Distributed underwater environment adaptive sensing method for amphibious vehicle

Abstract

The invention relates to a distributed underwater environment adaptive sensing method for an amphibious vehicle, belongs to the technical field of underwater environment sensing, and solves the problems of poor adaptability of underwater environment sensing and obstacle recognition under different turbidity conditions, limited sensing capability of a single sensor and poor data fusion effect in the prior art. Constructing a turbidity cost matrix based on historical detection data, and performing turbidity detection on the current underwater environment by using a turbidity meter to obtain a turbidity detection result; detection information of a target obstacle is obtained through a preset sensor; and fusing the visual distance data, the visual direction data, the visual size data, the ultrasonic distance data, the ultrasonic direction data, the ultrasonic size data, the mechanical direction data and the mechanical distance data through a heterogeneous data fusion algorithm to obtain the distance, the direction and the size of the target obstacle.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater environment perception technology, and in particular to a distributed underwater environment adaptive perception method for amphibious vehicles. Background Art

[0002] With the continuous development of marine resources and underwater engineering operations, higher requirements are placed on underwater environment perception and obstacle recognition technology. Underwater environment perception technology is one of the key technologies to ensure underwater operation safety and improve operation efficiency. However, the existing technologies still have outstanding problems such as poor adaptability of underwater environment perception and obstacle recognition under different turbidity conditions, limited perception capability of a single sensor, and poor data fusion effect.

[0003] At present, underwater environment perception mainly relies on single-type sensors such as sonar and visual sensors. These sensors can provide effective perception information under certain conditions, but in complex underwater environments, the perception ability of a single sensor is limited and it is difficult to meet the needs of all-round obstacle recognition. In addition, the turbidity changes in the underwater environment have a great impact on the performance of the sensor, and the existing technology lacks the adaptive adjustment of sensor performance under different turbidity conditions.

[0004] First, a single sensor is difficult to cope with the changing underwater environment, especially when the turbidity is high, the performance of the visual sensor will drop significantly; second, the existing technology lacks an effective data fusion algorithm and cannot fully utilize the complementary information of multiple sensors, resulting in insufficient accuracy and robustness in obstacle recognition; finally, the existing system is difficult to dynamically adjust the working state of the sensor according to the real-time changes in the underwater environment to adapt to different operating conditions. These problems limit the performance and application scope of underwater operation equipment and increase the risk of underwater operations. Summary of the invention

[0005] In view of the above analysis, the embodiment of the present invention aims to provide a distributed underwater environment adaptive perception method for amphibious vehicles, so as to solve the problems in the existing prior art that underwater environment perception and obstacle recognition have poor adaptability under different turbidity conditions, the perception capability of a single sensor is limited, and the data fusion effect is poor. The method includes:

[0006] S1. Construct a turbidity cost matrix based on historical detection data, and use a turbidity meter to detect the turbidity of the current underwater environment to obtain the turbidity detection result;

[0007] S2. Acquiring detection information of the target obstacle through a preset sensor, including: collecting binocular visual information of the target obstacle based on a binocular visual sensor, collecting ultrasonic information of the target obstacle based on an ultrasonic sensor, and collecting mechanical information of the target obstacle based on a mechanical sensor array;

[0008] S3, obtaining visual distance data, visual direction data, and visual size data of the target obstacle based on binocular visual information, processing ultrasonic information to obtain ultrasonic distance data, ultrasonic direction data, and ultrasonic size data of the target obstacle, and processing mechanical information to obtain mechanical direction data and mechanical distance data of the target obstacle;

[0009] S4. Based on the turbidity detection results and the turbidity cost matrix, the visual distance data, visual direction data, visual size data, ultrasonic distance data, ultrasonic direction data, ultrasonic size data, mechanical direction data and mechanical distance data are fused through a heterogeneous data fusion algorithm to obtain the distance, direction and size of the target obstacle.

[0010] As a further improvement of the present application, S4 further comprises:

[0011] Based on the turbidity detection result and the turbidity cost matrix, a turbidity cost vector matching the current turbidity detection result is determined; based on the turbidity cost vector, the visual distance data, the visual direction data, the visual size data, the ultrasonic distance data, the ultrasonic direction data, the ultrasonic size data, the mechanical direction data and the mechanical distance data are fused through a heterogeneous data fusion algorithm to obtain the distance, direction and size of the target obstacle;

[0012] The turbidity test results include mild, moderate and severe.

[0013] As a further improvement of this application, the turbidity cost matrix is:

[0014]

[0015] Among them, U1 is the light turbidity cost vector, U2 is the moderate turbidity cost vector, and U3 is the heavy turbidity cost vector;

[0016] U1 = [0.6, 0.3, 0.1];

[0017] U2 = [0, 0.7, 0.3];

[0018] U3=[0, 0.6, 0.4].

[0019] As a further improvement of the present application, the distance of the target obstacle is calculated as shown in formula (1);

[0020] d=min{min(d v ,d s , d F ), U i ·[d v ,d s , d F ] T} (1)

[0021] Where d is the distance of the target obstacle, d v , d s , d F They are visual distance data, ultrasonic distance data and mechanical distance data, U i is the turbidity cost matrix, i=1, 2, 3.

[0022] As a further improvement of the present application, the direction of the target obstacle is as shown in calculation formula (2);

[0023] θ=U i ·[θ v ,θ s ,θ F ] T (2)

[0024] Among them, θ is the direction of the target obstacle, θ v ,θ s ,θ F They are visual direction data, ultrasonic direction data and mechanical direction data, U i is the turbidity cost matrix, i=1, 2, 3.

[0025] As a further improvement of the present application, the size of the target obstacle is as shown in calculation formula (3);

[0026] L=max{max(L v , L s ), U i ·[[L v , L s ]] T} (3)

[0027] Where L is the size of the target obstacle, L v , L s They are visual size data and ultrasonic size data, U i is the turbidity cost matrix, i=1, 2, 3.

[0028] As a further improvement of the present application, the visual distance data, visual direction data, and visual size data of the target obstacle obtained by processing the binocular visual information in S3 include:

[0029] Based on the internal and external parameters of the binocular vision sensor and Snell's law, an underwater imaging model is constructed;

[0030] Restoring and correcting the binocular vision information based on the underwater imaging model to obtain corrected binocular vision information;

[0031] Construct the spatial three-dimensional coordinates of the obstacle based on the corrected binocular vision information;

[0032] Obtain visual distance data and visual size data of the target obstacle based on the spatial three-dimensional coordinates of the obstacle;

[0033] The angle between the turntable deflection direction and the vehicle driving direction is determined as the visual direction data of the target obstacle.

[0034] As a further improvement of the present application, the ultrasonic distance data, ultrasonic direction data, and ultrasonic size data of the target obstacle obtained by processing the ultrasonic information in S3 include:

[0035] Calculate the horizontal coordinate distance and the vertical coordinate distance of each position of the target obstacle from the ultrasonic sensor based on the ultrasonic information;

[0036] The ultrasonic direction data of the target obstacle is shown in formula (4);

[0037]

[0038] Among them, θ s is the ultrasonic direction data of the target obstacle, L sx is the lateral coordinate distance of the target obstacle, L sy is the longitudinal coordinate distance of the target obstacle;

[0039] The ultrasonic size data of the target obstacle is shown in formula (5);

[0040]

[0041] Among them, L s The ultrasonic size data of the target obstacle;

[0042] The ultrasonic direction data of the target obstacle is shown in formula (6);

[0043]

[0044] Among them, d s It is the ultrasonic distance data of the target obstacle.

[0045] As a further improvement of the present application, calculating the lateral coordinate distance and the longitudinal coordinate distance of each position of the target obstacle from the ultrasonic sensor based on the ultrasonic information includes:

[0046] Calculate the lateral coordinate distance of each position of the target obstacle from the ultrasonic sensor based on calculation formula (7);

[0047] L sx =|Ct / 2cos(π / 2-ωt)| (7)

[0048] Calculate the longitudinal coordinate distance of each position of the target obstacle from the ultrasonic sensor based on calculation formula (8);

[0049] L sy =Ct1 / 2sin(π / 2-ωt2) (8)

[0050] Among them, C is the propagation speed of ultrasound in water, ω is the angular velocity of the turntable of the ultrasonic sensor, t1 is the detection time of the current transmission cycle, and t2 is the total time of the current scan.

[0051] As a further improvement of the present application, the mechanical direction data and mechanical distance data of the target obstacle obtained by processing the mechanical information in S3 include:

[0052] The mechanical distance data is calculated based on the maximum pressure value collected by the mechanical sensor array in the mechanical information, as shown in formula (9);

[0053]

[0054] Among them, d F is the mechanical distance data, P max is the maximum pressure collected by the mechanical sensor array, ρ is the density of water, v v is the velocity component of the amphibious vehicle, and f is the acquisition frequency of the mechanical sensor;

[0055] The mechanical direction data is obtained based on the sensor position corresponding to the maximum pressure value collected in the sensor array, as shown in formula (10);

[0056] θ F =180k / n-120 (10)

[0057] Among them, θ F is the mechanical direction data, k is the sensor number corresponding to the maximum pressure, and n is the number of mechanical sensors. There are 3 mechanical sensors, numbered 1, 2, and 3, with angles of -6o, 0, and 60 degrees respectively.

[0058] The beneficial effects of the above technical solution are as follows:

[0059] 1. The present invention constructs a turbidity cost matrix and combines the data collected by binocular vision sensors, ultrasonic sensors and mechanical sensor arrays to dynamically adjust sensor weights according to different turbidity conditions, thereby solving the problem of poor adaptability of underwater environment perception under different turbidity conditions in the prior art and improving the accuracy of perception.

[0060] 2. The present invention integrates data from multiple sensors through a heterogeneous data fusion algorithm, fully utilizes the information of binocular vision, ultrasonic and mechanical sensors, significantly improves the accuracy and reliability of obstacle identification, overcomes the defect of limited perception capability of a single sensor in the prior art, and can comprehensively identify and locate underwater obstacles.

[0061] 3. The heterogeneous data fusion algorithm adopted in the present invention can effectively fuse the data from different sensors according to the real-time turbidity detection results and the turbidity cost matrix to obtain the distance, direction and size information of the target obstacle, thus solving the problem of poor data fusion effect in the prior art and improving the overall performance of underwater environment perception.

[0062] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The accompanying drawings are only used for the purpose of illustrating specific embodiments and are not to be considered as limiting the present invention. In the entire drawings, the same reference symbols represent the same components;

[0064] Figure 1 This is a schematic diagram of the process flow of a distributed underwater environment adaptive perception method for amphibious vehicles in an embodiment of the present application;

[0065] Figure 2 This is a schematic diagram of the arrangement of sensors in the embodiment of the present invention;

[0066] Figure 3 This is a schematic diagram of an underwater imaging model in an embodiment of the present invention;

[0067] Figure 4 This is a schematic diagram of restoring and correcting binocular vision information in an embodiment of the present invention;

[0068] Figure 5 This is a schematic diagram of the mechanical sensor information in the embodiment of the present invention. DETAILED DESCRIPTION

[0069] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0070] A specific embodiment of the present invention, as Figure 1As shown, a distributed underwater environment adaptive perception method for amphibious vehicles is disclosed, comprising:

[0071] S1. Construct a turbidity cost matrix based on historical detection data, and use a turbidity meter to detect the turbidity of the current underwater environment to obtain the turbidity detection result;

[0072] The historical test data is the performance of the sensor under different turbidity conditions, including the pre-collected records of various sensor parameters about the underwater environment at different time points and different turbidity conditions, including the accuracy, sensitivity and stability of the sensor. Turbidity reflects the number and size of suspended particles in the water, measured in NTU (scattered turbidity units). Turbidity is measured by a turbidity meter, which determines the turbidity value by measuring the degree of light scattering in the water sample. The higher the turbidity value, the more turbid the water sample. The underwater environment is divided into different levels according to the turbidity value, mild (≤50NTU), moderate (50~500NTU), and severe (500~1000NTU). The turbidity meter can be installed directly above the amphibious vehicle.

[0073] The turbidity cost matrix is ​​used to quantify the performance of the sensor under different turbidity conditions. The turbidity cost matrix is:

[0074]

[0075] Among them, U1 is the light turbidity cost vector, U2 is the moderate turbidity cost vector, and U3 is the heavy turbidity cost vector;

[0076] Specifically,

[0077] U1 = [0.6, 0.3, 0.1];

[0078] U2 = [0, 0.7, 0.3];

[0079] U3=[0, 0.6, 0.4].

[0080] S2, obtaining detection information of the target obstacle through the preset sensor, including: collecting binocular visual information of the target obstacle based on the binocular vision sensor, collecting ultrasonic information of the target obstacle based on the ultrasonic sensor, and collecting mechanical information of the target obstacle based on the mechanical sensor array.

[0081] like Figure 2 As shown, the preset sensors include binocular vision sensors, ultrasonic sensors and mechanical sensor arrays.

[0082] The binocular vision sensor is installed on a turntable and can be set in front of the amphibious vehicle to collect visual images of the target obstacle. Before collection, the shooting direction of the binocular vision sensor is consistent with the driving direction of the amphibious vehicle. When the binocular vision sensor collects binocular visual information of the target obstacle, the turntable is controlled to rotate to keep the target obstacle in the middle of the visual image.

[0083] The ultrasonic sensor is set on the turntable. When the turntable rotates, the ultrasonic sensor sends and receives signals according to the preset detection cycle to ensure the time synchronization between the angular velocity of the turntable and the detection distance. When the turntable rotates 180°, the ultrasonic probe completes a scan, and then the turntable rotates in the opposite direction to continue scanning.

[0084] The mechanical sensor array is used to detect pressure changes in the underwater environment and sense the mechanical properties of the surrounding underwater environment, such as the flow velocity and flow direction of the flow field. It can include 3 uniformly arranged mechanical sensors, numbered 1, 2, and 3, with angles of -60, 0, and 60 degrees respectively.

[0085] S3, obtaining visual distance data, visual direction data, and visual size data of the target obstacle based on binocular visual information, processing ultrasonic information to obtain ultrasonic distance data, ultrasonic direction data, and ultrasonic size data of the target obstacle, and processing mechanical information to obtain mechanical direction data and mechanical distance data of the target obstacle;

[0086] The visual distance data, visual direction data, and visual size data of the target obstacle obtained by binocular visual information processing include:

[0087] Based on the internal and external parameters of the binocular vision sensor and combined with Snell's law, an underwater imaging model is constructed; based on the underwater imaging model, the binocular vision information is restored and corrected to obtain the corrected binocular vision information; based on the corrected binocular vision information, the spatial three-dimensional coordinates of the obstacle are constructed; based on the spatial three-dimensional coordinates of the obstacle, the visual distance data and visual size data of the target obstacle are obtained; and the angle between the turntable deflection direction and the vehicle driving direction is determined as the visual direction data of the target obstacle.

[0088] Specifically, the visual direction data is the angle between the deflection direction of the turntable and the driving direction of the vehicle when the turntable is controlled to rotate to keep the target obstacle in the middle position of the visual image.

[0089] The internal and external parameters of the binocular vision sensor can be calibrated by Zhang's calibration method. The internal parameter matrix B and external parameter matrix P of the binocular vision sensor are:

[0090]

[0091] Among them, (u0, v0) is the coordinate of the pixel in the center of the image, R and T are the rotation and translation matrices respectively, and dx and dy are the dimensions of each pixel in the x-axis and y-axis directions of the image coordinate system.

[0092] like Figure 3 As shown, the underwater imaging model is:

[0093]

[0094] Where n0 is the refractive index of light in water or air; s is the scale factor, B is the camera intrinsic parameter matrix, P is the camera extrinsic parameter matrix, f is the camera focal length; d is the distance parameter from the camera center to the refractive plane; (x w ,y w ,z w ) is the world coordinate of a point on the object; (x r ,y r ,z r ) is the camera coordinate of the intersection of the incident light and the refraction plane; (u, v) is the coordinate of the imaging plane; d is the distance parameter from the camera center to the refraction plane; r is the model variable;

[0095] Among them, (x u ,y u ) is the two-dimensional physical coordinate of the imaging point in the camera coordinate system.

[0096] like Figure 4 As shown, based on the underwater imaging model, the underwater image can be restored and corrected to obtain the corrected binocular vision information as shown in formula (3);

[0097]

[0098] z is the depth of field, d and f are the distance between the refraction plane and the center of the lens and the focal length of the binocular vision sensor respectively; (u a ,v a ) is the pixel coordinate after restoration; (u w ,v w ) is the underwater pixel coordinate; n air and n water is the refractive index of light in air and water, θ air is the angle between the refracted light and the normal vector in the air medium; θ water is the angle between the incident light and the normal vector in the water medium.

[0099] Furthermore, the spatial three-dimensional coordinates of the obstacle are constructed based on the corrected binocular vision information:

[0100]

[0101] Among them, (u l,v l ) and (u r ,v r ) are the pixel coordinates of a point on the object in the image formed by the left and right cameras, (α x ,α y ) are the scale factors of the image in the horizontal and vertical directions.

[0102] The visual distance data and visual size data of the target obstacle are obtained based on the spatial three-dimensional coordinates of the obstacle.

[0103] Visual distance data refers to the straight-line distance between the target obstacle and the amphibious vehicle, which is obtained by calculating the distance between the three-dimensional coordinates of the target obstacle and the optical center coordinates of the binocular vision sensor. Specifically, the distance between the three-dimensional coordinates of each position of the target obstacle and the optical center coordinates of the binocular vision sensor is calculated respectively, and the optical center coordinates of the sensor are subtracted from the three-dimensional coordinates of the target obstacle to obtain the difference on the X, Y, and Z axes. The difference is squared and added, and then the square root is taken to obtain the visual distance data of each position, and the minimum value of the visual distance data of each position is selected as the visual distance between the target obstacle and the amphibious vehicle.

[0104] The visual size data is the actual size of the target obstacle measured by the binocular vision sensor. To calculate the visual size data, the edge point coordinates of the target obstacle are first selected from the three-dimensional coordinates of the target obstacle, and the geometric shape of the target obstacle in the three-dimensional space is defined by the edge points. Each size parameter of the visual size data can be obtained by calculating the maximum and minimum values ​​of the edge point coordinates. Specifically, the length of the target obstacle can be determined by calculating the maximum difference in the X coordinates of the edge points, the width of the target obstacle can be determined by calculating the maximum difference in the Y coordinates of the edge points, and the height of the target obstacle can be determined by calculating the maximum difference in the Z coordinates of the edge points.

[0105] Further, the lateral coordinate distance and the longitudinal coordinate distance of each position of the target obstacle from the ultrasonic sensor are calculated based on the ultrasonic information;

[0106] The ultrasonic direction data of the target obstacle is shown in formula (4);

[0107]

[0108] Among them, θ s is the ultrasonic direction data of the target obstacle, L sx is the lateral coordinate distance of the target obstacle, L sy is the longitudinal coordinate distance of the target obstacle;

[0109] The ultrasonic size data of the target obstacle is shown in formula (5);

[0110]

[0111] Among them, L s The ultrasonic size data of the target obstacle;

[0112] The ultrasonic direction data of the target obstacle is shown in formula (6);

[0113]

[0114] Among them, d s It is the ultrasonic distance data of the target obstacle.

[0115] The calculation of the horizontal coordinate distance and the vertical coordinate distance of each position of the target obstacle from the ultrasonic sensor based on the ultrasonic information includes:

[0116] Calculate the lateral coordinate distance of each position of the target obstacle from the ultrasonic sensor based on calculation formula (7);

[0117] L sx =|Ct1 / 2cos(π / 2-ωt2)| (7)

[0118] Calculate the longitudinal coordinate distance of each position of the target obstacle from the ultrasonic sensor based on calculation formula (8);

[0119] L sy =Ct1 / 2sin(π / 2-ωt2) (8)

[0120] Among them, C is the propagation speed of ultrasound in water, ω is the angular velocity of the turntable of the ultrasonic sensor, t1 is the detection time of the current transmission cycle, and t2 is the total time of the current scan.

[0121] Ultrasonic waves propagate at a constant speed in water. When they encounter a target obstacle, they are reflected back. The sensor receives the reflected signal. The detection time t1 of the current transmission cycle is the time interval between the recorded transmission and reception.

[0122] As the turntable rotates, the ultrasonic sensor's direction changes continuously. When emitting ultrasonic waves at different time points, its angle relative to the target obstacle is different. When the sensor rotates, the lateral coordinate distance L sx It can be expressed by the projection distance of the target obstacle on the rotation plane of the ultrasonic sensor, the longitudinal coordinate distance L sy It can be expressed by the projection distance of the target obstacle in the direction perpendicular to the rotation plane of the ultrasonic sensor.

[0123] Furthermore, the mechanical direction data and mechanical distance data of the target obstacle obtained by processing the mechanical information include:

[0124] like Figure 5 As shown, the mechanical distance data is calculated based on the maximum pressure value collected by the mechanical sensor array in the mechanical information, as shown in calculation formula (9);

[0125]

[0126] Among them, d F is the mechanical distance data, P max is the maximum pressure collected by the mechanical sensor array, ρ is the density of water, v v is the velocity component of the amphibious vehicle, and f is the acquisition frequency of the mechanical sensor;

[0127] The mechanical direction data is obtained based on the sensor position corresponding to the maximum pressure value collected in the sensor array, as shown in formula (10);

[0128] θ F =180k / n-120 (10)

[0129] Among them, θ F is the mechanical direction data, k is the sensor position number corresponding to the maximum pressure, and n is the number of mechanical sensors.

[0130] The velocity of the underwater flow field can be calculated based on the maximum value and distribution position of the pressure sensor

[0131]

[0132] v w is the velocity of the underwater flow field.

[0133] S4. Based on the turbidity detection results and the turbidity cost matrix, the visual distance data, visual direction data, visual size data, ultrasonic distance data, ultrasonic direction data, ultrasonic size data, mechanical direction data and mechanical distance data are fused through a heterogeneous data fusion algorithm to obtain the distance, direction and size of the target obstacle.

[0134] Specifically, based on the turbidity detection results and the turbidity cost matrix, the visual distance data, ultrasonic distance data and mechanical distance data are fused to obtain the distance of the target obstacle;

[0135] The distance of the target obstacle is shown in formula (12);

[0136] d=min{min(d v ,d s , d F ), U i ·[d v , d s , d F ] T} (12)

[0137] Where d is the distance of the target obstacle, d v , d s , d F They are visual distance data, ultrasonic distance data and mechanical distance data, U i is the turbidity cost matrix, i=1, 2, 3. When the turbidity detection result is mild, i=1; when the turbidity detection result is moderate, i=2; when the turbidity detection result is severe, i=3.

[0138] Based on the turbidity detection results and the turbidity cost matrix, the visual direction data, ultrasonic direction data and mechanical direction data of the target obstacle are fused through the heterogeneous data fusion algorithm to obtain the direction of the target obstacle; the direction of the target obstacle is shown in formula (13);

[0139] θ=U i ·[θ v ,θ s ,θ F ] T (13)

[0140] Among them, θ is the direction of the target obstacle, θ v ,θ s ,θ F They are visual direction data, ultrasonic direction data and mechanical direction data, U i is the turbidity cost matrix, i = 1, 2, 3, U i is the turbidity cost matrix, i=1, 2, 3. When the turbidity detection result is mild, i=1; when the turbidity detection result is moderate, i=2; when the turbidity detection result is severe, i=3.

[0141] Based on the turbidity detection results and the turbidity cost matrix, the visual size data and ultrasonic size data of the target obstacle are fused through the heterogeneous data fusion algorithm to obtain the size data of the target obstacle. The size of the target obstacle is shown in formula (14);

[0142] L=max{max(L v , L s ), U i ·[[L v , L s ]] T} (3)

[0143] Where L is the size of the target obstacle, L v , L s They are visual size data and ultrasonic size data, U i is the turbidity cost matrix, i = 1, 2, 3, U iis the turbidity cost matrix, i=1, 2, 3. When the turbidity detection result is mild, i=1; when the turbidity detection result is moderate, i=2; when the turbidity detection result is severe, i=3.

[0144] Compared with the prior art, the distributed underwater environment adaptive perception method for amphibious vehicles provided in this embodiment constructs a turbidity cost matrix and combines the data collected by binocular vision sensors, ultrasonic sensors and mechanical sensor arrays to dynamically adjust the sensor weights according to different turbidity conditions, thereby solving the problem of poor adaptability of underwater environment perception under different turbidity conditions in the prior art and improving the accuracy of perception; integrates the data of multiple sensors through a heterogeneous data fusion algorithm, fully utilizes the information of binocular vision, ultrasonic and mechanical sensors, significantly improves the accuracy and reliability of obstacle recognition, overcomes the defect of limited perception capability of a single sensor in the prior art, and can comprehensively identify and locate underwater obstacles; the heterogeneous data fusion algorithm can effectively fuse data from different sensors according to real-time turbidity detection results and the turbidity cost matrix to obtain the distance, direction and size information of the target obstacle, solves the problem of poor data fusion effect in the prior art, and improves the overall performance of underwater environment perception.

[0145] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0146] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A distributed underwater environment adaptive perception method for amphibious vehicles, characterized in that: The method comprises: S1. Construct a turbidity cost matrix based on historical detection data, and use a turbidity meter to detect the turbidity of the current underwater environment to obtain the turbidity detection result; S2. Acquiring detection information of the target obstacle through a preset sensor, including: collecting binocular visual information of the target obstacle based on a binocular visual sensor, collecting ultrasonic information of the target obstacle based on an ultrasonic sensor, and collecting mechanical information of the target obstacle based on a mechanical sensor array; S3, obtaining visual distance data, visual direction data, and visual size data of the target obstacle based on binocular visual information, processing ultrasonic information to obtain ultrasonic distance data, ultrasonic direction data, and ultrasonic size data of the target obstacle, and processing mechanical information to obtain mechanical direction data and mechanical distance data of the target obstacle; S4. Based on the turbidity detection results and the turbidity cost matrix, the visual distance data, visual direction data, visual size data, ultrasonic distance data, ultrasonic direction data, ultrasonic size data, mechanical direction data and mechanical distance data are fused through a heterogeneous data fusion algorithm to obtain the distance, direction and size of the target obstacle.

2. The method according to claim 1, characterized in that S4 further includes: Based on the turbidity detection result and the turbidity cost matrix, a turbidity cost vector matching the current turbidity detection result is determined; based on the turbidity cost vector, the visual distance data, the visual direction data, the visual size data, the ultrasonic distance data, the ultrasonic direction data, the ultrasonic size data, the mechanical direction data and the mechanical distance data are fused through a heterogeneous data fusion algorithm to obtain the distance, direction and size of the target obstacle; The turbidity test results include mild, moderate and severe.

3. The method according to claim 2, characterized in that The turbidity cost matrix is: Among them, when the turbidity detection result is mild, the turbidity cost vector is U1; when the turbidity detection result is moderate, the turbidity cost vector is U2; when the turbidity detection result is severe, the turbidity cost vector is U3; U1＝[0.6，0.3，0.1]； U2＝[0，0.7，0.3]； U3＝[0，0.6，0.4]。 4. The method according to claim 3, characterized in that The distance of the target obstacle is shown in formula (1); d=min{min(d v ,d s ,d F ),U i ·[d v ,,d s ,d F ] T } (1) Where d is the distance of the target obstacle, d v , d s , d F They are visual distance data, ultrasonic distance data and mechanical distance data, U i is the turbidity cost vector, i=1, 2, 3.

5. The method according to claim 3, characterized in that: The direction of the target obstacle is shown in formula (2); θ=U i ·[θ v ,the s ,the F ] T (2) Among them, θ is the direction of the target obstacle, θ v ,θ s ,θ F They are visual direction data, ultrasonic direction data and mechanical direction data, U i is the turbidity cost vector, i=1, 2, 3.

6. The method according to claim 3, characterized in that The size of the target obstacle is shown in formula (3); L=max{max(L v ,L s ),U i ·[[L v ,,L s ]] T } (3) Where L is the size of the target obstacle, L v , L s They are visual size data and ultrasonic size data, U i is the turbidity cost vector, i=1, 2, 3.

7. The method according to claim 1, characterized in that In S3, the binocular visual information is processed to obtain the visual distance data, visual direction data, and visual size data of the target obstacle, including: Based on the internal and external parameters of the binocular vision sensor and Snell's law, an underwater imaging model is constructed; Restoring and correcting the binocular vision information based on the underwater imaging model to obtain corrected binocular vision information; Construct the spatial three-dimensional coordinates of the obstacle based on the corrected binocular vision information; Obtain visual distance data and visual size data of the target obstacle based on the spatial three-dimensional coordinates of the obstacle; The angle between the turntable deflection direction and the vehicle driving direction is determined as the visual direction data of the target obstacle.

8. The method according to claim 1, characterized in that In S3, ultrasonic information is processed to obtain ultrasonic distance data, ultrasonic direction data, and ultrasonic size data of the target obstacle, including: Calculate the horizontal coordinate distance and the vertical coordinate distance of each position of the target obstacle from the ultrasonic sensor based on the ultrasonic information; The ultrasonic direction data of the target obstacle is shown in formula (4); Among them, θ s is the ultrasonic direction data of the target obstacle, L sx is the lateral coordinate distance of the target obstacle, L sy is the longitudinal coordinate distance of the target obstacle; The ultrasonic size data of the target obstacle is shown in formula (5); Among them, L s The ultrasonic size data of the target obstacle; The ultrasonic direction data of the target obstacle is shown in formula (6); Among them, d s It is the ultrasonic distance data of the target obstacle.

9. The method according to claim 8, characterized in that The calculation of the horizontal coordinate distance and the vertical coordinate distance of each position of the target obstacle from the ultrasonic sensor based on the ultrasonic information includes: Calculate the lateral coordinate distance of each position of the target obstacle from the ultrasonic sensor based on calculation formula (7); L sx =|Ct1 / 2cos(π / 2-ωt2)| (7) Calculate the longitudinal coordinate distance of each position of the target obstacle from the ultrasonic sensor based on calculation formula (8); L sy =Ct1 / 2sin(π / 2-ωt2) (8) Among them, C is the propagation speed of ultrasound in water, ω is the angular velocity of the turntable of the ultrasonic sensor, t1 is the detection time of the current transmission cycle, and t2 is the total time of the current scan.

10. The method according to claim 1, characterized in that The mechanical direction data and mechanical distance data of the target obstacle obtained by processing the mechanical information in S3 include: The mechanical distance data is calculated based on the maximum pressure value collected by the mechanical sensor array in the mechanical information, as shown in formula (9); Among them, d F is the mechanical distance data, P max is the maximum pressure collected by the mechanical sensor array, ρ is the density of water, v v is the velocity component of the amphibious vehicle, and f is the acquisition frequency of the mechanical sensor; The mechanical direction data is obtained based on the sensor position corresponding to the maximum pressure value collected in the sensor array, as shown in formula (10); i F =180k / n-120 (10) Among them, θ F is the mechanical direction data, k is the sensor number corresponding to the maximum pressure, and n is the number of mechanical sensors. There are 3 mechanical sensors, which are numbered 1, 2, and 3, and the angles are -60, 0, and 60 degrees, respectively.

Images