Large-scale irregular boundary fast fitting method based on guide point neighborhood search

Abstract

The application discloses a large-range irregular boundary fast fitting method based on a guide point neighborhood search, and comprises the following steps: S1, drawing a guide line segment in a drivable area of a grid map, and taking a point as an initial guide point on the guide line segment; S2, taking all the initial guide points as centers to perform neighborhood search, obtaining boundary points corresponding to all the initial guide points, and sequentially storing the boundary points in a boundary point array; S3, taking a guide point corresponding to a terminal boundary point in the boundary point array as an extended guide point, taking the extended guide point as a center to perform neighborhood search within a certain range, and gradually moving the extended guide point; S4, storing boundary points obtained by performing neighborhood search according to the extended guide point in a terminal of the boundary point array; S5, judging whether boundary point acquisition is completed, if yes, performing S6, and if not, returning to S3; and S6, performing boundary fitting according to the boundary point array, and obtaining a drivable area boundary.

Description

[0001] Technology Neighborhood This invention relates to the field of intelligent driving technology, and more specifically to a method for fast fitting of large-scale irregular boundaries based on guide point neighborhood search. Background Technology

[0002] In recent years, with the development of intelligent driving technology, drivable area boundary detection has received increasing attention. It can provide a basis for intelligent vehicles' path planning, behavior decision-making, and motion planning, helping intelligent vehicles to achieve semi-autonomous or fully autonomous driving within safe areas.

[0003] Currently, drivable area data is commonly acquired through cameras, LiDAR, or a fusion of both. Using cameras with a vision-based approach is low-cost, but its effectiveness is significantly affected by lighting and weather conditions. Fusion methods require consideration of LiDAR point cloud and image registration. Therefore, LiDAR is more commonly used for drivable area detection. LiDAR data acquisition offers advantages such as strong anti-interference capabilities, high accuracy, high speed, high resolution, and rich information, but it also presents significant challenges. The acquired point cloud data is extremely dense, especially in large-scale environments such as mining areas with wall boundaries. The amount of data acquired is enormous. When converting the point cloud data into a raster map for path planning, the need to repeatedly search through the massive amount of data to confirm the safety of path points results in low planning efficiency. If the most relevant drivable area boundary information could be quickly extracted from the large dataset beforehand, direct path planning using this boundary data would save considerable time.

[0004] Commonly used boundary extraction methods include latitude and longitude scanning, grid division, image processing, and minimum convex edge method. However, latitude and longitude scanning and grid division still require searching a large space, while image processing and minimum convex edge method have the problem of large computational load. Neither is suitable for large-scale raster maps and cannot quickly and accurately extract the boundary information of the drivable area of ​​interest. Summary of the Invention

[0005] The purpose of this invention is to provide a fast fitting method for large-scale irregular boundaries based on guide point neighborhood search, so as to quickly and accurately extract the boundary information of the drivable area of ​​intelligent vehicles in a large-scale grid map.

[0006] To address the aforementioned technical problems, this invention provides a technical solution: a fast fitting method for large-scale irregular boundaries based on guide point neighborhood search, comprising the following steps: S1. Draw guide lines in the drivable area of ​​the grid map according to certain rules, and take points at certain intervals on the guide lines as initial guide points; S2. Sequentially perform neighborhood searches within a certain range using all initial guiding points as centers, obtain the boundary points corresponding to all initial guiding points, and store them sequentially in the boundary point array; S3. Take the leading point corresponding to the last boundary point in the boundary point array as the expansion leading point, and perform a neighborhood search within a certain range with the expansion leading point as the center. The expansion leading point moves according to its own position, the position of several boundary points at the end of the boundary point array, and the neighborhood search results of the expansion leading point. S4. Store the boundary points obtained by neighborhood search based on the extended guide points into the end of the boundary point array. S5. Determine whether the boundary point acquisition is complete based on the distance between the end boundary point and the beginning boundary point in the boundary point array at this time. If yes, proceed to S6; otherwise, return to S3. S6. Perform boundary fitting based on the boundary point array to obtain the drivable area boundary.

[0007] The method for fast fitting of large-scale irregular boundaries based on guide point neighborhood search according to claim 1 is characterized in that: the process of expanding the movement of the guide point in S3 is specifically as follows: S301. First, determine the slope of the line connecting the last two boundary points in the boundary point array. If the slope does not exist, execute S302; otherwise, execute S303. S302, then the guide point will be moved longitudinally a certain distance; S303. When the slope of the connecting line is less than or equal to a certain value, the expansion guide point is first moved horizontally a certain distance, and then the expansion guide point is moved vertically according to the slope of the connecting line and the horizontal moving distance of the expansion guide point; otherwise, the expansion guide point is first moved vertically a certain distance, and then the expansion guide point is moved horizontally according to the slope of the connecting line and the vertical moving distance of the expansion guide point. S304. If the extended guide point after moving according to the slope of the connecting line finds a boundary point within a certain range, then execute S305; otherwise, execute S306. S305. If the distance between the boundary point searched in S304 and the boundary point at the end of the current boundary point array is greater than a certain value, then the boundary point searched in S304 is defined as the opposite boundary point relative to the boundary point at the end of the current boundary point array. At this time, the expansion guide point is moved a certain distance towards the boundary point at the end of the current boundary point array along the normal direction of the boundary of the traversable area boundary at the position of the boundary point at the end of the current boundary point array, and then returns to S304. If the distance between the boundary point searched in S304 and the boundary point at the end of the current boundary point array is less than or equal to a certain value, then the boundary point searched in S304 is defined as the same boundary point relative to the boundary point at the end of the current boundary point array. Then it is determined whether the boundary point searched in S304 and the boundary point at the end of the current boundary point array are the same boundary point. If not, S4 is executed; if so, S301 is returned. S306. Move the extended guide point a certain distance along the normal direction of the boundary point at the end of the boundary point array of the current drivable area boundary, and return to S304.

[0008] According to the above scheme, in S305, when the boundary point is the opposite boundary point to the end of the boundary point array at this time, the distance that the expansion guide point moves is 1 / 4 of the distance between the expansion guide point and the boundary point searched in S304.

[0009] According to the above scheme, the distance that the expansion guide point moves in S306 is 1 / 4 of the distance between the expansion guide point and the end of the boundary point in the boundary point array at this time.

[0010] According to the above scheme, the neighborhood search method specifically involves finding the boundary point closest to the center of the rectangle within a rectangular area centered on the initial guide point or the extended guide point.

[0011] According to the above scheme, the method also includes a noise reduction process for the boundary point array before boundary fitting. Specifically, boundary points in the surrounding 8 grids that are occupied by obstacles and do not meet a certain number of requirements are regarded as noise and removed.

[0012] According to the above scheme, in S5, when the distance between the end boundary point and the beginning boundary point in the boundary point array does not exceed the distance represented by two grid cells, the boundary point acquisition is considered complete.

[0013] According to the above scheme, B-spline curves are used for boundary fitting in S6.

[0014] According to the above scheme, the boundary is segmented during the boundary fitting process, and only boundary points at a certain interval are selected for fitting in each segment.

[0015] A vehicle that uses the drivable region boundary obtained by the fast fitting method for large-scale irregular boundaries based on guide point neighborhood search described above for path planning.

[0016] The beneficial effects of this invention are: by drawing guide line segments, partial boundary points are quickly obtained; then, by moving and expanding the guide points, a neighborhood search is performed with the expanded guide points as the center; and based on the results of the neighborhood search, the expanded guide points are gradually moved to complete the acquisition of all boundary points. This method directly utilizes a raster map without requiring other information, resulting in low computational cost. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method for fast fitting of large-scale irregular boundaries based on guide point neighborhood search according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the movement of the guide point according to an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0019] To overcome the shortcomings of existing technologies, this invention proposes a method for rapidly fitting large-scale irregular boundaries based on a guide point neighborhood search. First, a starting guide point is set to determine the location of the target area boundary to be extracted. Then, a neighborhood search is used to find the coordinates of the boundary point closest to the guide point to determine the initial direction of the boundary. Next, automatic searching is initiated, and the guide point is automatically moved and adjusted according to specific circumstances, thereby achieving the goal of automatically and rapidly searching for boundary points. This method can quickly and accurately extract boundary information of drivable areas for intelligent vehicles in large-scale raster maps.

[0020] See Figure 1 , Figure 2 The specific technical solution of the present invention is as follows: The fast fitting method for large-scale irregular boundaries based on guide point neighborhood search can be divided into three parts: boundary selection, automatic search, and fitting and saving. The boundary selection part involves drawing two points in the target drivable area of ​​the grid map using the right mouse button. The two points must simultaneously meet two conditions: the line connecting them does not touch any obstacles, and every point on the line can find a boundary point within a specified range. This ensures safety and proximity to the target boundary. Each point on the line is used as a guide point for finding the boundary. The neighborhood search method is used to find the coordinates of the boundary point closest to the guide point. After deduplication, the coordinates are stored in a boundary point array, thus obtaining the initial direction of the boundary. The neighborhood search method involves finding the nearest boundary point to the center within a small square formed by expanding a specified range in four directions (up, down, left, and right) centered on the guide point. The number of grid cells around the found boundary point that are occupied by obstacles is checked to determine if it is a noise point.

[0021] The core of the automatic search section is to automatically move the guide point to the next suitable position based on the found boundary direction. The specific steps are as follows: S1: Calculate the slope of the line connecting the last two points in the boundary point array to determine the boundary direction. If the slope does not exist, execute S2; otherwise, execute S3. S2: Move the previous guiding point up / down by c(m), update its y-coordinate value, and keep the x-coordinate unchanged, then proceed to S4; S3: Calculate the equation of the line passing through the previous guide point based on the slope. At the same time, classify the two moving schemes according to the magnitude of the slope. If the absolute value of the slope does not exceed 1, the previous guide point is first moved left / right by c (m), the value of x is updated, and then the value of y is adjusted according to the equation of the line. If the absolute value of the slope is greater than 1, the point is first moved up / down by c (m), the value of y is updated, and then the value of x is obtained according to the equation of the line. S4: Determine whether the new guiding point can find the boundary point through neighborhood search after the move. If the boundary point can be found, execute S5; otherwise, execute S7. S5: Determine whether the new boundary point is on the same side or the opposite side based on the distance from the previous boundary point. If it is on the opposite side, adjust the coordinates of the guide point along the boundary normal and towards the target boundary. The adjustment distance is 1 / 4 of the distance from the guide point to the corresponding boundary point. After adjustment, return to S4 to check again. If it is on the same side, check whether this point is the same as the previous boundary point. If they are the same, return to S1 to update the position of the guide point again. If they are different, it means that a boundary point that meets the requirements has been found, and it is stored in the boundary point array. S6: Calculate the distance from the new guide point to the corresponding boundary point. The guide point and the boundary need to be kept close to each other to ensure that the movement trajectory of the guide point always follows the direction of the boundary. If this distance exceeds the specified range, adjust the guide point to the trisection point on the line connecting this point and the corresponding boundary point. Then check whether the distance is within the specified range again. If this distance is within the specified range, store it in the guide point array and proceed to S8. S7: Adjust the coordinates of the guide point along the boundary normal and toward the target boundary. The adjustment distance is 1 / 4 of the distance from the guide point to the previous boundary point. After adjustment, return to S4 to judge again. S8: Calculate the distance from the latest boundary point to the first boundary point. If it does not exceed the actual distance represented by two grids, it means that a complete boundary has been found. Save the data and end the automatic search. If it still exceeds the actual distance represented by two grids, return to S1 to continue moving the guide point to find a new boundary point.

[0022] The up / down or left / right movement refers to selecting the specific movement direction based on the relative position of the last two points in the boundary point array.

[0023] After the automatic search ends, the process proceeds to the fitting and saving section. The boundary is segmented according to the number of boundary points, and B-spline curves are used for segmented boundary fitting. To further improve the fitting calculation speed, the number of control points is reduced, and only boundary points at certain intervals are selected for fitting each boundary segment. After fitting, the equally spaced boundary data is saved, and the saved boundary data can be directly extracted to draw the drivable area of ​​the intelligent vehicle.

[0024] The technical advantages of this invention include: narrowing the search range through a guide point neighborhood search method to quickly locate boundary points; automatically searching for corresponding complete boundaries in sequence after setting the starting point and completing the target boundary selection, improving efficiency, and directly drawing complete boundaries using the search results; extracting the most relevant intelligent vehicle drivable area boundary information from the massive data volume of a large-scale raster map, compressing the data, and facilitating subsequent improvements in the efficiency of intelligent vehicle path planning; applicable to all raster environments with one or more closed boundaries, allowing direct processing of raster maps without the need for other information.

[0025] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A fast fitting method for large-scale irregular boundaries based on guide point neighborhood search, characterized by: Includes the following steps, S1. Draw guide lines in the drivable area of ​​the grid map according to certain rules, and take points at certain intervals on the guide lines as initial guide points; S2. Sequentially perform neighborhood searches within a certain range using all initial guiding points as centers, obtain the boundary points corresponding to all initial guiding points, and store them sequentially in the boundary point array; S3. Take the leading point corresponding to the last boundary point in the boundary point array as the expansion leading point, and perform a neighborhood search within a certain range with the expansion leading point as the center. The expansion leading point moves according to its own position, the position of several boundary points at the end of the boundary point array, and the neighborhood search results of the expansion leading point. S4. Store the boundary points obtained by neighborhood search based on the extended guide points into the end of the boundary point array. S5. Determine whether the boundary point acquisition is complete based on the distance between the end boundary point and the beginning boundary point in the boundary point array at this time. If yes, proceed to S6; otherwise, return to S3. S6. Perform boundary fitting based on the boundary point array to obtain the drivable area boundary; The specific process of expanding the guide point movement in S3 is as follows: S301. First, determine the slope of the line connecting the last two boundary points in the boundary point array. If the slope does not exist, execute S302; otherwise, execute S303. S302, then the guide point will be moved longitudinally a certain distance; S303. When the slope of the connecting line is less than or equal to a certain value, the expansion guide point is first moved horizontally a certain distance, and then the expansion guide point is moved vertically according to the slope of the connecting line and the horizontal moving distance of the expansion guide point; otherwise, the expansion guide point is first moved vertically a certain distance, and then the expansion guide point is moved horizontally according to the slope of the connecting line and the vertical moving distance of the expansion guide point. S304. If the extended guide point after moving according to the slope of the connecting line finds a boundary point within a certain range, then execute S305; otherwise, execute S306. S305. If the distance between the boundary point searched in S304 and the boundary point at the end of the current boundary point array is greater than a certain value, then the boundary point searched in S304 is defined as the opposite boundary point relative to the boundary point at the end of the current boundary point array. At this time, the expansion guide point is moved a certain distance towards the boundary point at the end of the current boundary point array along the normal direction of the boundary of the traversable area boundary at the position of the boundary point at the end of the current boundary point array, and then returns to S304. If the distance between the boundary point searched in S304 and the boundary point at the end of the current boundary point array is less than or equal to a certain value, then the boundary point searched in S304 is defined as the same boundary point relative to the boundary point at the end of the current boundary point array. Then it is determined whether the boundary point searched in S304 and the boundary point at the end of the current boundary point array are the same boundary point. If not, S4 is executed; if so, S301 is returned. S306. Move the extended guide point a certain distance along the normal direction of the boundary point at the end of the boundary point array of the current drivable area boundary, and return to S304.

2. The method for fast fitting of large-scale irregular boundaries based on guide point neighborhood search according to claim 1, characterized in that: In S305, when the boundary point is the opposite boundary point to the end of the boundary point array at this time, the distance the expansion guide point moves is 1 / 4 of the distance between the expansion guide point and the boundary point searched in S304.

3. The method for fast fitting of large-scale irregular boundaries based on guide point neighborhood search according to claim 1, characterized in that: In S306, the distance the extended guide point moves is 1 / 4 of the distance between the extended guide point and the end of the boundary point in the current boundary point array.

4. The method for fast fitting of large-scale irregular boundaries based on guide point neighborhood search according to claim 1, characterized in that: The neighborhood search method specifically involves finding the boundary point closest to the center of a rectangle centered on the initial or extended guide point.

5. The method for fast fitting of large-scale irregular boundaries based on guide point neighborhood search according to claim 1, characterized in that: The method also includes a denoising process for the boundary point array before boundary fitting. Specifically, boundary points in the surrounding 8 grids that are occupied by obstacles and do not meet a certain number of requirements are regarded as noise and removed.

6. The method for fast fitting of large-scale irregular boundaries based on guide point neighborhood search according to claim 1, characterized in that: In S5, the boundary point acquisition is completed when the distance between the end boundary point and the beginning boundary point in the boundary point array does not exceed the distance represented by two grid cells.

7. The method for fast fitting of large-scale irregular boundaries based on guide point neighborhood search according to claim 1, characterized in that: B-spline curves are used for boundary fitting in S6.

8. The method for fast fitting of large-scale irregular boundaries based on guide point neighborhood search according to claim 7, characterized in that: During the boundary fitting process, the boundary is segmented, and only boundary points at certain intervals are selected for fitting in each segment.

9. A car, characterized in that: The vehicle uses the drivable area boundary obtained by the fast fitting method based on guide point neighborhood search for large-scale irregular boundaries as described in any one of claims 1-8 for path planning.

Images