Method and system for detecting point cloud shape features

Abstract

The application discloses a kind of detection method and system based on point cloud shape feature.The detection method includes the following steps:S1: obtaining 2D point cloud data;S2: the data cutting of 2D point cloud data obtained in step S1 is carried out;S3: the data after cutting in step S2 is carried out mean filtering;S4: the data after mean filtering in step S3 is obtained by Hough transform, target line segment;And S5: target line segment is sorted, and adjacent target line segment between angle relationship is extracted target shape feature.The main beneficial effects of the application are: 1) the single-line laser radar on AGV for positioning navigation and obstacle avoidance is used for shelf detection positioning in the scheme, which reduces the additional sensor and processor hardware cost;2) the detection precision of single-line laser on shelf in the scheme is high, fast, and the structure is simple, stable, reliable, and not affected by environmental light changes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic guided vehicle positioning detection, in particular to a detection method and system based on point cloud shape features. BACKGROUND

[0002] In the process of latent lifting carrying AGV (automatic guided vehicle) carrying material boxes, the material boxes are generally placed on a shelf slightly higher than the AGV. The AGV needs to move to the shelf, lift the material box from the bottom, and carry it on the top of the AGV. This requires the AGV to accurately identify the position of the shelf to stably carry out the latent lifting.

[0003] For the detection of the shelf, generally based on stereo vision, such as ToF, structured light or binocular camera, the processor will extract and match features in the collected 3D point cloud depth image to detect the shape and position of the shelf.

[0004] There are also some solutions to place a two-dimensional code on the column of the shelf, so that the camera on the AGV can determine the position of the shelf by detecting the two-dimensional code.

[0005] The disadvantages of the method based on stereo vision are: the depth camera used to collect stereo vision data has high cost, and the amount of 3D point cloud data collected is large, which increases the difficulty of algorithm development and the hardware cost of the processor.

[0006] The disadvantages of the method based on two-dimensional code are: the installation of two-dimensional code consumes additional manpower, and manual maintenance is also required to prevent the two-dimensional code from falling off or being damaged, increasing the later cost.

[0007] The above-mentioned vision-based methods are easily affected by changes in environmental light. SUMMARY

[0008] The embodiments of the present application provide a detection method and system based on point cloud shape features, which can realize low-cost detection of the shelf of the lifting carrying AGV, without additional hardware to realize detection and positioning of the shelf, thereby saving the additional hardware cost, avoiding the installation and maintenance of the two-dimensional code, and using 2D point cloud data instead of 3D point cloud and environmental pictures collected by two-dimensional code camera, so the data processing complexity will be significantly reduced, thereby reducing the cost of the data processor and not being affected by changes in environmental light. The present application solves the technical problems of large computation, high equipment cost, high maintenance cost and influence of environmental light changes in the current shelf detection and positioning of the lifting carrying AGV.

[0009] The embodiments of the present application provide a detection method based on point cloud shape features, comprising the following steps:

[0010] S1: obtaining 2D point cloud data;

[0011] S2: data clipping on the 2D point cloud data obtained in step S1;

[0012] S3: mean filtering on the data clipped in step S2;

[0013] S4: obtaining target line segments through Hough transformation on the data mean filtered in step S3; and

[0014] S5: sorting the target line segments and extracting target shape features according to the angle relationship between adjacent target line segments.

[0015] In some embodiments, the specific method of sorting the target line segments comprises: taking the front of the radar as 0 degrees, and taking the counterclockwise direction as the positive direction, and sorting the adjacent target line segments in the order from small to large according to the angle between the adjacent target line segments.

[0016] In some embodiments, the method of obtaining the target shape features comprises: according to all the line segments obtained, sequentially calculating whether two continuous target line segments form an L shape, whether three continuous target line segments form a U shape, and whether the length of the target line segments meets the requirements; the two target line segments forming an L shape deny the determination standard that the angle range of the two straight lines is 90±5 degrees, and the three continuous target line segments forming a U shape are based on the two target line segments forming an L shape, and whether the three target line segments meet the condition that two parallel target line segments are connected at the two ends of another target line segment.

[0017] In some embodiments, the target shape features are a U shape formed by three target line segments, and the target shape features are output as the result of real-time positioning detection of the shelf.

[0018] In some embodiments, the method of obtaining 2D point cloud data comprises: using the angle and distance of the radar scanning to the obstacle points.

[0019] In some embodiments, the specific method of mean filtering is: directly deleting the aggregated point cloud data whose difference from the mean value is greater than a first threshold value.

[0020] In some embodiments, before the step of obtaining 2D point cloud data, the method further comprises:

[0021] setting an AGV body, a processor mounted on the AGV body, and a plurality of single-line radars, wherein the plurality of single-line radars are used to scan the environment around the shelf to obtain 2D point cloud data, and the processor is used to obtain the 2D point cloud data and sequentially perform data clipping, mean filtering, Hough transformation, and extraction of target shape features on the obtained 2D point cloud data; and

[0022] The middle part of the goods shelf is the running path of the AGV body; the goods shelf is provided with cover plates on three sides, and only the front part is provided with an opening to form a U-shaped structure, the AGV body can enter the goods shelf from the opening position, and a material box is arranged above the U-shaped structure of the goods shelf.

[0023] According to the detection method based on the point cloud shape features, the application further provides a detection system based on the point cloud shape features, which comprises an AGV body, a processor mounted on the AGV body, and a plurality of single-line radars for scanning the surrounding environment of the goods shelf to obtain 2D point cloud data, and the processor is used for acquiring the 2D point cloud data and sequentially performing data cutting, mean filtering, Hough transformation and target shape feature extraction on the acquired 2D point cloud data.

[0024] In some embodiments, the plurality of single-line radars are arranged at the front end of the AGV body, and the scanning planes of the plurality of single-line radars are parallel to the bottom surface of the AGV body.

[0025] In some embodiments, the detection system based on the point cloud shape features further comprises a goods shelf, the middle part of the goods shelf is the running path of the AGV body; the goods shelf is provided with cover plates on three sides, and only the front part is provided with an opening to form a U-shaped structure, the AGV body can enter the goods shelf from the opening position, and a material box is arranged above the U-shaped structure of the goods shelf.

[0026] The detection method and system based on the point cloud shape features provided by the embodiments of the application utilize the existing single-line laser radar on the jacking and carrying AGV for positioning and navigation to detect and position the goods shelf, thereby saving the additional hardware cost and avoiding the installation and maintenance of the two-dimensional code. Since the amount of 2D point cloud data collected by the single-line laser radar is far less than the amount of 3D point cloud collected by the stereoscopic vision sensor and the amount of environment pictures collected by the two-dimensional code camera, the data processing complexity will be significantly reduced, thereby reducing the cost of the data processor and avoiding the influence of the change of environmental light.

[0027] Therefore, the main beneficial effects of the application are as follows: 1) the scheme utilizes the single-line laser radar on the AGV for positioning, navigation and obstacle avoidance to detect and position the goods shelf, thereby reducing the additional sensor and processor hardware cost; 2) the scheme utilizes the single-line laser to detect the goods shelf with high precision and high speed, and the structure is simple, stable and reliable, and is not affected by the change of environmental light. BRIEF DESCRIPTION OF DRAWINGS

[0028] The technical scheme and other beneficial effects of the application will be apparent through the following detailed description of the specific embodiments of the application with reference to the accompanying drawings.

[0029] Figure 1A structural schematic diagram of a detection system based on point cloud shape features is provided for the embodiments of the present application.

[0030] Figure 2 A structural schematic diagram of relative positions of a shelf and a material box in a detection system based on point cloud shape features is provided for the embodiments of the present application.

[0031] Figure 3 A flowchart of a detection method based on point cloud shape features is provided for the embodiments of the present application.

[0032] Figure 4 A structural schematic diagram of a detection system based on point cloud shape features is provided for the embodiments of the present application. Figure 3 A structural schematic diagram of a detection system based on point cloud shape features is provided for the embodiments of the present application.

[0033] The identification in the figure is as follows:

[0034] A detection system based on point cloud shape features 10,

[0035] An AGV body 1, a single-line radar 2, a carrying plate 3,

[0036] Wheels 4, a shelf 5, a material box 6,

[0037] A scanning plane 21, a cover plate 51, an opening 52. DETAILED DESCRIPTION

[0038] The technical solutions in the embodiments of the present application will be clearly and completely described in combination with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0039] In the description of the present application, it should be understood that the orientations or positional relationships indicated by the terms “up”, “down”, “front”, “back”, “left”, “right”, “vertical”, “horizontal”, “top”, “bottom”, “inner”, “outer”, “clockwise”, “counterclockwise” and the like are the orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and therefore cannot be understood as indicating or implying that the devices or elements indicated must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms “first” and “second” are only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features limited by “first” and “second” can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of “multiple” is two or more, unless otherwise specifically limited.

[0040] Specifically, please refer toFigure 1 、 Figure 2 The application also provides a detection system 10 based on point cloud shape features, which comprises an AGV body 1, a processor (not shown) mounted on the AGV body 1, and a plurality of single-line radars 2 for scanning the surrounding environment of a shelf to obtain 2D point cloud data, and the processor is used for acquiring the 2D point cloud data and sequentially performing data cutting, mean filtering, Hough transformation and target shape feature extraction on the acquired 2D point cloud data. The processor is preferably arranged in the AGV body 1, a carrying plate 3 is arranged on the top of the AGV body 1, and wheels 4 are arranged at the bottom of the AGV body 1, all the wheels 4 are in a plane to facilitate ground travel, and the plane where all the wheels 4 are located is the bottom surface of the AGV body 1.

[0041] In the embodiment, the plurality of single-line radars 2 are arranged at the front end of the AGV body 1, and the scanning planes 21 of the plurality of single-line radars 2 are parallel to the bottom surface of the AGV body 1. The single-line radars 2 are preferably single-line laser radars and can be used for positioning and navigation. The scheme utilizes the existing single-line laser radars 2 on the jacking carrying AGV for positioning and navigation to detect and position the shelf, thereby saving the cost of additional hardware and avoiding the installation and maintenance of two-dimensional codes. Since the amount of 2D point cloud data collected by the single-line laser radars 2 is much smaller than the amount of 3D point cloud collected by a stereo vision sensor and the amount of environment pictures collected by a two-dimensional code camera, the data processing complexity will be significantly reduced, thereby reducing the cost of the data processor.

[0042] In the embodiment, referring to Figure 2 The detection system 10 based on point cloud shape features further comprises a shelf 5, the middle part of the shelf 5 is the travel path of the AGV body 1; the shelf 5 is provided with cover plates 51 on three sides and is provided with an opening 52 only at the front part to form a U-shaped structure, the AGV body 1 can enter the shelf 5 from the opening 52, and a material box 6 is correspondingly arranged above the U-shaped structure of the shelf 5. It can be understood that the U-shaped structure of the shelf 5 facilitates the AGV to detect and position the shelf 5 by using the single-line radars 2, and when the AGV body 1 travels to the inside of the U-shaped structure of the shelf 5, the material box 6 is correspondingly arranged, at this time, the bottom of the material box 6 is correspondingly arranged with the carrying plate 3, and the carried piece falls from the bottom of the material box 6 to the carrying plate 3 to realize material transfer.

[0043] Based on the same inventive concept, referring to Figure 3 The embodiment of the application provides a detection method based on point cloud shape features, which comprises the following steps:

[0044] S1: a point cloud data acquisition step, acquiring 2D point cloud data;

[0045] S2: a data clipping step of clipping data of the 2D point cloud data obtained in step S1;

[0046] S3: a mean filtering step of performing mean filtering on the data clipped in step S2;

[0047] S4: a Hough transform line segment detection step of obtaining a target line segment through Hough transform on the data subjected to the mean filtering in step S3; and

[0048] S5: a line segment sorting step of sorting the target line segment and extracting a target shape feature according to an angle relationship between adjacent target line segments.

[0049] Please refer to Figure 3 Based on the extracted target shape feature, the detection and positioning of the shelf 5 can be realized, and therefore the line segment sorting step is followed by:

[0050] S6: a three-segment method shelf detection and positioning step, the target shape feature being a U-shaped structure formed by three target line segments, and the target shape feature being output as a result of real-time detection and positioning of the shelf 5.

[0051] In the embodiment, the specific method of sorting the target line segments includes taking the front of the radar 2 as 0 degrees, taking the counterclockwise direction as the positive direction, and sorting the adjacent target line segments in the order from small to large according to the angle between the adjacent target line segments.

[0052] In the embodiment, the method of obtaining the target shape feature includes: according to all the line segments obtained, sequentially calculating whether two continuous target line segments form an L-shaped structure, whether three continuous target line segments form a U-shaped structure, and whether the length of the target line segment meets the requirements; the two target line segments forming an L-shaped structure is a criterion for denying the determination, which is that the angle range of the two straight lines is within 90±5 degrees, indicating that complete perpendicularity or similar perpendicularity is acceptable; the three continuous target line segments forming a U-shaped structure is a criterion for determining that, on the basis of the two target line segments forming an L-shaped structure, it is determined whether the three target line segments satisfy the condition that two parallel target line segments are connected to the two ends of another target line segment. In the embodiment, the two parallel target line segments refer to the angle range of the two straight lines being within 0±5 degrees, indicating that complete parallelism or similar parallelism is acceptable.

[0053] In the embodiment, the 2D point cloud data includes the angle and distance of the obstacle point scanned by the radar 2.

[0054] In the embodiment, the specific method of mean filtering is to directly delete the aggregated point cloud data whose difference from the mean value is greater than a first threshold value.

[0055] Please refer to Figure 3 , Figure 4Since the positioning and navigation function has moved the AGV to a position in front of the shelf 5, when the processor collects 2D point cloud data of the surrounding environment from the radar 2, it can first crop and remove point cloud data in certain directions, such as point cloud data towards the rear of the AGV. After cropping, the point cloud data will contain the shape features of the shelf 5. Since there are errors and noise points in the measurement process of the laser radar 2, the cropped point cloud data will be subjected to mean filtering to remove observation points that deviate significantly from the actual position. At this time, the left and right sides and the inner cover plate 51 of the U-shaped shelf 5 will appear in the form of line segments in the point cloud data, and the line segments in the point cloud can be detected by Hough transform, and then all the line segments are arranged in clockwise order according to the angle of the starting point. Finally, by detecting the perpendicular angle relationship between adjacent line segments, the continuous three line segments with U-shaped features are extracted as the detection result of the shelf 5. Therefore, the shelf 5 detection method using single-line laser can use the single-line laser radar 2 on the AGV for positioning and navigation and obstacle avoidance to detect and position the shelf 5, reducing the additional sensor and processor hardware cost; the design of the U-shaped shelf 5 improves the stability of the detection, and the detection accuracy of the single-line laser radar 2 on the shelf 5 is high, fast, and simple in structure, stable and reliable, and is not affected by changes in environmental light; the algorithm flow of the shelf 5 detection and positioning is simple and fast in operation speed.

[0056] In the present embodiment, before the step of obtaining 2D point cloud data, the following steps are further included:

[0057] An AGV body 1, a processor mounted on the AGV body 1, and a plurality of single-line radars 2 are provided, wherein the plurality of single-line radars 2 are used to scan the surrounding environment of the shelf 5 to obtain 2D point cloud data, and the processor is used to obtain the 2D point cloud data and sequentially perform data cropping, mean filtering, Hough transform, and target shape feature extraction on the obtained 2D point cloud data; and

[0058] The middle part of the shelf 5 is set as the travel path of the AGV body 1; the shelf 5 is provided with a cover plate 51 on three sides and only a front opening 52, forming a U-shaped structure, and the AGV body 1 can enter the shelf 5 from the opening 52 position, and a material box 6 is provided above the U-shaped shelf 5.

[0059] The method and system for detecting based on point cloud shape features provided by the embodiments of the present application utilize the single-line laser radar on the jacking carrier AGV, which is used for positioning and navigation, to detect and position the shelves, thereby saving the additional hardware cost and avoiding the installation and maintenance of the two-dimensional code. Since the amount of 2D point cloud data collected by the single-line laser radar is far less than the amount of 3D point cloud collected by the stereo vision sensor and the amount of environment pictures collected by the two-dimensional code camera, the data processing complexity will be significantly reduced, thereby reducing the cost of the data processor and avoiding the influence of the change of ambient light.

[0060] Therefore, the main beneficial effects of the present application are:

[0061] 1) The present application utilizes the single-line laser radar on the AGV for positioning, navigation and obstacle avoidance to detect and position the shelves, thereby reducing the additional sensor and processor hardware cost.

[0062] 2) The present application utilizes the single-line laser to detect the shelves, which is high in detection accuracy, fast in speed, simple in structure, stable and reliable, and not affected by the change of ambient light.

[0063] In the above embodiments, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0064] The above describes in detail the method and system for detecting based on point cloud shape features provided by the embodiments of the present application, and the principle and implementation mode of the present application are described by applying specific examples. The above description of the embodiments is only used to help understand the technical solutions and core ideas of the present application; the ordinary skilled in the art should understand that the technical solutions recorded in the above embodiments can be modified or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for detecting shape features based on point cloud, characterized in that, The method comprises the following steps: S1: obtaining 2D point cloud data; S2: performing data cutting on the 2D point cloud data obtained in step S1; S3: performing mean filtering on the data after cutting in step S2; S4: obtaining target line segments through Hough transformation on the data after mean filtering in step S3; and S5: sorting the target line segments and extracting target shape features according to the angle relationship between adjacent target line segments; Before the step of obtaining 2D point cloud data, an AGV body, a processor installed on the AGV body and a plurality of single-line laser radars are further provided, wherein the plurality of single-line laser radars are used to scan the environment around the shelf to obtain 2D point cloud data, and the processor is used to obtain the 2D point cloud data and sequentially perform data cutting, mean filtering, Hough transformation and target shape feature extraction on the obtained 2D point cloud data; According to all obtained line segments, whether two continuous target line segments constitute an L type, whether three continuous target line segments constitute a U type, and whether the length of the target line segment meets the requirements are sequentially calculated, and whether is_L>0 is judged, whether is_U>0 is judged, and if is_L>0 and is_U>0, the probability that the current object is a shelf is calculated through IS_SHELF=is_L||is_U, whether IS_SHELF>0 is judged, and if yes, it is determined that the object has been pressed into the shelf, the most reliable shelf is screened, and the warehousing is started.

2. The method of claim 1, wherein, The specific method of sorting the target line segments comprises: taking the front of the radar as 0 degrees, taking the counterclockwise as the positive direction, and sorting the adjacent target line segments in the order from small to large according to the angle between them.

3. The method of claim 1, wherein, The method for obtaining the target shape features comprises: the denial judgment standard for two target line segments constituting an L type is that the angle range of the two straight lines is 90±5 degrees, and the judgment standard for three continuous target line segments constituting a U type is that, on the basis of the two target line segments constituting an L type, it is judged whether two parallel target line segments in the three target line segments are connected at the two ends of the other target line segment.

4. The method of claim 3, wherein, The target shape features are a U type formed by three target line segments, and the target shape features are output as the result of real-time positioning detection of the shelf.

5. The method of claim 1, wherein, The 2D point cloud data is obtained by scanning the angle and distance of the obstacle point by the radar.

6. The method of claim 1, wherein, The specific method of mean filtering is to directly delete the aggregated point cloud data with a difference greater than a first threshold from the mean value.

7. The method of claim 1, wherein, Before the step of obtaining 2D point cloud data, the following steps are further included: The middle part of the shelf is set as the travel path of the AGV body, three sides of the shelf are provided with cover plates, only the front part is provided with an opening to form a U-shaped structure, the AGV body can enter the shelf from the opening position, and a material box is arranged above the U-shaped structure.

8. A point cloud shape feature based detection system for implementing the point cloud shape feature based detection method of any one of claims 1 to 7, characterized in that, The AGV body, a processor mounted on the AGV body, and a plurality of single-line laser radars for scanning the environment around the shelf to obtain 2D point cloud data, the processor is used for obtaining the 2D point cloud data and sequentially performing data cutting, mean filtering, Hough transformation and target shape feature extraction on the obtained 2D point cloud data.

9. The system for detecting shape features based on point clouds of claim 8, wherein, The plurality of single-line laser radars are arranged at the front end of the AGV body, and the scanning planes of the plurality of single-line laser radars are parallel to the bottom surface of the AGV body.

10. The system for detecting shape features based on point cloud of claim 8, wherein, The detection system based on the point cloud shape feature further comprises a shelf, the middle part of the shelf is the running path of the AGV body, the shelf is provided with cover plates on three sides, only the front part is provided with an opening, forming a U-shaped structure, the AGV body can enter the shelf from the opening position, and a material box is arranged above the U-shaped structure.

Images