Adaptive edge adjustment method and adjustment system for unmanned sweeper

Abstract

The present invention discloses an adaptive edge adjustment method and adjustment system for an unmanned road sweeper. The method comprises: step S1: obtaining an image of the outer edge of the sweeping brush on one side of the unmanned road sweeper and the road edge or curb; step S2: grayscale processing, edge detection, edge contour extraction, and forming a grayscale matrix; step S3: classifying the two contours, clustering to obtain corresponding contour cluster points, quickly looping through and calculating the minimum distance between the pixel cluster points in the two contours, obtaining the minimum distance deviation value, and converting the minimum distance of the pixel cluster points into an actual engineering deviation distance parameter; step S4: if the distance stability deviation value is 0, directly exit and re-perform real-time detection; otherwise, output to the control module, perform offset compensation on the original driving plan path or automatically adjust the sweeping brush deployment angle to achieve adaptive dynamic adjustment of the edge. The system is implemented based on the above method. The present invention has the advantages of high intelligence, precise adjustment, and a wide range of applications.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of unmanned sanitation equipment, and in particular to an adaptive edge adjustment method and adjustment system for an unmanned sweeper. Background Art

[0002] As the technology of unmanned sanitation vehicles continues to mature, their applications are becoming increasingly widespread. For unmanned sanitation vehicles, how to ensure that the onboard cleaning actuator can clean as close to the edge as possible is one of their key technologies and a key technical indicator.

[0003] In the existing technology, two methods are generally used. The first is to generate global path information through high-precision maps. When encountering a curb section during the cleaning process, the boundary position is obtained through boundary information to achieve curb-side cleaning. The second is to obtain curb and other boundary information through multi-sensor fusion to achieve curb cleaning.

[0004] However, in unmanned road sweepers, the above traditional technologies still have some technical deficiencies:

[0005] 1. The current algorithm for edge cleaning in unmanned road sweepers is not capable of adaptively and dynamically adjusting edge cleaning. After a sweeper has been in operation for a while, the brush wears out, the bristles shorten, and the edge cleaning effect deteriorates. A gap exists between the brush extension and the roadside, preventing it from fully contacting the roadside. Manual adjustment of the limiter is required to further expand the brush opening to achieve the same edge cleaning effect as before wear. Dynamic adaptive matching is not possible.

[0006] 2. The manual method is time-consuming and laborious. If you do not have enough control or experience in adjusting the opening of the sweeping brush, you need to repeatedly adjust the sweeping brush opening and match and debug it with the automatic driving system many times. The effect is not satisfactory and it is not intelligent enough.

[0007] 3. After replacing the sweeping brush, it is still necessary to re-set parameters and match them with the automatic driving system to achieve the edge cleaning effect, which is time-consuming and labor-intensive and not intelligent enough.

[0008] In other words, there is currently no relevant content on adaptively adjusting the edge cleaning technology within the life cycle of the sweeping brush. Summary of the Invention

[0009] The technical problem to be solved by the present invention is: in response to the technical problems existing in the prior art, the present invention provides an adaptive edge adjustment method and adjustment system for an unmanned sweeper with simple principle, high degree of intelligence, precise adjustment and wide range of applications.

[0010] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0011] An adaptive edge adjustment method for an unmanned sweeper, comprising:

[0012] Step S1: Acquire an image of the outer edge of the sweeping brush on one side of the unmanned sweeper and the road sideline or curb;

[0013] Step S2: grayscale processing is performed on the acquired image to detect edges and extract edge contours to form a grayscale matrix;

[0014] Step S3: The two contours are classified using the K-means clustering algorithm, and the corresponding contour cluster points are obtained using DBSCAN clustering. Then, the improved Euclidean distance algorithm is used to quickly loop through and calculate the minimum distance between the pixel cluster points in the two contours. After obtaining the stable minimum distance deviation value, the pixel cluster point distance is converted into the actual engineering deviation distance parameter;

[0015] Step S4: If the distance stability deviation value is 0, exit directly and re-perform real-time detection (start step S1); if the deviation is greater than 0, it is output to the control module of the unmanned sweeper to compensate for the offset of the original planned path or automatically adjust the deployment angle of the sweeping brush to achieve adaptive dynamic adjustment of the edge.

[0016] As a further improvement of the method of the present invention: in step S2, the acquired image is gray-scale processed by an image processing module, a Gaussian filter is used for blurring and noise reduction, and an edge detector is used to detect edges and extract edge contours to form a gray-scale matrix.

[0017] As a further improvement of the method of the present invention: in step S3, the edge contour classification method is the K-means clustering method, that is:

[0018]

[0019] in, Indicates the classification result; Indicates that there are n data points; Indicates that there are k initialized data centers.

[0020] As a further improvement of the method of the present invention: in step S3, the cluster point generation and classification method is the DBSCAN clustering algorithm, and a fixed domain radius is used for a given For the m objects in the inner area, a cluster of core objects is generated. The DBSCAN algorithm repeatedly searches for other objects that are directly density-reachable by the core object and merges the clusters that may be directly density-reachable until there are no other clusters around it, thus achieving the generation of cluster points of two contours.

[0021] As a further improvement of the method of the present invention: in step S3, the neighborhood parameters introduced by the DBSCAN cluster points are Introducing the standardized Euclidean distance formula, an improved standardized Euclidean distance formula is formed:

[0022]

[0023]

[0024] in, Refers to the brush outer edge contour cluster point vector, the number is 1 to n; Refers to the inner contour cluster point vector of the edge of the road or edge line, the number is 1 to m; is the domain parameter vector; Refers to the horizontal coordinate of the i-th cluster point; Refers to the horizontal coordinate of the j-th cluster point; Refers to the ordinate of the i-th cluster point; Refers to the ordinate of the jth cluster point The minimum distance should belong to the improved Euclidean distance of the distance set D.

[0025] As a further improvement of the method of the present invention, one or more cameras are arranged at the front right side of the vehicle body of the unmanned sweeper to obtain images of the sweeping brush of one side of the unmanned sweeper and the road edge.

[0026] As a further improvement of the method of the present invention: the road edge line includes one or more of a curb edge line, a curb edge line, and a sidewalk edge line.

[0027] The present invention further provides an adaptive edge adjustment system for an unmanned sweeper, comprising:

[0028] An image acquisition module is used to acquire images of the sweeping brush on one side of the unmanned sweeper and the road edge;

[0029] The image processing and calculation module is used to perform grayscale processing on the acquired image, detect edges and extract edge contours to form a grayscale matrix; the two contours are classified using the K-means clustering algorithm, and the corresponding contour cluster points are obtained using DBSCAN clustering. Then, the improved standardized Euclidean distance algorithm is used to quickly loop through the pixel cluster points within the two contours to calculate the minimum distance between the two contour pixel cluster points. After obtaining a stable minimum distance deviation value, the pixel cluster point distance is converted into an actual engineering deviation distance parameter; if the distance stability deviation value is 0, the module directly exits and re-performs real-time detection (starting step S1); if the deviation is greater than 0, the module outputs it to the autonomous driving system planning and control module of the unmanned sweeper;

[0030] The planning control module is used to control the actuator to perform offset compensation after obtaining the input value of the image processing calculation module to achieve smooth adaptive edge fitting.

[0031] As a further improvement of the system of the present invention: it also includes a command input end, which is a web page in the cloud, or an app of a mobile terminal, or software that is automatically powered on by default settings.

[0032] As a further improvement of the system of the present invention, the planning control module performs offset compensation on the driving path or adjusts the deployment angle of the sweeping brush to achieve adaptive dynamic adjustment of the edge.

[0033] Compared with the prior art, the advantages of the present invention are:

[0034] 1. The adaptive edge adjustment method and adjustment system of the unmanned road sweeper of the present invention have a simple principle, a high degree of intelligence, precise adjustment, a wide range of applications, and can be applied to all automatic driving road sweepers or road sweepers that require sweeping adjustment and edge cleaning.

[0035] 2. The present invention's adaptive edge adjustment method and system for unmanned road sweepers utilizes a camera mounted on the upper right front side of one side to capture images of the brush's outer edge and the curb, curb, or sidewalk boundary. Image processing and a distance detection algorithm are used to calculate the distance between the brush's outer edge and the curb, curb, or sidewalk boundary. After obtaining a stable deviation, the feedback planning and control system compensates for the offset in the direction of travel, dynamically adjusting the deviation in real time to achieve adaptive, dynamic edge-adjusted sweeping. This eliminates the need for manual adjustment and allows for adaptive edge-adjusted sweeping, ensuring a perfect fit for curbs and other boundaries.

[0036] 3. The key innovation of the adaptive edge adjustment method and adjustment system of the unmanned sweeper of the present invention is that it is based on visual detection and an improved distance measurement method, which realizes the adaptive adjustment of edge cleaning according to the wear of the sweeping brush during the life cycle of the sweeping brush, and always satisfies the edge cleaning function until the end of the sweeping brush life cycle. After replacing the new sweeping brush, there is no need to make any manual adjustment of the edge setting, and it can be used immediately after replacement. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a flow chart of the self-adaptive edge-adjusting cleaning method of the present invention.

[0038] Figure 2 It is a schematic diagram comparing the bristles of the adaptive edge cleaning brush before and after wear in a specific application example of the present invention.

[0039] Figure 3 It is a schematic diagram of the composition and execution flow of the adaptive edge adjustment cleaning system of the present invention.

[0040] Figure 4 It is a schematic diagram of a grayscale image and edge contour in a specific application example of the present invention.

[0041] Figure 5 It is a schematic diagram of two contours of K-means value classification in a specific application example of the present invention.

[0042] Figure 6 It is a schematic diagram of cluster point creation of the DBSCAN clustering algorithm in a specific application example of the present invention.

[0043] Figure 7 It is a schematic diagram of the improved Euclidean distance traversal calculation of short distances between clusters in a specific application example of the present invention.

[0044] Figure 8 It is a schematic diagram of adaptive edge adjustment in a specific application example of the present invention.

[0045] Figure 9 It is a schematic diagram of the composition and execution flow of the adaptive edge adjustment cleaning system of another embodiment of the present invention. DETAILED DESCRIPTION

[0046] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0047] like Figure 1 and Figure 2 As shown, the adaptive edge adjustment method of the unmanned sweeper of the present invention is mainly applicable to the unmanned sweeper, and is used to ensure that the sweeping brush of the cleaning actuator of the unmanned sweeper can be kept as close to the edge of the road as possible in real time during the cleaning operation; the method of the present invention includes:

[0048] Step S1: Acquire an image of the outer edge of a sweeping brush on one side of the unmanned sweeper and the road edge or curb; the road edge includes boundary lines such as curbs / curbs / sidewalk edges;

[0049] Step S2: grayscale processing is performed on the acquired image to detect edges and extract edge contours to form a grayscale matrix. Figure 4 ;

[0050] Step S3: The two contours are classified using the K-means clustering algorithm, and the corresponding contour cluster points are obtained using DBSCAN clustering. Then, the improved Euclidean distance algorithm is used to quickly loop through the pixel cluster points in the two contours to calculate the minimum distance between the two contour pixel cluster points. After obtaining the stable minimum distance deviation value, the pixel cluster point distance is converted into the actual engineering deviation distance parameter;

[0051] Step S4: If the distance stability deviation value is 0, exit directly and re-perform real-time detection (start step S1); if the deviation is greater than 0, it is output to the autonomous driving system planning and control module of the unmanned sweeper to compensate for the offset of the driving path or adjust the deployment angle of the sweeping brush to achieve dynamic and smooth reduction of the boundary deviation value between the outermost contour of the sweeping brush and the boundary line / curb / sideline, etc., to achieve an adaptive dynamic adjustment of the edge effect.

[0052] In a specific application example, the present invention only requires setting one or more cameras on the right front of the unmanned sweeper to obtain boundary line images such as the brush on one side and the curb / teeth / sidewalk edge.

[0053] In a specific application example, in step S2, the present invention performs grayscale processing on the acquired image through an image processing module, blurs and reduces noise with a Gaussian filter, and uses an edge detector to detect edges and extract edge contours to form a grayscale matrix.

[0054] See also Figure 2 As shown, after an unmanned sweeper has been operating for a period of time, the brush wears out and becomes shorter, leaving a gap between the actual edge and the actual edge. By employing the above-described method, the edge can be adaptively and dynamically adjusted to achieve a clean sweep. In other words, with the adjustment method of the present invention, manual adjustment of the brush opening is no longer necessary, and no further adjustment is required throughout the lifecycle of the brush. Furthermore, after replacing a new brush, the system can operate directly without the need for parameter settings or matching with the automated driving system.

[0055] In a specific application example, the clustering algorithm in step S3 is described in Figure 5 , the edge contour classification method is the K-means clustering method, namely:

[0056]

[0057] Indicates the classification result;

[0058] Indicates that there are n data points;

[0059] Indicates that there are k initialized data centers.

[0060] In a specific application example, the cluster point generation and classification method in step S3 is the DBSCAN clustering algorithm, see Figure 6 , a fixed domain radius is used for a given In the inner area objects, generating a cluster of core objects. The DBSCAN algorithm repeatedly searches for other objects that are directly accessible to the core object, and merges the clusters that may be directly accessible to the core object until there are no other clusters around it, thus achieving the generation of cluster points of two contours.

[0061] Since the position error of the cluster point itself is related to the neighborhood and the number of core points in the neighborhood area. In specific application examples, see Figure 8 In step S3, the distance calculated based on the cluster points in the present invention still has a certain error. If it is not optimized, the edge adjustment effect will not be perfect. In order to make the deviation distance calculation result more accurate, the neighborhood parameter introduced by the DBSCAN cluster point is Introducing the standardized Euclidean distance formula to improve calculation accuracy, improve the standardized Euclidean distance formula:

[0062]

[0063]

[0064] in, Refers to the brush outer edge contour cluster point vector, the number is 1 to n;

[0065] Refers to the inner contour cluster point vector of the edge of the road or edge line, the number is 1 to m;

[0066] is the domain parameter vector;

[0067] Refers to the horizontal coordinate of the i-th cluster point;

[0068] Refers to the horizontal coordinate of the j-th cluster point;

[0069] Refers to the ordinate of the i-th cluster point;

[0070] Refers to the ordinate of the jth cluster point;

[0071] The minimum distance should belong to the improved Euclidean distance of the distance set D.

[0072] In a specific application example, in step S4, see Figure 8 After receiving a stable deviation value, the system performs a smooth offset on the cleaning path, controlling the driving system to slowly reduce the deviation. When the real-time feedback deviation value approaches 0, it stops offset compensation and continues cleaning for a certain distance. If the effect remains consistent and close to the edge, the adaptive adjustment system exits, retaining the offset parameter value and writing it into the control system.

[0073] See also Figure 3 The present invention further provides a camera-based adaptive edge adjustment system, which includes: the present invention is based on a road sweeper with an automatic driving system function, mainly comprising a function start terminal part, an image acquisition module, an image processing and calculation module, and a planning and control module. Among them:

[0074] An image acquisition module is used to acquire images of the sweeping brush on one side of the unmanned sweeper and the road edge;

[0075] The image processing and calculation module is used to perform grayscale processing on the acquired image, detect edges and extract edge contours to form a grayscale matrix; the two contours are classified using the K-means clustering algorithm, and the corresponding contour cluster points are obtained using DBSCAN clustering. Then, the improved Euclidean distance algorithm is used to quickly loop through the pixel cluster points within the two contours to calculate the minimum distance between the two contour pixel cluster points. After obtaining a stable minimum distance deviation value, the pixel cluster point distance is converted into an actual engineering deviation distance parameter; if the distance stability deviation value is 0, the module directly exits and re-performs real-time detection (starting step S1); if the deviation is greater than 0, it is output to the autonomous driving system planning and control module of the unmanned sweeper;

[0076] The planning control module is used to control the actuator to perform offset compensation after obtaining the input value of the image processing calculation module to achieve smooth adaptive edge fitting.

[0077] As a preferred embodiment, the present invention also includes a command input terminal, which can be a web page in the cloud, or an app in a tablet computer, or the software default setting is that the automatic driving is powered on and the adaptive edge adjustment function is automatically started in real time.

[0078] In a specific application example, the adjustment system of the present invention is activated through the command input end, and the automatic driving road sweeper starts to operate along the edge. After the automatic driving road sweeper works stably along the edge, the image acquisition module (camera) starts to work, and obtains the automatic driving system's edge cleaning image in real time and transmits it to the image processing and calculation module through the LVDS data cable (coaxial cable).

[0079] See also Figure 9 In another embodiment, after obtaining a stable distance deviation value, the sweeping brush deployment mechanism control module controls the sweeping brush deployment execution motor to adjust the opening of the deployment mechanism in real time to achieve the overlap of the sweeping brush with the boundaries such as the curb, thereby achieving the purpose of complete edge cleaning.

[0080] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. An adaptive edge adjustment method for an unmanned sweeper, characterized in that: include: Step S1: Acquire an image of the outer edge of the sweeping brush on one side of the unmanned sweeper and the road sideline or curb; Step S2: grayscale processing is performed on the acquired image to detect edges, extract edge contours, and form a grayscale matrix; Step S3: Classify the two contours using the K-means clustering algorithm, and use DBSCAN clustering to obtain the corresponding contour cluster points. Then, use the improved standardized Euclidean distance algorithm to quickly loop through and calculate the minimum distance between the two contour pixel cluster points to obtain the minimum distance deviation value. The minimum distance of the pixel cluster points is converted into the actual engineering deviation distance parameter; Step S4: If the distance stability deviation value is 0, exit directly and re-perform real-time detection; if the deviation is greater than 0, it is output to the control module of the unmanned sweeper to compensate for the offset of the original driving plan path or automatically adjust the sweeping angle to achieve adaptive dynamic adjustment of the edge.

2. The method for adaptively adjusting the edge of an unmanned sweeper according to claim 1, characterized in that: In step S2, the acquired image is gray-scale processed by an image processing module, a Gaussian filter is used for blurring and noise reduction, and an edge detector is used to detect edges and extract edge contours to form a gray-scale matrix.

3. The method for adaptively adjusting the edge of an unmanned sweeper according to claim 1, characterized in that: In step S3, since the edge contours are relatively clear and continuous, the two contours are quickly divided here using the K-means clustering method, namely: in, Indicates the classification result; Indicates that there are n data points; Indicates that there are k initialized data centers.

4. The method for adaptively adjusting the edge of an unmanned sweeper according to claim 1, characterized in that: In step S3, cluster points are generated and classified in each contour using the DBSCAN clustering algorithm. In the inner area objects, generating a cluster of core objects. The DBSCAN algorithm repeatedly searches for other objects that are directly density-reachable by the core object, and merges the clusters that may be directly density-reachable until there are no other clusters around, thus realizing the generation of cluster points for the two contours.

5. The method for adaptively adjusting the edge of an unmanned sweeper according to claim 4, characterized in that: In step S3, the neighborhood parameters introduced by the DBSCAN cluster points are Introduced into the Euclidean distance formula, an improved Euclidean distance formula is formed: in, Refers to the brush outer edge contour cluster point vector, the number is 1 to n; Refers to the inner contour cluster point vector of the edge of the road or edge line, the number is 1 to m; is the domain parameter vector; Refers to the horizontal coordinate of the i-th cluster point; Refers to the horizontal coordinate of the j-th cluster point; Refers to the ordinate of the i-th cluster point; Refers to the ordinate of the jth cluster point; It means that the minimum distance should belong to the improved Euclidean distance of distance set D.

6. The method for adaptively adjusting the edge of an unmanned sweeper according to any one of claims 1 to 5, characterized in that: One or more cameras are set on the right front of the unmanned sweeper to obtain images of the sweeping brush on one side of the unmanned sweeper and the road sideline.

7. The method for adaptively adjusting the edge of an unmanned sweeper according to any one of claims 1 to 5, characterized in that: The road edge line includes one or more of a curb edge line, a curb edge line, and a sidewalk edge line.

8. An adaptive edge adjustment system for an unmanned sweeper, characterized in that: include: An image acquisition module is used to acquire images of the sweeping brush on one side of the unmanned sweeper and the road edge; The image processing and calculation module is used to perform grayscale processing on the acquired image, detect edges and extract edge contours to form a grayscale matrix; the two contours are classified using the K-means clustering algorithm, and the corresponding contour cluster points are obtained using DBSCAN clustering. Then, the improved standardized Euclidean distance algorithm is used to quickly loop through the pixel cluster points within the two contours to calculate the minimum distance between the two contour pixel cluster points. After obtaining the stable minimum distance deviation value, the pixel cluster point distance is converted into the actual engineering deviation distance parameter; if the distance stability deviation value is 0, it exits directly; if the deviation is greater than 0, it is output to the autonomous driving system planning and control module of the unmanned sweeper; The planning control module is used to control the actuator to perform offset compensation after obtaining the input value of the image processing calculation module to achieve smooth adaptive edge fitting.

9. The adaptive edge adjustment system for an unmanned sweeper according to claim 8, characterized in that: It also includes a command input terminal, which is a web page in the cloud, or an app on a mobile terminal, or software that is set to power on automatically by default.

10. The adaptive edge adjustment system for an unmanned sweeper according to claim 8 or 9, characterized in that: The planning control module compensates for the offset of the driving path or adjusts the deployment angle of the sweeping brush to achieve adaptive dynamic adjustment of the edge.

Images