A method for modeling marine obstacle environment for unmanned boats based on rough set data fusion

Abstract

The present invention belongs to the field of unmanned boat navigation and relates to a method for modeling an unmanned boat's marine obstacle environment based on rough set data fusion. The method comprises: (1) constructing a two-dimensional grid map with the unmanned boat's position as the center; (2) the unmanned boat is equipped with m Different types of obstacle sensors are used to divide the two-dimensional grid map into n regions; (3) for the two-dimensional grid map n area, establish n (4) Using the degree of dependence of the decision attributes on each condition attribute in the information table, n (5) Using the information table of each region to simplify the conditional attributes; n A simplified information table is generated to classify and summarize the decision fusion rules of whether the grid is an obstacle area; (6) whether each grid is an obstacle area is determined to complete the modeling. The method of the present invention provides a new idea for the modeling of the marine obstacle environment of the unmanned boat and improves the detection accuracy.

Description

Technical Field

[0001] The invention belongs to the technical field of unmanned boat navigation, and in particular relates to a method for modeling an unmanned boat marine obstacle environment based on rough set data fusion. Background Art

[0002] As an emerging unmanned surface system, unmanned boats are widely used in fields such as ocean exploration, environmental monitoring and maritime surveillance. In practical applications, accurately sensing obstacles in the environment is the key to ensuring the safe navigation of unmanned boats. However, traditional obstacle detection technology usually relies on single sensor data and is difficult to meet the needs under complex ocean conditions. For this reason, multi-sensor data fusion technology has emerged. Multi-sensor data fusion refers to the comprehensive processing of data from different sensors (such as lidar, millimeter-wave radar, vision, etc.) to obtain more comprehensive and accurate environmental information, so as to make full use of the advantages of various sensors and improve the accuracy and robustness of obstacle detection.

[0003] With the development of artificial intelligence and machine learning technologies, deep learning-based methods have also been widely used in obstacle detection. However, these methods usually require a large amount of labeled data and are prone to misjudgment in dynamic environments.

[0004] As an effective knowledge representation and processing tool, rough sets can perform information analysis in environments with strong uncertainty and ambiguity. Through attribute simplification and data classification of rough sets, key features can be effectively extracted, the dimension of data can be reduced, and data processing efficiency can be improved. Patent CN 107180432 A discloses a navigation method and device. The patent realizes image region segmentation through image grayscale fuzzy characteristic analysis and rough sets combined with OTSU method. It cannot accurately perceive the depth information in the front scene and has the problem of low accuracy. Summary of the invention

[0005] In view of the above technical problems, the present invention provides a method for modeling an unmanned boat's marine obstacle environment based on rough set data fusion, and realizes higher-precision modeling of an unmanned boat's marine obstacle environment based on rough set multi-sensor data fusion.

[0006] The present invention is achieved through the following technical solutions:

[0007] A method for modeling an unmanned boat's marine obstacle environment based on rough set data fusion, the method comprising the following steps:

[0008] (1) Construct a two-dimensional grid map centered on the location of the unmanned boat;

[0009] (2) The unmanned boat is equipped with m Different types of obstacle sensors, mis an integer greater than or equal to 2, and the two-dimensional grid map is divided into n Regions;

[0010] (3) For the two-dimensional grid map n Regions, established in different regions n The corresponding area n Information table;

[0011] (4) Using the degree of dependence of the decision attributes on each condition attribute in the information table, n The information table of each region is simplified by conditional attributes, and redundant conditional attribute information is removed to obtain n A simplified information table;

[0012] (5) Using the two-dimensional grid map n A simplified information table is used to classify and summarize the decision fusion rules of whether the grid is an obstacle area according to different condition attributes;

[0013] (6) Based on the actual obstacle sensor data obtained and the decision fusion rules, determine whether each grid is an obstacle area and complete the construction of the unmanned boat's maritime obstacle environment model.

[0014] Furthermore, in step (2), the two-dimensional grid map is divided into n Areas, respectively R 1 , R 2 ,…, R n indicates that the n The areas include: m The area that can be sensed by all obstacle sensors, m -The area that can be sensed by 1 obstacle sensor, ..., the area that can be sensed by 1 obstacle sensor and the area with no sensed data.

[0015] In step (3), the domain is represented by the region information table U With conditional attributes C , decision attributes D The mapping relationship between them.

[0016] Furthermore, the domain U is a finite set of objects, the first n The domain of a region is represented by ,in l Indicates the number of objects in the domain set; conditional attribute CFor each sensor data, if the sensor has measurement data, it is represented by 1, and if there is no measurement data, it is represented by 0; decision attribute D Indicates whether the grid is an obstacle area. If the grid is an obstacle area, it is represented by 1, and if it is a passable area, it is represented by 0.

[0017] Furthermore, in step (4), the decision attribute D For each condition attribute C The degree of dependence is calculated by the following formula:

[0018] ;

[0019] in, Decision attribute D Conditional attributes C The degree of dependence, Representation Domain U According to the decision attribute D Classification, Representing a collection X For conditional attributes C The lower approximation of Represents the modulus operation on a set.

[0020] Furthermore, in step (4), the redundant condition attribute information is removed, specifically:

[0021] When the decision attribute D Conditional attributes C When the formula of the degree of dependence is calculated to be 0, the corresponding conditional attribute is regarded as a non-dependent conditional attribute;

[0022] In the information table, the information corresponding to the independent conditional attribute is used as redundant conditional attribute information.

[0023] After removing redundant conditional attribute information, a simplified information table is obtained.

[0024] Furthermore, in step (5), for n The method for determining whether a grid in a certain area of ​​the regions is an obstacle area by the decision fusion rule includes:

[0025] According to the result of simplifying the conditional attributes of the information table of each area in step (4), a single conditional attribute or a combination of conditional attributes having a dependency relationship with the decision attribute in each area is obtained, and the detection data of the single conditional attribute or the combination of conditional attributes having a dependency relationship is used as the effective obstacle detection data;

[0026] It is determined whether there is valid obstacle detection data in each grid in the area; if there is valid obstacle detection data in the grid, the corresponding grid is determined to be an obstacle area; if there is no valid obstacle detection data, the corresponding grid is determined to be a passable area.

[0027] Beneficial technical effects of the present invention:

[0028] The unmanned boat maritime obstacle environment modeling method based on rough set data fusion provided by the present invention can make full use of the complementary information of various obstacle sensors, reduce the false detection rate and missed detection rate, and improve the accuracy and reliability of the unmanned boat maritime obstacle environment model construction.

[0029] The method provided by the present invention adopts rough sets, and the model of rough sets has good adaptability and can maintain high detection accuracy with fewer samples. The present invention combines rough sets with multi-sensor data fusion, which can realize accurate detection of sea surface obstacles of unmanned boats in complex marine environments, and provides a new idea for modeling the marine obstacle environment of unmanned boats, which can not only improve the detection accuracy, but also enhance the stability and reliability of the system, and provide strong support for the safe navigation of unmanned boats. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of the method for modeling the marine obstacle environment of an unmanned boat based on rough set data fusion of the present invention;

[0031] Figure 2 The three-dimensional laser radar, millimeter wave radar, and binocular vision detection data in the typical obstacle scene of the present invention;

[0032] Figure 3 This is a schematic diagram of the division of the three-dimensional laser radar, millimeter wave radar, and binocular vision detection areas of the present invention;

[0033] Figure 4 The modeling result of the unmanned boat marine obstacle environment based on rough set data fusion of the present invention;

[0034] Figure 5 The modeling result of the marine obstacle environment of the unmanned boat based on the three-dimensional laser radar of the present invention;

[0035] Figure 6 The modeling result of the marine obstacle environment of the unmanned boat based on the millimeter wave radar of the present invention;

[0036] Figure 7 This is the result of the binocular vision-based modeling of the unmanned boat's marine obstacle environment. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0038] On the contrary, the present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention as defined by the claims. Further, in order to make the public have a better understanding of the present invention, some specific details are described in detail in the detailed description of the present invention below. Those skilled in the art can fully understand the present invention without the description of these details.

[0039] This application provides a method for modeling the marine obstacle environment of an unmanned boat based on rough set data fusion, specifically Figure 1 As shown, the method comprises the following steps:

[0040] (1) Construct a two-dimensional grid map centered on the location of the unmanned boat;

[0041] (2) The unmanned boat is equipped with m Different types of obstacle sensors, m is an integer greater than or equal to 2, and the two-dimensional grid map is divided into n Regions;

[0042] (3) For the two-dimensional grid map n Regions, established in different regions n The corresponding area n Information table;

[0043] (4) Using the degree of dependence of the decision attributes on each condition attribute in the information table, n The information table of each region is simplified by conditional attributes, and redundant conditional attribute information is removed to obtain n A simplified information table;

[0044] (5) Using the two-dimensional grid map n A simplified information table is used to classify and summarize the decision fusion rules of whether the grid is an obstacle area according to different condition attributes;

[0045] (6) Based on the actual obstacle sensor data obtained and the decision fusion rules, determine whether each grid is an obstacle area and complete the construction of the unmanned boat's maritime obstacle environment model.

[0046] In step (1) of this embodiment, a two-dimensional grid map is constructed with the position of the unmanned boat as the center. The size of the grid map can be set according to the maximum effective detection distance of each obstacle sensor, and the area of ​​the two-dimensional grid map is larger than the effective detection area of ​​each obstacle sensor; for example, in this embodiment, the obstacle sensor includes a three-dimensional laser radar, a millimeter wave radar, and a binocular vision, but is not limited to the above three sensors. Since the effective detection distance of the three-dimensional laser radar used is 100m, and the effective detection distance of the millimeter wave radar and the binocular vision is 200m, the size of the two-dimensional grid map constructed with the position of the unmanned boat as the center is 400m×400m.

[0047] Specifically, in step (2) of this embodiment, the obstacle sensor includes a three-dimensional laser radar, a millimeter wave radar, and a binocular vision sensor, but other embodiments are not limited to the above three sensors; Figure 2 It is the detection data of 3D laser radar, millimeter wave radar and binocular vision in typical obstacle scenes. Figure 2 It can be seen that 3D laser radar, millimeter wave radar and binocular vision can all provide information such as obstacle distance and direction, but there are certain differences. Among them, the effective ranging range of the VLP-16 3D laser radar used is about 100 m, and the effective ranging range of the ARS 408-21 millimeter wave radar and the self-developed binocular vision camera is about 200 m; the obstacle information obtained by different detection sensors is complementary, and there are false detections or missed detections. According to the differences in the detection capabilities of the above 3D laser radar, millimeter wave radar and binocular vision, the grid map is divided into 4 areas, respectively. R 1 , R 2 , R 3 , R 4 express. Figure 3 This is a schematic diagram of the division of the 3D laser radar, millimeter wave radar, and binocular vision detection areas used. R 1 Indicates the area that can be sensed by the above three obstacle detection sensors. R 2 Indicates the perception area of ​​only millimeter-wave radar and binocular vision. R 3 Indicates that only the 3D laser radar’s sensing area is present. R 4 Indicates that there is no perception data.

[0048] In step (3) of this embodiment, for the grid map n Regions (in this example, the main analysis R 1 , R 2, R 3 Three areas, R 4 There is no perception data in the area, so no analysis is done). Information tables are established for each area. In this embodiment, the sensor includes a laser radar. L , millimeter wave radar M and binocular vision S ; Decision attributes D Is the grid an obstacle area? If the grid is an obstacle area, it is represented by 1, and if it is a passable area, it is represented by 0. In this embodiment, the decision attribute value is determined based on the prior knowledge of the measurement range and measurement accuracy of different sensors, and the decision attribute value is established. R 1 , R 2 , R 3 Information table of three areas; the decision attributes are mainly determined by the measurement accuracy of different sensors (i.e. conditional attributes) in a certain area. R 1 Area, LiDAR L The measurement accuracy is much higher than that of millimeter wave radar M and binocular vision S , therefore, R 15 For example, although M and S are all 0, but because L is 1, therefore, the final obstacle area is 1; R 16 For example, although M is 1, but due to L is 0, and S is also 0, so the final obstacle area is 0.

[0049] Among them, Table 1 is the area R 1 Information table, Table 2 is the area R 2 Information table, Table 3 is the area R 3 Information table.

[0050] Table 1 Region R 1 Information Sheet

[0051]

[0052] Table 2 Region R 2 Information Sheet

[0053]

[0054] Table 3 Region R 3 Information Sheet

[0055]

[0056] In step (4) of this embodiment, the degree of dependence of the decision attribute on each condition attribute is used to determine the n The information table of each region is simplified by conditional attributes, and redundant conditional attribute information is removed to obtain a simplified information table.

[0057] 1) Targeting R 1 The conditional attribute simplification of the region information table is as follows:

[0058] Domain U 1 = { R 11 , R 12 , R 13 , R 14 , R 15 , R 16 , R 17 , R 18};

[0059] Domain U 1 By decision attribute D The classification is U 1 / D = { D 1 , D 2},in, D 1 = { R 11 , R 12 , R 13 , R 14 , R 15}, D 2 = { R 16 , R 17 , R18};

[0060] Domain U 1 By conditional attributes L The classification is U 1 / L = {{ R 11 , R 12 , R 13 , R 15}, { R 14 , R 16 , R 17 , R 18}};

[0061] Domain U 1 By conditional attributes M The classification is U 1 / M = {{ R 11 , R 12 , R 14 , R 16}, { R 12 , R 15 , R 16 , R 18}};

[0062] Domain U 1 By conditional attributes S The classification is U 1 / S = {{ R 11 , R 13 , R 14 , R 17}, { R 13 , R 15 , R17 , R 18}};

[0063] Domain U 1 By conditional attribute { M , S} is classified as U 1 / { M , S} = {{ R 11 , R 14}, { R 12 , R 16}, { R 13 , R 17}, { R 15 , R 18}}.

[0064] (1) Determine decision attributes D Conditional attributes L Dependence:

[0065] gather D 1 of L The lower approximation is ;

[0066] gather D 2 of L The lower approximation is ;

[0067] Decision attributes D Conditional attributes L The degree of dependence is calculated by the following formula:

[0068] ;

[0069] (2) Determine decision attributes D Conditional attributes M Dependence:

[0070] gather D 1 of M The lower approximation is ;

[0071] gather D 2 ofM The lower approximation is ;

[0072] Decision attributes D Conditional attributes M The degree of dependence is calculated by the following formula:

[0073] ;

[0074] (3) Determine decision attributes D Conditional attributes S Dependence:

[0075] gather D 1 of S The lower approximation is ;

[0076] gather D 2 of S The lower approximation is ;

[0077] Decision attributes D Conditional attributes S The degree of dependence is calculated by the following formula:

[0078] ;

[0079] (4) Determine decision attributes D For conditional attributes { M , S}Degree of dependency:

[0080] gather D 1 of{ M , S} is approximately ;

[0081] gather D 2 of{ M , S} is approximately ;

[0082] Decision attributes D For conditional attributes { M , S The degree of dependence of} is calculated by the following formula:

[0083] ;

[0084] From the above calculation process, we can see that R 1 Regional decision attributesD With conditional attributes L and the combined conditional attributes { M , S} has a dependency relationship, and with a single conditional attribute M and S No dependencies.

[0085] 2) Targeting R 2 The conditional attribute simplification of the region information table is as follows:

[0086] Domain U 2 = { R 21 , R 22 , R 23 , R 24};

[0087] Domain U 2 By decision attribute D The classification is U 2 / D = { D 1 , D 2},in, D 1 = { R 21 , R 22 , R 23}, D 2 ={ R 24};

[0088] Domain U 2 By conditional attributes L The classification is U 2 / L = { R 21 , R 22 , R 23 , R 24};

[0089] Domain U 2 By conditional attributes MThe classification is U 2 / M = {{ R 21 , R 22}, { R 23 , R 24}};

[0090] Domain U 2 By conditional attributes S The classification is U 2 / S = {{ R 21 , R 23}, { R 22 , R 24}}.

[0091] (1) Determine decision attributes D Conditional attributes L Dependence:

[0092] gather D 1 of L The lower approximation is ;

[0093] gather D 2 of L The lower approximation is ;

[0094] Decision attributes D Conditional attributes L The degree of dependence is calculated by the following formula:

[0095] ;

[0096] (2) Determine decision attributes D Conditional attributes M Dependence:

[0097] gather D 1 of M The lower approximation is ;

[0098] gather D 2 of M The lower approximation is ;

[0099] Decision attributes D Conditional attributes M The degree of dependence is calculated by the following formula:

[0100] ;

[0101] (3) Determine decision attributes D Conditional attributes S Dependence:

[0102] gather D 1 of S The lower approximation is ;

[0103] gather D 2 of S The lower approximation is ;

[0104] Decision attributes D Conditional attributes S The degree of dependence is calculated by the following formula:

[0105] ;

[0106] From the above calculation process, we can see that R 2 Regional decision attributes D With conditional attributes M and conditional attributes S There is a dependency relationship, and the conditional attribute L No dependencies.

[0107] 3) Targeting R 3 The conditional attribute simplification of the region information table is as follows:

[0108] Domain U 3 = { R 31 , R 32};

[0109] Domain U 3 By decision attribute D The classification is U 3 / D = { D 1 , D 2},in, D1 = { R 31}, D 2 = { R 32};

[0110] Domain U 3 By conditional attributes L The classification is U 3 / L = {{ R 31}, { R 32}};

[0111] Domain U 3 By conditional attributes M The classification is U 3 / M = { R 31 , R 32};

[0112] Domain U 3 By conditional attributes S The classification is U 3 / S = { R 31 , R 32};

[0113] (1) Determine decision attributes D Conditional attributes L Dependence:

[0114] gather D 1 of L The lower approximation is ;

[0115] gather D 2 of L The lower approximation is ;

[0116] Decision attributes D Conditional attributes L The degree of dependence is calculated by the following formula:

[0117] ;

[0118] (2) Determine decision attributes D Conditional attributes M The degree of dependence

[0119] gather D 1 of M The lower approximation is ;

[0120] gather D 2 of M The lower approximation is ;

[0121] Decision attributes D Conditional attributes M The degree of dependence is calculated by the following formula:

[0122] ;

[0123] (3) Determine decision attributes D Conditional attributes S Dependence:

[0124] gather D 1 of S The lower approximation is ;

[0125] gather D 2 of S The lower approximation is ;

[0126] Decision attributes D Conditional attributes S The degree of dependence is calculated by the following formula:

[0127]

[0128] From the above calculation process, we can see that R 3 Regional decision attributes D Conditional attributes L The degree of dependence is 1, that is, it is completely dependent on the conditional attribute L , and with the conditional attribute M and S No dependencies.

[0129] In step (5) of this embodiment, the grid map is used to nThe information table after attribute simplification of each region is used to classify the decision fusion rules of whether the grid is an obstacle region according to different conditional attributes. According to the result of conditional attribute simplification of the information table of each region in step 4, it can be seen that R 1 Regional decision attributes D With conditional attributes L and the combined conditional attributes { M , S} has a dependency relationship, and with a single conditional attribute M and S No dependencies; R 2 Regional decision attributes D With conditional attributes M and conditional attributes S There is a dependency relationship, and the conditional attribute L No dependencies; R 3 Regional decision attributes D Conditional attributes L The degree of dependence is 1, that is, it is completely dependent on the conditional attribute L , and with the conditional attribute M and S There is no dependency. Based on this, we can summarize the following decision fusion rules for whether a grid is an obstacle area:

[0130] 1) Targeting R 1 The decision fusion rule for whether the grid of a region is an obstacle region is:

[0131] If the area R 1 If there is laser radar detection data in the grid, then the grid is an obstacle area;

[0132] If the area R 1 If there is no laser radar detection data in the grid, but there is both millimeter wave radar and binocular vision detection data, then the grid is an obstacle area;

[0133] Otherwise, the grid is a passable area.

[0134] 2) Targeting R 2 The decision fusion rule for whether the grid of a region is an obstacle region is:

[0135] If the area R 2 If there is millimeter-wave radar detection data or binocular vision detection data in the grid, the grid is an obstacle area;

[0136] Otherwise, the grid is a passable area.

[0137] 3) Targeting R 3 The decision fusion rule for whether the grid of a region is an obstacle region is:

[0138] If the area R 3 If there is laser radar detection data in the grid, then the grid is an obstacle area;

[0139] Otherwise, the grid is a passable area.

[0140] In step (6) of this embodiment, whether each grid is an obstacle area is determined based on the actual acquired obstacle sensor data and the decision fusion rule, and the unmanned boat marine obstacle environment model is constructed. Figure 4 This is the modeling result of the unmanned boat's marine obstacle environment based on rough set data fusion. For comparison, Figure 5 The results of modeling the marine obstacle environment of unmanned boats based on three-dimensional laser radar are given. Figure 6 The results of modeling the marine obstacle environment of unmanned boats based on millimeter-wave radar are given. Figure 7 The results of modeling the marine obstacle environment of unmanned boats based on binocular vision are given. Figure 4 and Figure 5 , Figure 6 , Figure 7 By comparison, we can see that the obstacles in the scene are basically beyond the ranging range of the 3D laser radar. Through sensor information fusion, it is possible to perceive obstacles farther ahead (about 100m to 200m), while reducing the false detection of millimeter-wave radar and binocular vision. The present invention provides a new method for modeling the obstacle environment of unmanned boats at sea, which has a wide range of application prospects and can effectively improve the safe navigation capability of unmanned boats at sea.

[0141] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. The unmanned boat marine obstacle environment modeling method based on rough set data fusion is characterized by: The method comprises the following steps: (1) Construct a two-dimensional grid map centered on the location of the unmanned boat; (2) The unmanned boat is equipped with m different types of obstacle sensors, where m is an integer greater than or equal to 2, and the two-dimensional grid map is divided into n areas according to the different detection capabilities of each obstacle sensor; (3) for the n regions in the two-dimensional grid map, establishing n information tables corresponding to the n regions by region; (4) using the degree of dependence of the decision attribute on each condition attribute in the information table, simplifying the condition attributes of the information tables of the n regions in the two-dimensional grid map, removing redundant condition attribute information, and obtaining n simplified information tables; (5) respectively using the n simplified information tables in the two-dimensional grid map to classify and summarize the decision fusion rules of whether the grid is an obstacle area according to different condition attributes; (6) Determine whether each grid is an obstacle area based on the actual obstacle sensor data and decision fusion rules, and complete the construction of the unmanned boat marine obstacle environment model; In step (2), the two-dimensional grid map is divided into n regions according to the different detection capabilities of each obstacle sensor, and R1, R2, ..., R n It means that the n areas include: the area that can be sensed by m obstacle sensors, the area that can be sensed by m-1 obstacle sensors, ..., the area that can be sensed by 1 obstacle sensor and the area with no sensed data.

2. The method for modeling an unmanned boat's marine obstacle environment based on rough set data fusion according to claim 1 is characterized in that: In step (3), the region information table is used to represent the mapping relationship between the domain U and the condition attribute C and decision attribute D.

3. The method for modeling an unmanned boat's marine obstacle environment based on rough set data fusion according to claim 2 is characterized in that: The domain U is a finite set of objects. The domain of the nth region in the two-dimensional grid map is represented by U n = {R n1 ,R n2 ,…,R nl }, where l represents the number of objects in the domain set; the conditional attribute C is the data of each sensor. If the sensor has measurement data, it is represented by 1, and if it has no measurement data, it is represented by 0; the decision attribute D represents whether the grid is an obstacle area. If the grid is an obstacle area, it is represented by 1, and if it is a passable area, it is represented by 0.

4. The method for modeling an unmanned boat's marine obstacle environment based on rough set data fusion according to claim 1 is characterized in that: In step (4), the degree of dependence of the decision attribute D on each condition attribute C is calculated by the following formula: Among them, r(C,D) is the degree of dependence of decision attribute D on condition attribute C, U / D represents the classification of domain U according to decision attribute D, CX represents the lower approximation of set X for condition attribute C, and || represents the modulo operation on the set.

5. The method for modeling an unmanned boat's marine obstacle environment based on rough set data fusion according to claim 4 is characterized in that: In step (4), the redundant condition attribute information is removed, specifically: When the formula of the degree of dependence of the decision attribute D on the condition attribute C is calculated to be 0, the corresponding condition attribute is regarded as an independent condition attribute; In the information table, the information corresponding to the independent conditional attribute is used as redundant conditional attribute information, and after removing the redundant conditional attribute information, a simplified information table is obtained.

6. The method for modeling an unmanned boat's marine obstacle environment based on rough set data fusion according to claim 1 is characterized in that: In step (5), the method for determining the decision fusion rule of whether a grid of a certain area among the n areas is an obstacle area includes: According to the result of simplifying the conditional attributes of the information table of each area in step (4), a single conditional attribute or a combination of conditional attributes having a dependency relationship with the decision attribute in each area is obtained, and the detection data of the single conditional attribute or the combination of conditional attributes having a dependency relationship is used as the effective obstacle detection data; It is determined whether there is valid obstacle detection data in each grid in the area; if there is valid obstacle detection data in the grid, the corresponding grid is determined to be an obstacle area; if there is no valid obstacle detection data, the corresponding grid is determined to be a passable area.

Images