, and is an example of a candidate data generation unit.

The plurality of candidate data generating units preferably include two or more of radar, AIS, sonar, image sensor, and the like for detecting target object candidates in a polygonal manner. Further, since radar is excellent in the detection target object candidate, it is preferable that the plurality of candidate data generating units include at least one radar.

Further, the radar preferably has different straightness, attenuation, and the like depending on the wavelength of the radio wave, and the plurality of candidate data generating units preferably include a plurality of radars having different use frequency bands. However, the plurality of candidate data generating units may be a plurality of radars having the same use frequency band and different antenna installation locations.

The target candidate data generated by the radar 2, 3 or sonar 5 does not generally include the category data, but the category data may be included in the target candidate data by combining with the target recognition unit 25. That is, the target object recognition unit 25 may estimate the type of the target object candidate expressed in the echo image generated by the radar 2, 3 or sonar 5.

In this case, the target recognition unit 25 estimates the range and the type of the target candidate in the echo image generated by the radar 2, 3 or sonar 5 using the echo image as input data and a training model generated in advance by machine learning using the range and the type of the target in the echo image as teaching data.

Fig. 3 is a diagram showing an example of the candidate radar management DB 31. The radar candidate management DB31 is a database for managing target object candidate data generated by the radars 2 and 3. The radar candidate management DB31 may be stored in the computers of the radars 2 and 3.

The radar candidate management DB31 includes fields such as "candidate identifier", "position", "category", "speed", "heading", "track", "time", "size", "strength", "attribute", and "indication", for example.

The "candidate identifier" is an identifier for identifying a candidate of the object. The "position" indicates a position of the target object candidate. The "category" indicates the category of the target object candidate. Specifically, the "category" is a category of a target candidate estimated by the target recognition unit 25 from the echo image.

"speed", "heading", and "track" respectively indicate the speed, heading, and track of the target object candidate. "time" indicates an elapsed time from detection of the target candidate. The "size" indicates the size of an image appearing in the echo image. "intensity" represents the signal intensity of the reflected wave.

The "attribute" indicates which radar of the plurality of radars 2, 3 detected the attribute. The "instruction" indicates that the user has instructed the echo image displayed on the display device 7.

Fig. 4 is a diagram showing an example of the AIS candidate management DB 32. The AIS candidate management DB32 is a database for managing target candidate data generated by AIS 4. The AIS candidate management DB32 may be stored in the computer of the AIS 4. The AIS candidate management DB32 includes fields such as "candidate identifier", "position", "category", "ship name", "heading", "speed", and "destination".

Fig. 5 is a diagram showing an example of sonar candidate management DB 33. The sonar candidate management DB33 is a database for managing target candidate data generated by sonar 5. The sonar candidate management DB33 may be stored in a computer of sonar 5. The sonar candidate management DB33 includes fields such as "candidate identifier", "position", "category", "time", "size", "intensity", and "instruction".

Fig. 6 is a diagram showing an example of the image candidate management DB 34. The image candidate management DB34 is a database for managing target candidate data generated by the target recognition unit 25 recognizing target candidates in the image captured by the image sensor 6. The image candidate management DB34 includes fields such as "candidate identifier", "position", "category", "speed", "heading", "track", "time", "size", and "indication".

Fig. 7 is a diagram showing an example of the common candidate management DB 35. The common candidate management DB35 is a higher-level database for integrally managing the radar candidate management DB31, the AIS candidate management DB32, the sonar candidate management DB33, and the image candidate management DB 34. The common candidate management DB35 includes fields such as "common candidate identifier", "position", "category", "attribute", "indication", "presence reliability", and "category reliability".

The "common candidate identifier" is an identifier for identifying a common target candidate. The common candidate identifiers are used to collectively represent candidate identifiers of the same target object candidate. The "position" indicates a position of the target object candidate. Whether or not the same object candidate is indicated is determined based on whether or not the positions are common. The "category" indicates the category of the target object candidate.

The "attribute" indicates an attribute of the candidate data generating unit that produces the target candidate data. That is, "attribute" indicates which of the radars 2, 3, AIS4, sonar 5, image sensor 6, and the like is detected. The "existence reliability" and the "category reliability" indicate the existence reliability and the category reliability of the target candidate (described later in detail).

Fig. 8 is a flowchart showing an example of the steps of the marine target detection method implemented in the marine target detection system 100. The reliability estimation device 1 executes the information processing shown in the figure in accordance with a program.

First, the reliability estimation device 1 collects target candidate data generated by a plurality of candidate data generation units such as radars 2 and 3, AIS4, sonar 5, and image sensor 6 (S11: processing by the candidate data acquisition unit 21). Specifically, the reliability estimation device 1 integrates target candidate data managed by the radar candidate management DB31, the AIS candidate management DB32, the sonar candidate management DB33, and the image candidate management DB34 into the common candidate management DB 35.

Next, the reliability estimation device 1 selects a plurality of target candidate data indicating the same target candidate based on the position data included in the target candidate data (S12: processing by the same candidate selection unit 22). Specifically, the reliability estimation device 1 selects a plurality of target candidate data having a common position as data indicating the same target candidate from among a plurality of target candidate data integrated into the common candidate management DB35, and assigns a common candidate identifier to the target candidate data. The reliability estimation device 1 also assigns a common candidate identifier to individual target candidate data in which there is no target candidate data common to other positions.

Next, the reliability estimating device 1 calculates the existence reliability of the target candidate based on the attribute of the candidate data generating unit that generated the target candidate data (S13: as the processing of the existence reliability calculating unit 23). That is, the existence reliability is calculated from which target candidate data is generated from the radar 2, 3, AIS4, sonar 5, image sensor 6, and the like. The calculated existence reliability is stored in the common candidate management DB 35.

When a plurality of target candidate data belong to one common candidate identifier, the reliability estimation device 1 calculates the existence reliability based on the attributes of the plurality of candidate data generation units that have generated the target candidate data. The radars 2, 3, AIS4, sonar 5, image sensor 6, and the like are assigned weights indicating degrees of reliability, and the presence reliability is calculated by weighting them.

For example, since the target candidate data (AIS data) generated by the AIS4 is data relating to another ship received from another ship, it is preferable to set the weight of the AIS4 higher than the weight of the other candidate data generation units such as radars 2 and 3.

Further, since the target candidate data indicated by the user in the image displayed on the display device 7 is a target actually recognized by the user, it is preferable to set the weight of the user's indication higher than the weight of the other candidate data generating unit.

In addition, when the frequency bands used by the radars 2 and 3 are different from each other, since the shorter the wavelength, the higher the linearity and directivity of the radio wave, and the gain are advantageous, it is preferable to set the weight of the radar in the X band to be higher than the weight of the radar in the S band, for example.

In addition, when the antennas of the radars 2 and 3 are installed at different positions from each other, since the shield on the ship is less as the antennas are installed at higher positions such as the main column, it is preferable to set the weight of the radar to be higher as the antennas are installed at higher positions.

The presence reliability may be adjusted based on not only the attribute of the candidate data generating unit but also the speed, track, time, size, or intensity of the target candidate. For example, the presence reliability may be adjusted to be higher as the moving speed is higher as the size of the target object candidate is larger, as the elapsed time from the detection is longer, or as the detection signal intensity is higher.

Further, when the target candidate is located on the planned route of the own ship, the presence reliability may be adjusted to be higher than when the target candidate is not located on the planned route.

The presence reliability may be adjusted based on a chart, weather, time, tide, or the like. For example, when the target object candidate is in an unsustainable area such as a land area or a seaweed net on the sea map, the presence reliability may be adjusted to be high, or the weight of the image sensor may be reduced and the weight of the radar may be increased in the case of weather such as rainy weather or at night.

Next, the reliability estimating device 1 calculates the class reliability of the target candidate based on the class data included in the target candidate data and the attribute of the candidate data generating unit that generated the target candidate data (S14: as the processing of the class reliability calculating unit 24). That is, the category reliability is calculated from the content of the category data and from which of the radars 2, 3, AIS4, sonar 5, image sensor 6, and the like the category data is generated. The calculated category reliability is stored in the common candidate management DB 35.

For example, since the target candidate data (AIS data) generated by the AIS4 includes the identification code of another ship as the category data, it is preferable that the category reliability be evaluated higher than those of the other candidate data generation units such as radars 2 and 3.

The category data of the radars 2 and 3, sonar 5, and image sensor 6 is the category of the target object candidate estimated by the target object recognition unit 25. The target object recognition unit 25 may estimate a type related to the size of a ship such as a small ship, a medium ship, a large ship, or an ultra-large ship, and may estimate a type related to the type of a ship such as a cruise ship, a fishing ship, a business ship, or a tanker. The target object recognition unit 25 may estimate the type of floating objects on water, such as fishing nets, driftwood, people, driftice, and floating containers.

The category reliability may be adjusted based on not only the category data included in the target candidate data and the attribute of the candidate data generator but also the track, the position on the chart, the tide, or the like of the target candidate. For example, in the case where the target object candidate is going into a fishing harbor, or sailing around, or is unnatural reversing tide, the fishing vessel is highly likely, and therefore the type may be determined as a fishing vessel. Further, when the heading of the target candidate is significantly different from the destination, the category reliability of the AIS4 may be lowered.

Next, the reliability estimation device 1 performs calculation based on the ship avoidance algorithm based on the target object candidate data, the existence reliability, and the category reliability stored in the common candidate management DB35, and calculates the ship avoidance planned route (S15: processing performed by the ship avoidance calculation unit 27).

When avoidance of a ship is required as a result of the calculation based on the avoidance driving algorithm (yes in S16), the reliability estimation device 1 performs avoidance control so that the own ship travels along the avoidance planning route by the automatic steering device 12 (S17: as processing by the ship control unit 28).

According to the above-described embodiment, the presence reliability of the target candidate existing around the own ship is calculated by using the plurality of candidate data generating units such as the radars 2, 3, the AIS4, and the sonar 5, and thereby the presence of the target can be grasped accurately and the course avoidance can be performed. Further, by further calculating the category reliability of the target object candidate, the navigation avoidance can be performed more efficiently.

Further, simply obtaining target candidate data individually from a plurality of candidate data generating units such as radars 2, 3, AIS4, sonar 5, and the like results in necessity of a person to determine the reliability thereof, and further, depending on the situation, the reliability may vary. Therefore, as in the present embodiment, the target object candidate data obtained from the plurality of candidate data generating units are indexed by comprehensively considering their attributes and the like, and it is possible to effectively assist the person and the navigation system.

While the embodiments of the present invention have been described above, the present invention is not limited to the embodiments described above, and it is a matter of course that various modifications can be made by those skilled in the art.

In the above embodiment, the presence reliability and the category reliability calculated by the reliability estimating device 1 are used for evasive use, but the present invention is not limited thereto, and may be used for observation use. Specifically, the reliability estimating device 1 may change the display form of the target candidate data according to the presence reliability and the category reliability, thereby facilitating observation.

For example, the higher the confidence level of existence, the brighter the marker of the target candidate is, and when the confidence level of existence is equal to or lower than the threshold value, the marker of the target candidate may not be displayed. The threshold may be a fixed value or may be changed according to the weather, time, or other conditions.

The target candidate mark is displayed on the screens of the radars 2 and 3, the ECDIS10, the display device 7, or the like. In addition, the markers of the target object candidates may be displayed on the image from the image sensor 6 displayed on the display device 7 by AR (Augmented Reality).

In addition, when the presence reliability is calculated, as described above, when the target candidate is located on the planned route of the own ship, the presence reliability is adjusted to be higher than when the target candidate is not located on the planned route, and therefore, the present invention is useful not only for evasion purposes but also for observation purposes.