AUV (Autonomous Underwater Vehicle) adaptive surveying and mapping path planning method based on mixed grid map and dynamic sampling

Abstract

The invention relates to an AUV (Autonomous Underwater Vehicle) adaptive surveying and mapping path planning method based on a mixed grid map and dynamic sampling, which comprises the following specific steps of: S1, constructing a grid map mixed model based on the interrelation of AUV, terrain and sonar; s2, after the AUV sails for a section of path, re-sampling path points on the basis of the grid map hybrid model, and planning a next section of path on the new path points; and S3, constructing an evaluation function, calculating a route formed by the sampled candidate path points, evaluating all routes meeting conditions through a graph search algorithm, and selecting an optimal path. According to the method, a grid map hybrid model is adopted, and a confidence coefficient layer is the core of map data and is also the basis of a path planning algorithm; the terrain coefficient layer provides terrain complexity information for the confidence coefficient layer; the saliency layer identifies an area with a saliency terrain, so that the AUV can improve the positioning precision through closed-loop detection, and a reference basis is provided for path planning.

Description

Technical Field

[0001] The invention belongs to the technical field of AUV underwater navigation and perception, and relates to an AUV adaptive mapping path planning method based on hybrid grid graph and dynamic sampling. Background Art

[0002] With the continuous development and growth of the marine industry, underwater robots, as an indispensable tool for the development of marine resources, have been greatly developed. Among the key technologies of underwater robots, the positioning and perception capabilities of underwater robots are still the key to restricting the development of underwater robots. In underwater environments, the positioning of underwater robots mainly relies on inertial navigation systems, combined with technologies such as hydroacoustic positioning systems, geophysical navigation systems, and terrain-assisted navigation systems. In this case, the positioning of underwater robots is subject to many restrictions, with poor autonomy and a relatively low level of intelligence. Summary of the invention

[0003] In order to solve the above problems existing in the current technology, the present invention provides an AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling, which can improve positioning accuracy, high autonomy and improved planning efficiency.

[0004] The technical solution adopted by the present invention is:

[0005] The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling has the following specific steps:

[0006] S1, constructs a grid map hybrid model based on the interrelationship between AUV, terrain and sonar;

[0007] S2, after navigating a path, the AUV resamples the waypoints based on the grid map hybrid model and plans the next route on the new waypoints;

[0008] S3, constructs an evaluation function, calculates the route formed by the sampled candidate path points, and evaluates all routes that meet the conditions through a graph search algorithm to select the optimal path.

[0009] Furthermore, the grid map hybrid model includes a depth data layer, a confidence layer, a terrain coefficient layer and a significance layer.

[0010] Furthermore, the steps of constructing the grid map hybrid model are as follows:

[0011] S11, rasterizing the mission area, converting the sonar data into a geodetic coordinate system, and filling the raster map;

[0012] S12, the degree of mapping completion of a certain point on the map is evaluated by the grid confidence. The confidence of the blank area is 0, and the confidence of the mapped area is 0-1. The larger the confidence, the better the mapping completion of the grid. The confidence of the obstacle area is set to 2.

[0013] Furthermore, the coordinate conversion formula of the sonar data in step S11 is shown in formula (1):

[0014]

[0015] Where P e , P s are the coordinates of the data points in the geodetic coordinate system and the sonar coordinate system, respectively. It is the transformation matrix from the sonar coordinate system to the AUV body coordinate system. After the sonar is installed, this matrix is ​​determined. It is the transformation matrix from the AUV body coordinate system to the earth coordinate system, which is related to the position and attitude of the AUV.

[0016] Further, in step S12, the grid confidence C p The calculation formula is shown in formula (2):

[0017]

[0018] In the formula, C k is the confidence of the kth data in the grid, M k is the kth measurement data of the grid,

[0019] h b is the terrain coefficient, T is the threshold, S(M k ) is the standard deviation of each measurement data.

[0020] Furthermore, the confidence level C of the confidence layer k The calculation of is as follows:

[0021] The ratio of the beam projection area directly below the sonar to the beam projection area at the specified position is proposed to be used. s As a measure of confidence in sonar data:

[0022]

[0023] In the formula, it is assumed that the projection surface is a plane, h is the height from the sonar to the projection surface, α is the angle between the sonar centerline and the vertical direction (beam angle), and β is the half beam width; the numerator is the projection area of ​​the beam directly below, and the denominator is the projection area when the beam angle is; C s The larger it is, the higher the confidence of the data is, and the more it can truly reflect the seabed topography;

[0024] Using the offset distance e from the data point to the bottom of the AUV as a variable, α = atan (e / h), and substituting it into equation (3), we can get the confidence level of each data on the sonar scan line;

[0025] C of a certain measurement data s It is the confidence C of the kth measurement of the corresponding grid k .

[0026] Furthermore, the terrain coefficient h of the terrain coefficient layer b The acquisition is as follows:

[0027] The area with a radius of R around the grid is examined. If the number of data points in the area exceeds the threshold, the standard deviation S of these data is calculated. R , and normalize it to serve as the terrain coefficient h of the grid b .

[0028] Furthermore, the saliency layer is labeled by a saliency region extraction algorithm, as follows:

[0029] Find the grids whose data values ​​are lower than the set value 1 or higher than the set value 2, and mark these grids as 1, and mark the rest of the grids as 0 to form a binary image;

[0030] Use the connected domain algorithm in image processing to find the area larger than the set value and mark it as a significant area;

[0031] When the AUV passes through a marked significant area, the historical data corresponding to the current surveying area is found according to the distribution characteristics of the grid data, and the coordinate difference of the corresponding grid is calculated to calibrate the map data and the positioning information of the AUV.

[0032] Furthermore, the resampled path points in step S2 are distributed in areas where the confidence level is lower than the set value. The sampling points are evenly distributed at the edge of the unknown area, and the map is opened and closed to achieve edge connectivity. The contour of the unknown area can then be extracted through a mature edge extraction algorithm.

[0033] Furthermore, the evaluation function in step S3 assigns a value to each line segment connecting the path points:

[0034] V i =W s S i +W M M i +W L L i (4)

[0035] V i represents the evaluation function value of the i-th line segment,

[0036] S i is the expected mapping efficiency, taking the average scan line width of the existing data As the expected scan line width of the sonar, count the number of The number of low-confidence grids N contained in the rectangular area with width , assuming that the line segment length is L, the expected mapping efficiency S i =N / L;

[0037] M i is the interpolation gain, which counts the average distance from the candidate line segment to the known area;

[0038] L i is the positioning gain, positioning gain L i The area where the expected scanning area intersects with the significant area is counted. If the area is greater than the threshold, L i =1; the smaller the intersection area, the smaller the L i The smaller the size, the less they intersect. i =0;

[0039] W s , W M , W L is the weight of each item.

[0040] Beneficial effects of the present invention:

[0041] (1) The present invention adopts a grid map hybrid model, in which the confidence layer is the core of the map data and the basis of the path planning algorithm. After the surveying and mapping is completed, the mean and variance of the confidence layer can be statistically analyzed to evaluate the surveying and mapping effect; the terrain coefficient layer provides terrain complexity information for the confidence layer; the significance layer identifies the area with significant terrain, which can enable the AUV to improve the positioning accuracy through closed-loop detection and provide a reference basis for path planning.

[0042] (2) The present invention adopts a dynamic sampling and dynamic programming method with strong adaptability. Moreover, through edge sampling, the number of sampling points can be greatly reduced, thereby improving the efficiency of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the relationship between the AUV, terrain, sonar and map model of the present invention.

[0044] Figure 2 It is a schematic diagram of the raster map data format of the present invention.

[0045] Figure 3 It is a schematic diagram of the multi-beam sonar imaging effect of the present invention.

[0046] Figure 4 The figure is a flow chart of the salient region extraction algorithm of the present invention.

[0047] Figure 5 Schematic diagram of edge extraction and sampling of unknown areas in the present invention.

[0048] Figure 6 This is a schematic diagram of route search according to the present invention. The numerical values ​​in the figure are the reward values ​​for each line segment. DETAILED DESCRIPTION

[0049] The present invention is further described below in conjunction with specific embodiments, but the present invention is not limited to these specific embodiments. Those skilled in the art should recognize that the present invention covers all possible alternatives, improvements and equivalents within the scope of the claims.

[0050] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "clockwise", "counterclockwise" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, the meaning of "multiple" is two or more, unless otherwise clearly defined.

[0051] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0052] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes that the first feature is directly below and obliquely below the second feature, or simply indicates that the first feature is lower in level than the second feature.

[0053] This embodiment provides an AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling, and the specific steps are as follows:

[0054] S1, constructs a grid map hybrid model based on the interrelationship between AUV, terrain and sonar;

[0055] The interrelationships among AUV, terrain and sonar are as follows: Figure 1 As shown in the figure, the AUV sails above the terrain, and the terrain features can provide auxiliary positioning information for the AUV; the sonar generates data by sensing the terrain, and the undulation of the terrain affects the sonar scanning line width and data quality; the sonar data provides information support for the AUV's path planning, and the AUV's operating parameters and positioning accuracy will affect the sonar data quality. The above relationship is reflected in the grid map hybrid model to form Figure 2 The four-layer map model shown.

[0056] The grid map hybrid model includes a depth data layer, a confidence layer, a terrain coefficient layer, and a significance layer. The confidence layer is the core of the map data and the basis of the path planning algorithm. After the surveying and mapping is completed, the mean and variance of the confidence layer can be statistically analyzed to evaluate the surveying and mapping effect. The terrain coefficient layer provides terrain complexity information for the confidence layer. The significance layer identifies areas with significant terrain, which can improve the positioning accuracy of the AUV through closed-loop detection and provide a reference for path planning.

[0057] The steps for constructing the grid map hybrid model are as follows:

[0058] S11, rasterizing the task area according to the resolution requirements of the survey map (if there are sonar data points in the grid, it means that the grid has been surveyed), converting the sonar data into the geodetic coordinate system, and filling the grid map;

[0059] The coordinate transformation formula of sonar data is shown in formula (1):

[0060]

[0061] Where P e , P s are the coordinates of the data points in the geodetic coordinate system and the sonar coordinate system, respectively. It is the transformation matrix from the sonar coordinate system to the AUV body coordinate system. After the sonar is installed, this matrix is ​​determined. It is the transformation matrix from the AUV body coordinate system to the earth coordinate system, which is related to the position and attitude of the AUV.

[0062] S12, the degree of mapping completion of a certain point on the map is evaluated by the grid confidence. The confidence of the blank area is 0, and the confidence of the mapped area is 0-1. The larger the confidence, the better the mapping completion of the grid. The confidence of the obstacle area is set to 2.

[0063] The grid confidence C p The calculation formula is shown in formula (2):

[0064]

[0065] In the formula, C k is the confidence of the kth data in the grid, M k is the kth measurement data of the grid,

[0066] h b is the terrain coefficient, T is the threshold, S(M k ) is the standard deviation of each measurement data.

[0067] The model considers four factors: the confidence of sensor data, the undulation of terrain, the number of times the grid is measured, and the standard deviation S (M k The basic idea is that if a grid is measured multiple times and the measured data are relatively consistent, each additional measurement can increase the confidence of the grid. If the terrain near a grid is undulating, its confidence will be appropriately reduced, but C p It can still be improved by measuring the number of times.

[0068] The surveying accuracy of multi-beam sonar is closely related to the beam angle, such as Figure 3 As shown in the figure, the same beam width covers different areas at different beam angles. The larger the beam angle, the larger the covered area and the lower the mapping accuracy. In addition, when the beam angle is large, the scattered waves generated by the sound waves on the seabed are also weak. If the scattered waves are not detected by the sonar, the beam will not be able to detect depth data.

[0069] The ratio of the beam projection area directly below the sonar to the beam projection area at the specified position is proposed to be used. s As a measure of confidence in sonar data:

[0070]

[0071] In the formula, it is assumed that the projection surface is a plane, h is the height from the sonar to the projection surface, α is the angle between the sonar centerline and the vertical direction (beam angle), and β is the half beam width; the numerator is the projection area of ​​the beam directly below, and the denominator is the projection area when the beam angle is; C s The larger it is, the higher the confidence of the data is, and the more it can truly reflect the seabed topography;

[0072] When the sonar is installed on the AUV, the beam angle relative to the seabed will change with the roll of the AUV, but the data confidence relative to the distribution directly below the AUV remains unchanged. Therefore, the offset distance e from the data point to the bottom of the AUV is used as a variable, α = atan (e / h). Substituting it into formula (3), the confidence of each data on the sonar scan line can be obtained;

[0073] C of a certain measurement data s It is the confidence C of the kth measurement of the corresponding grid k .

[0074] When the terrain is undulating, the area is surveyed more carefully to ensure the accuracy of the data. When the terrain is undulating, the area can be interpolated to improve the efficiency of surveying. In order to quantitatively describe the degree of terrain undulation, the area with a radius of R around the grid is examined. If the number of data points in the area exceeds the threshold, the standard deviation S of these data is calculated. R , and normalize it to serve as the terrain coefficient h of the grid b .

[0075] Grid h b If it is large, it means that the terrain nearby is undulating. In formula (2), h b The grid confidence will be appropriately reduced. The more times the grid is measured, the higher the h b In addition, if the h of most grids in the current measurement data is b If it is too large, the navigation speed of the AUV should be appropriately reduced to increase the density of data points.

[0076] When a small blank area is within a certain range of h b When both are small, the area can be interpolated using methods such as Kriging interpolation, and there is no need to plan and map the path of this area in the subsequent path. The size of the interpolation area and the terrain coefficient threshold are determined according to the actual task requirements and the results of simulation and experiments.

[0077] AUV can dynamically avoid obstacles without marking the obstacle area in advance. When the AUV detects an obstacle through the front sonar, it will go around the obstacle and mark the obstacle area while mapping. The confidence of the grid in the obstacle area is represented by 2, and the planned route cannot pass through these grids.

[0078] The positioning accuracy of AUV is the basic condition for achieving accurate surveying and mapping. Traditional positioning methods such as dead reckoning have cumulative errors. We hope to use terrain information to reduce these errors and thus improve surveying and mapping accuracy. The terrain in some areas has relatively significant features, such as hill-like protrusions or lake-like depressions. Such landmarks are easy to distinguish and can be used as reference areas for positioning. The grids in these areas are marked with 1, and the rest of the non-significant areas are marked with 0. The significant areas are planned to be marked with Figure 4 The algorithm shown is searched.

[0079] Specifically, find the grids whose data values ​​are lower than the set value 1 or higher than the set value 2, and mark these grids as 1, and mark the rest of the grids as 0, to form a binary image;

[0080] The connected domain algorithm in image processing is used to find the area larger than the set value and mark it as a significant area;

[0081] When the AUV passes through a marked significant area, the historical data corresponding to the current surveying area is found according to the distribution characteristics of the grid data, and the coordinate difference of the corresponding grid is calculated to calibrate the map data and the positioning information of the AUV. That is, after marking the significant area, the closed-loop detection algorithm can be used to improve the positioning accuracy.

[0082] S2, after navigating a path, the AUV resamples the waypoints based on the grid map hybrid model and plans the next route on the new waypoints;

[0083] Specifically, the surveying path is a full-coverage path, and path planning needs to be continued after the local area mapping is completed. For this purpose, a dynamic sampling and dynamic planning method is adopted. That is, after the AUV navigates a path, it resamples the path points based on the completed map data model and plans the next route on the new path points until the overall confidence of the map meets the task requirements.

[0084] In order to effectively map the unknown area, the resampled path points are distributed in the areas where the confidence is lower than the set value. If the path points are set inside these unknown areas, a single path will not be able to completely pass through the unknown area, and more paths and more turns are required to complete the mapping of the area. Therefore, it is proposed to evenly distribute the sampling points at the edge of the unknown area, such as Figure 5As shown. Considering the grid as pixels in the image, the map can be opened and closed to achieve edge connectivity, and then the outline of the unknown area can be extracted through a mature edge extraction algorithm. This edge sampling method can greatly reduce the number of sampling points and improve the efficiency of the algorithm. Moreover, as the unknown area decreases, the number of sampling points will also decrease. The number of sampling points is set based on multiple simulation experiments.

[0085] S3, constructs an evaluation function, calculates the route formed by the sampled candidate path points, and evaluates all routes that meet the conditions through a graph search algorithm to select the optimal path.

[0086] Specifically, after sampling candidate path points, an optimal path needs to be found from them. The path points are connected according to the set range, and the line segments passing through obstacles are automatically deleted. Traditional graph search algorithms evaluate the quality of paths by route length. In surveying and mapping tasks, the goal of path planning is obviously not the shortest path. Each line segment connecting the path points is assigned a value through the following evaluation function:

[0087] V i =W s S i +W M M i +W L L i (4)

[0088] V i represents the evaluation function value of the i-th line segment, S i is the expected mapping efficiency, M i is the interpolation gain, L i is the positioning gain, W s , W M , W L is the weight of each item.

[0089] Get the average scan line width of the existing data As the expected scan line width of the sonar, count the number of The number of low-confidence grids N contained in the rectangular area of ​​width, assuming that the line segment length is L, S i =N / L, which is the expected surveying and mapping efficiency.

[0090] Some tasks require obtaining as much terrain information as possible in a specified area in the shortest possible time. For areas that cannot be surveyed in time, interpolation can be performed. To deal with this situation, an interpolation gain M is added. i , which counts the average distance from the candidate line segment to the known area. A larger interpolation gain means that the route is farther from the known area, and the information obtained through the route is more helpful for interpolation.

[0091] Positioning gain L i The area where the expected scanning area intersects with the significant area is counted. If the area is greater than the threshold, L i =1. The smaller the intersection area, the greater the L i The smaller the size, the less they intersect. i =0, which is beneficial for the AUV to calibrate its position when passing through a significant area and improve the positioning accuracy.

[0092] The result of the evaluation function is the reward value of each line segment. Next, the optimal path is found through graph search algorithms such as Dijkstra. The starting point is set at the end point of the current path, and the average reward of each route leading to each sampling point is calculated. The number of nodes in the route is limited to 2 to 4. The relevant term L of the turning angle between the line segments must also be subtracted from the average reward of the route. angle , because a large turning angle will affect the position accuracy of the AUV and cause uneven distribution of multi-beam data. When turning, the AUV must also reduce its navigation speed to increase the density of outer data points. Finally, the route with the largest average reward is selected as the next route. The above sampling and route selection process is repeated until the confidence index of the map meets the task requirements.

[0093] Figure 6 Several candidate paths are shown in the figure. The values ​​marked on the lines are the rewards calculated based on the evaluation function. The average reward value of the red route on the far left is (0.8+0.6+0.3) / 3-L angle The graph search algorithm will evaluate all routes that meet the conditions and obtain the route with the largest average reward.

Claims

1. The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling, the specific steps are as follows: S1, constructs a grid map hybrid model based on the interrelationship between AUV, terrain and sonar; S2, after navigating a path, the AUV resamples the waypoints based on the grid map hybrid model and plans the next route on the new waypoints; S3, constructs an evaluation function, calculates the route formed by the sampled candidate path points, and evaluates all routes that meet the conditions through a graph search algorithm to select the optimal path.

2. The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling according to claim 1 is characterized in that: The grid map hybrid model includes a depth data layer, a confidence layer, a terrain coefficient layer and a significance layer.

3. The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling according to claim 2 is characterized in that: The steps for constructing the grid map hybrid model are as follows: S11, rasterizing the mission area, converting the sonar data into a geodetic coordinate system, and filling the raster map; S12, the degree of mapping completion of a certain point on the map is evaluated by the grid confidence. The confidence of the blank area is 0, and the confidence of the mapped area is 0-1. The larger the confidence, the better the mapping completion of the grid. The confidence of the obstacle area is set to 2.

4. The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling according to claim 3 is characterized in that: The coordinate conversion formula of the sonar data in step S11 is shown in formula (1): Where P e , P s are the coordinates of the data points in the geodetic coordinate system and the sonar coordinate system, respectively. It is the transformation matrix from the sonar coordinate system to the AUV body coordinate system. After the sonar is installed, this matrix is ​​determined. It is the transformation matrix from the AUV body coordinate system to the earth coordinate system, which is related to the position and attitude of the AUV.

5. The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling according to claim 3 is characterized in that: The grid confidence C in step S12 p The calculation formula is shown in formula (2): In the formula, C k is the confidence of the kth data in the grid, M k is the kth measurement data of the grid, h b is the terrain coefficient, T is the threshold, S(M k ) is the standard deviation of each measurement data.

6. The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling according to claim 5 is characterized in that: The confidence level C of the confidence layer k The calculation of is as follows: The ratio of the beam projection area directly below the sonar to the beam projection area at the specified position is proposed to be used. s As a measure of confidence in sonar data: In the formula, it is assumed that the projection surface is a plane, h is the height from the sonar to the projection surface, α is the angle between the sonar centerline and the vertical direction (beam angle), and β is the half beam width; the numerator is the projection area of ​​the beam directly below, and the denominator is the projection area when the beam angle is; C s The larger it is, the higher the confidence of the data is, and the more it can truly reflect the seabed topography; Using the offset distance e from the data point to the bottom of the AUV as a variable, α = atan (e / h), and substituting it into equation (3), we can get the confidence level of each data on the sonar scan line; C of a certain measurement data s It is the confidence C of the kth measurement of the corresponding grid k .

7. The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling according to claim 5 is characterized in that: The terrain coefficient h of the terrain coefficient layer b The acquisition is as follows: The area with a radius of R around the grid is examined. If the number of data points in the area exceeds the threshold, the standard deviation S of these data is calculated. R , and normalize it to serve as the terrain coefficient h of the grid b .

8. The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling according to claim 2 is characterized in that: The saliency layer is marked by a saliency region extraction algorithm, as follows: Find the grids whose data values ​​are lower than the set value 1 or higher than the set value 2, and mark these grids as 1, and mark the rest of the grids as 0 to form a binary image; Use the connected domain algorithm in image processing to find the area larger than the set value and mark it as a significant area; When the AUV passes through a marked significant area, the historical data corresponding to the current surveying area is found according to the distribution characteristics of the grid data, and the coordinate difference of the corresponding grid is calculated to calibrate the map data and the positioning information of the AUV.

9. The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling according to claim 1 is characterized in that: The resampled path points in step S2 are distributed in areas where the confidence level is lower than the set value. The sampling points are evenly distributed at the edge of the unknown area, and the map is opened and closed to achieve edge connectivity. The contour of the unknown area can then be extracted through a mature edge extraction algorithm.

10. The AUV adaptive mapping path planning method based on hybrid grid map and dynamic sampling according to claim 1 is characterized in that: The evaluation function in step S3 assigns a value to each line segment connecting the path points: V i =W s S i +W M M i +W L L i (4) V i represents the evaluation function value of the i-th line segment, S i is the expected mapping efficiency, taking the average scan line width of the existing data As the expected scan line width of the sonar, count the number of The number of low-confidence grids N contained in the rectangular area with width , assuming that the line segment length is L, the expected mapping efficiency S i =N / L; M i is the interpolation gain, which counts the average distance from the candidate line segment to the known area; L i is the positioning gain, positioning gain L i The area where the expected scanning area intersects with the significant area is counted. If the area is greater than the threshold, L i =1; the smaller the intersection area, the smaller the L i The smaller the size, the less they intersect. i =0; W s , W M , W L is the weight of each item.

Images