Bottom template foreign matter detection method and system based on laser radar scanning

Abstract

The invention provides a bottom template foreign matter detection method and system based on laser radar scanning, and relates to the field of radar detection, and the method comprises the steps: scanning the surface of a bottom template through a laser radar, collecting point cloud data, and storing the data; the method comprises the following steps of: acquiring data, performing denoising and down-sampling on the acquired data, analyzing geometrical characteristics, selecting a registration reference area, calculating FPFH characteristics, performing coarse registration by using an RANSAC algorithm, performing fine registration by using an ICP algorithm, clustering by using a DBSCAN algorithm, visualizing point cloud data by using a Plot library, and finally identifying foreign matters by combining priori knowledge so as to clean a foreign matter area by using washing equipment. And the bottom template is ensured to be clean through multiple times of scanning and processing. According to the technical scheme, the bottom formwork can be rapidly scanned, the plane point cloud model graph can be formed, the garbage distribution position and the residue number of the bottom formwork are determined, and therefore cleaning work is guided, redundant and repeated cleaning workloads are reduced, it is ensured that the cleaning effect is practical and effective, and it is ensured that the apparent quality of a concrete bottom plate reaches the standard.

Description

Technical Field

[0001] The present invention relates to the field of radar detection technology, and in particular to a method and system for detecting foreign matter in a bottom template based on laser radar scanning. Background Art

[0002] In recent years, with the vigorous development of transportation civil engineering, in order to achieve high-efficiency and high-quality concrete construction, componentized steel construction technology has been widely used in large prefabricated structures. After the steel skeleton and prestressed pipe system are prefabricated on the steel frame and the internal formwork of the structure is installed, the componentized steel construction technology of overall hoisting can directly hoist the complete steel skeleton into the formwork, reducing the installation work of on-site steel bars, improving the construction quality, improving the construction accuracy, and ensuring the construction efficiency. Due to its overall hoisting characteristics, the overall weight of the steel skeleton is huge and inconvenient to move. During the installation of componentized steel bars and the intersection of construction measures such as the template, some sporadic welding slag and other foreign matter are generated in the bottom template and leaked onto the bottom template. Due to the obstruction of the componentized steel cage, the foreign matter on the bottom template cannot be directly observed and detected, and it is difficult to observe the distribution state of foreign matter on the bottom template and the cleaning effect. Poor cleaning effect of foreign matter will seriously affect the apparent quality effect of the concrete bottom plate, reduce the thickness of the concrete protective layer, and have a certain impact on the crack resistance and durability of the structure. Therefore, it is particularly important to quickly and effectively determine the foreign matter on the bottom template. Traditional cleaning methods have the following adverse effects: First, it is impossible to effectively determine the distribution location and quantity of foreign matter on the bottom formwork, the cleaning work is not targeted, and the cleaning efficiency is low; second, it is impossible to effectively determine the effect of cleaning foreign matter on the bottom formwork, and the cleaning effect is often poor and cannot be effectively obtained, and the final cleaning effect is uncontrollable; Based on the above reasons, how to quickly detect foreign matter on the surface of the bottom formwork during the construction of modular steel bars has become a technical problem that needs to be solved. Summary of the invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art or related technology, and discloses a bottom template foreign body detection method and system based on laser radar scanning, which can quickly detect the construction site for visual foreign body identification, realize the rapid positioning of foreign bodies, and realize the visualization of foreign bodies.

[0004] The first aspect of the present invention discloses a method for detecting foreign matter in a bottom template based on laser radar scanning, comprising: radar scanning: deploying laser radars at multiple measuring points on the side wall of the bottom template, scanning the surface of the bottom template using the laser radar, and obtaining multiple groups of point cloud data corresponding to the multiple measuring points; processing point cloud data: selecting feature areas that appear in each group of point cloud data as registration reference features; calculating point FPFH features of multiple groups of point cloud data; coarse registration: according to the FPFH features, using the RANSAC algorithm to perform coarse registration on each group of point cloud data, and calculating the transformation matrix of each group of point cloud data; transforming each group of point cloud data according to the transformation matrix, aligning multiple groups of point cloud data to the same coordinate system, and generating coarse registration point clouds; Fine registration: Based on the coarse registration point cloud, the ICP algorithm is used for fine registration to generate a fine registration point cloud; Point cloud clustering: Based on the fine registration point cloud, the DBSCAN algorithm is used to cluster the point cloud to generate a clustered point cloud; Visualization: The clustered point cloud is visualized to generate a visualized point cloud image; Identify foreign objects: Based on the visualized point cloud image, combined with the prior knowledge of the shape, size and position of normal objects on the base template, the point cloud of componentized steel bars and pads is eliminated, and the location information of foreign objects is identified and recorded.

[0005] According to the bottom template foreign body detection method based on laser radar scanning disclosed by the present invention, preferably, it also includes: cleaning process control: according to the foreign body identification result, issuing an instruction to clean the bottom template, and performing radar scanning again after the cleaning is completed, so as to carry out a new round of foreign body detection and cleaning, until no foreign body is detected on the bottom template.

[0006] According to the base template foreign body detection method based on laser radar scanning disclosed in the present invention, preferably, the step of processing point cloud data also includes: denoising the collected original point cloud data to reduce the impact of environmental noise and measurement errors on subsequent alignment and analysis; down-sampling the denoised point cloud data to reduce the amount of data while retaining the point cloud features.

[0007] According to the base template foreign body detection method based on laser radar scanning disclosed in the present invention, preferably, the method for selecting the registration reference features includes: analyzing the geometric features of the point cloud data, and selecting the overlapping areas with rich and significant features in multiple groups of point cloud data as the registration reference features.

[0008] According to the base template foreign body detection method based on laser radar scanning disclosed by the present invention, preferably, the process of calculating the transformation matrix specifically includes: using the RANSAC algorithm to find a matching relationship between the FPFH features of multiple groups of point cloud data, by continuously randomly selecting multiple point pairs, and then calculating the consistency of the point pairs, counting the number of point pairs that meet the consistency conditions, and after multiple iterations, finding the transformation matrix with the most point pairs that meet the consistency conditions.

[0009] According to the base template foreign body detection method based on laser radar scanning disclosed by the present invention, preferably, the step of precise registration specifically includes: the ICP algorithm searches for the optimal rigid transformation between the two groups of point clouds through iterative optimization so that the distance between corresponding points is minimized; in each iteration, firstly, the nearest point pair between the two groups of point clouds is found according to the current point cloud posture estimation, and then the rigid transformation that minimizes the sum of the distances of the corresponding point pairs is calculated, and the transformation is applied to the point cloud to be registered, and this process is repeated until the convergence condition is met.

[0010] According to the base template foreign body detection method based on laser radar scanning disclosed by the present invention, preferably, the step of point cloud clustering specifically includes: the DBSCAN algorithm divides the point cloud into different clusters based on the density distribution of points in the point cloud, and each cluster represents an object or a point cloud area with similar characteristics.

[0011] According to the base template foreign body detection method based on laser radar scanning disclosed by the present invention, preferably, the visualization step specifically includes: using the Plotly visualization library to visualize the clustered point cloud data, and assigning different colors to different point clusters, so as to intuitively distinguish various objects or areas.

[0012] According to the bottom template foreign body detection method based on laser radar scanning disclosed by the present invention, preferably, the angular resolution of the laser radar meets the following conditions: ; Where, θ represents the angular resolution of the lidar, θ max represents the maximum resolution, r represents the detection distance, and d represents the minimum size of the foreign object.

[0013] The second aspect of the present invention discloses a bottom template foreign body detection system based on laser radar scanning, comprising: a memory for storing program instructions; a processor for calling the program instructions stored in the memory to implement a bottom template foreign body detection method based on laser radar scanning as any of the above technical solutions.

[0014] The beneficial effects of the present invention include at least: 1. The present invention can accurately identify foreign matter on the bottom mold through advanced point cloud data processing technology. Compared with the traditional method of relying on manual visual inspection or simple image processing, it greatly improves the accuracy and reliability of foreign matter detection, effectively avoids misjudgment and missed judgment caused by human factors or technical limitations, thereby ensuring that the cleanliness of the bottom mold reaches a higher standard and guarantees the quality and stability of subsequent production or processing.

[0015] Second, the present invention has strong adaptability. Whether in an outdoor environment with strong light or in a dimly lit indoor scene, the laser radar can work stably. In strong light, its laser-based detection principle is not affected by light intensity, and will not have problems such as overexposure or shadow interference like a camera, and can always accurately detect obstacles. In a dark environment, there is no need for additional lighting equipment to assist, and obstacles can be clearly identified by relying on the laser beam emitted by itself, ensuring reliable operation under various lighting conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic flow chart of a method for detecting foreign matter in a base template based on laser radar scanning according to an embodiment of the present invention is shown.

[0017] Figure 2 A schematic diagram of the installation position of a laser radar according to an embodiment of the present invention is shown.

[0018] Figure 3 A schematic diagram of a visualized point cloud according to an embodiment of the present invention is shown.

[0019] Figure 4 A schematic block diagram of a base template foreign body detection system based on laser radar scanning according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0020] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention can also be implemented in other ways different from those described herein, and therefore, the present invention is not limited to the limitations of the specific embodiments disclosed below.

[0021] like Figure 1 As shown, according to one embodiment of the present invention, a method for detecting foreign matter on a bottom template based on laser radar scanning is disclosed, comprising: Step 1, data collection: select a suitable laser radar according to the detection distance and required accuracy, and use the laser radar to scan the base template to collect point cloud data of the object on the base template; Step 2, data processing: process the collected point cloud data, select the overlapping areas with significant features in multiple groups of point cloud data, calculate the point FPFH features of multiple groups of point cloud data, use the RANSAC algorithm for rough registration, and calculate the transformation matrix of each group; after the point cloud data is transformed according to the transformation matrix obtained by rough registration, the ICP algorithm is used for fine registration, and then the DBSCAN algorithm is used for point cloud clustering; Step 3, cleaning control: visualize the clustered point cloud data, identify foreign matter based on the visualized point cloud data, and then rinse the area with debris on the bottom mold. Repeat the above steps until the debris on the bottom mold is cleared.

[0022] According to the above embodiment, further, step 1 specifically includes: selecting a laser radar that meets the requirements according to the required detection distance r, the minimum size d of the debris, and the required accuracy, mainly including the measurement radius, ranging accuracy, and angular resolution of the laser radar, wherein the expression of the angular resolution that meets the requirements is: ; like Figure 2 As shown in the figure, the laser radar is connected to the development board equipped with ROS system and Linux system and installed on the radar installation device. Several measuring points are selected on the base template, and the laser radar is installed on the measuring points. The objects on the base template are scanned, and the radar point cloud data is saved locally by reading the laser radar topic message in ROS. The point cloud data is saved in csv format, with 2 columns for each group of data. The first column is the angle of laser rotation, and the second column is the distance from the point to the radar.

[0023] According to the above embodiment, further, step 2 specifically includes: Step 2.1: De-noise the collected original point cloud data to reduce the impact of environmental noise and measurement errors on subsequent registration and analysis. Downsample the denoised point cloud data to reduce the amount of data and improve computational efficiency while retaining the key features of the point cloud. Step 2.2: Analyze the geometric features of the point cloud data, and select the areas with rich and significant features and large overlap in multiple groups of point cloud data as references for point cloud registration; Step 2.3: Use the RANSAC algorithm to find matching relationships between the FPFH features of multiple groups of point cloud data. By continuously randomly selecting a small number of point pairs, and then calculating the consistency of these point pairs, the number of point pairs that meet the consistency conditions is counted. After multiple iterations, the transformation matrix with the most point pairs that meet the consistency conditions is found.

[0024] In this step, the FPFH feature is a feature descriptor used for 3D point cloud processing, which is defined as the weighted average of the simplified point feature histogram (SPFH) between each point and its neighboring points. The RANSAC algorithm is a random sampling consensus algorithm. Its principle is to first randomly extract a minimum sample set from the data set to fit the transformation matrix, and then calculate the distance between the remaining points and the corresponding points after applying the transformation matrix. Points with a distance less than the set threshold are judged as inliers, and the transformation matrix with more inliers is better. After multiple sampling and evaluation, the transformation matrix with the most inliers is selected as the final transformation matrix.

[0025] Step 2.4: Apply the optimal transformation matrix of each group of point clouds to each group of point cloud data, so that all point clouds are initially aligned to the coordinate system of the first group of point cloud data, completing the rough registration process. Although there may still be some deviations in the point cloud data after rough registration, they have been aligned in the approximate position, providing a good initial condition for the subsequent fine registration; Step 2.5: Use ICP algorithm for precise registration: The ICP algorithm continuously iterates and optimizes to find the optimal rigid transformation between the two sets of point clouds P and Q so that the corresponding points {p i} and {q i The distance between them is the smallest.

[0026] Since it is impossible to accurately determine the matching corresponding points between two point clouds, in each iteration, firstly, according to the current point cloud pose estimation, the nearest point pair between the two groups of point clouds is found as the corresponding point, and then the rigid transformation that minimizes the sum of the distances of these corresponding point pairs is calculated and applied to the point cloud to be registered. At this time, the error e is expressed as: ; Repeat this process until the error e meets the convergence condition, and the transformation matrices r and t can be obtained.

[0027] Step 2.6: Apply the DBSCAN algorithm to cluster the point cloud data after precise registration: The DBSCAN algorithm divides the point cloud into different clusters based on the density distribution of points in the point cloud. Each cluster represents a possible object or a point cloud area with similar features.

[0028] According to the above embodiment, further, step 3 specifically includes: Step 3.1: If Figure 3 As shown in the figure, Plotly is used to visualize the clustered point cloud data, and different colors are assigned to different point clusters to intuitively distinguish various objects or areas.

[0029] Step 3.2: By visualizing the clustered point cloud data and combining the prior knowledge of the shape, size and position of normal objects on the base mold, the point cloud of componentized steel bars and pads is eliminated, and the location information of foreign objects is identified and recorded; Step 3.3: Use appropriate flushing equipment (such as high-pressure water gun, spray device, etc.) to flush the area on the bottom mold with debris. After flushing, wait for the bottom mold surface to dry.

[0030] Step 3.4: Collect the point cloud data of the base mold again, process, align, cluster, visualize and identify foreign objects, and check whether there are any remaining debris. If there are still debris, continue to rinse and repeat the inspection until the point cloud data inspection determines that the debris on the base mold has been completely removed.

[0031] According to another embodiment of the present invention, the implementation of the bottom template foreign body detection method based on laser radar scanning provided in the above embodiment in an actual application scenario is also disclosed: the laser radar measurement radius is 18m, the scanning frequency is 10Hz, and the angular resolution is 0.1125°; Figure 2 As shown in the figure, multiple measuring points are selected along the edge line of the bottom formwork of a short-line prefabricated segment beam, with 4 measuring points as a measuring area. Triangular objects are placed on both sides of each measuring area as a reference for registration; the objects on the bottom formwork are scanned at each measuring point in turn, and the radar point cloud data is saved locally by reading the lidar topic message in ROS; the point cloud data is read, and the point clouds of the triangular objects on both sides of the measuring area are selected as the reference features for registration, and the FPFH features of the reference features for registration are calculated, and then the RANSAC algorithm is used for rough registration; the second group of point clouds are transformed according to the results of the rough registration, and on this basis, the ICP algorithm is used for further registration, and the second group of point clouds are transformed to the coordinate system of the first group of point clouds according to the registration results, and the two groups of point clouds are fused into one group of point clouds. The fused point cloud is clustered using the DBSCAN algorithm, and the clustered point cloud is visualized using Plotly, as shown in the figure below. Figure 3 As shown. Figure 3 Combined with the on-site environment, it can be seen that the points with regular arrangement are concrete pads, and the points with disorderly distribution can be judged as debris. According to the visualization results, use a water gun to clean the debris on the bottom mold. Repeat the above steps until the debris is cleaned.

[0032] According to another embodiment of the present invention, a base template foreign object detection system based on laser radar scanning is also disclosed, including: a memory for storing program instructions; a processor for calling the program instructions stored in the memory to implement the base template foreign object detection method based on laser radar scanning as in the above embodiment.

[0033] like Figure 4 As shown, according to the above embodiment, further, the memory is also used to store point cloud data, and the processor is also used to run the ROS (Robot Operating System) system; a lidar node and a data acquisition node are established based on the ROS node manager, and the lidar is controlled by the lidar node after being connected to the ROS system, and the radar signal is fed back to the lidar node; the lidar node also sends the radar signal to the data acquisition node by publishing a point cloud data topic, and the point cloud data is stored in the memory.

[0034] In summary, the present invention first selects a laser radar according to the detection distance, the minimum size of debris and the required accuracy, connects it to a development board equipped with a ROS system and a Linux system and installs it at the measuring point, scans the bottom mold to collect point cloud data and saves it. Subsequently, the collected data is denoised and downsampled, the geometric features are analyzed to select the registration reference area, the FPFH features are calculated, and the RANSAC algorithm is used for rough registration, the ICP algorithm is used for fine registration, and then clustered by the DBSCAN algorithm and visualized using the Plotly library. Finally, foreign matter is identified in combination with prior knowledge, and the debris area is cleaned with a flushing device. The operation is repeated until the debris is removed. The bottom mold is ensured to be clean through multiple scans and processing to ensure the surface quality of the concrete bottom plate.

[0035] The bottom template foreign body detection method based on laser radar scanning provided in the above embodiment can quickly scan the bottom template and form a plane point cloud model map, determine the distribution position and residual amount of bottom template garbage, thereby guiding the cleaning work, reducing the redundant and repeated cleaning workload, ensuring that the cleaning effect is effective, and ensuring that the surface quality of the concrete bottom plate meets the standard. The technical solution provided by the present invention has the significant advantages of both speed and precision, giving full play to the rapidity of laser scanning, being able to quickly complete multiple repeated scans, and timely generate an accurate foreign body distribution position map, effectively avoiding meaningless and redundant cleaning operations.

[0036] All or part of the steps in the various methods of the above embodiments can be completed by controlling the relevant hardware through a program, and the program can be stored in a readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other readable medium that can be used to carry or store data.

[0037] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for detecting foreign matter on a bottom template based on laser radar scanning, characterized in that: include: LiDAR scanning: LiDAR is deployed at multiple measuring points on the side wall of the bottom template, and the surface of the bottom template is scanned by LiDAR to obtain multiple sets of point cloud data corresponding to multiple measuring points; Processing point cloud data: Selecting feature areas that appear in each set of point cloud data as registration reference features; Calculate the point FPFH features of multiple sets of point cloud data; Rough registration: Based on the FPFH features, the RANSAC algorithm is used to perform rough registration on each group of point cloud data, and the transformation matrix of each group of point cloud data is calculated; each group of point cloud data is transformed according to the transformation matrix, and multiple groups of point cloud data are aligned to the same coordinate system to generate a rough registration point cloud; Fine registration: Based on the coarse registration point cloud, the ICP algorithm is used to perform fine registration to generate a fine registration point cloud; Point cloud clustering: Based on the precisely registered point cloud, a DBSCAN algorithm is used to perform point cloud clustering to generate a clustered point cloud; Visualization: performing visualization processing on the clustered point cloud to generate a visualized point cloud image; Identify foreign objects: Based on the visualized point cloud image and combined with prior knowledge of the shape, size and position of normal objects on the bottom template, the point cloud of componentized steel bars and pads is eliminated, and the position information of foreign objects is identified and recorded.

2. The method for detecting foreign matter in a bottom template based on laser radar scanning according to claim 1, characterized in that: Also includes: Cleaning process control: Based on the foreign body identification results, an instruction to clean the bottom template is issued, and after the cleaning is completed, the radar scan is performed again to carry out a new round of foreign body detection and cleaning until no foreign body is detected on the bottom template.

3. The method for detecting foreign matter in a bottom template based on laser radar scanning according to claim 1, characterized in that: The step of processing point cloud data further includes: De-noising the collected raw point cloud data to reduce the impact of environmental noise and measurement errors on subsequent registration and analysis; The denoised point cloud data is downsampled to reduce the amount of data while retaining the point cloud features.

4. The method for detecting foreign matter in a bottom template based on laser radar scanning according to claim 1, characterized in that: The selection methods of registration reference features include: The geometric features of the point cloud data are analyzed, and the overlapping areas with rich and significant features in multiple groups of point cloud data are selected as the reference features for registration.

5. The method for detecting foreign matter in a bottom template based on laser radar scanning according to claim 1, characterized in that: The process of calculating the transformation matrix specifically includes: The RANSAC algorithm is used to find matching relationships between the FPFH features of multiple groups of point cloud data. By continuously randomly selecting multiple point pairs, calculating the consistency of the point pairs, and counting the number of point pairs that meet the consistency conditions, after multiple iterations, the transformation matrix with the most point pairs that meet the consistency conditions is found.

6. The method for detecting foreign matter in a bottom template based on laser radar scanning according to claim 1, characterized in that: The step of precise registration specifically includes: The ICP algorithm searches for the optimal rigid transformation between two groups of point clouds through iterative optimization, so that the distance between corresponding points is minimized. In each iteration, the nearest point pair between the two groups of point clouds is first found based on the current point cloud pose estimation, and then the rigid transformation that minimizes the sum of the distances of the corresponding point pairs is calculated, and the transformation is applied to the point cloud to be registered, and this process is repeated until the convergence condition is met.

7. The method for detecting foreign matter in a bottom template based on laser radar scanning according to claim 1, characterized in that: The step of point cloud clustering specifically includes: The DBSCAN algorithm divides the point cloud into different clusters based on the density distribution of points in the point cloud, and each cluster represents an object or a point cloud region with similar features.

8. The method for detecting foreign matter in a bottom template based on laser radar scanning according to claim 1, characterized in that: The visualization step specifically includes: The Plotly visualization library is used to visualize the clustered point cloud data, and different colors are assigned to different point clusters to intuitively distinguish various objects or areas.

9. The method for detecting foreign matter in a bottom template based on laser radar scanning according to any one of claims 1 to 8, characterized in that: The angular resolution of the laser radar meets the following conditions: ； Where, θ represents the angular resolution of the lidar, θ max represents the maximum resolution, r represents the detection distance, and d represents the minimum size of the foreign object.

10. A bottom template foreign body detection system based on laser radar scanning, characterized in that: include: A memory for storing program instructions; A processor is used to call the program instructions stored in the memory to implement the bottom template foreign body detection method based on laser radar scanning as described in any one of claims 1 to 9.

Images