Method for selecting optimal grabbing point for blast furnace stemming robot

Abstract

The invention discloses a method for selecting an optimal grabbing point for a blast furnace stemming robot, and the method comprises the following steps: 1) obtaining three-dimensional point cloud data of a stemming stacking area which comprises stemming, a stemming stacking container and a stemming area robot mechanical arm coverage area; (2) stemming point cloud in the three-dimensional point cloud data is extracted, the highest point is calculated, and the point cloud is sliced from the highest point; 3) performing point cloud segmentation on the slice data; 4) selecting an optimal stemming point cloud according to the segmented point cloud; and 5) calculating the center of the selected optimal stemming point cloud as a robot grabbing center point. The invention provides a method for selecting optimal stemming grabbing by a blast furnace clay adding gun robot, which can effectively solve the problem that the position of irregular stemming cannot be determined in the automatic grabbing process of a stemming robot, and improves the grabbing efficiency.

Description

Technical Field

[0001] The invention relates to metallurgical automation technology, in particular to a method for a blast furnace taphole robot to select an optimal grasping point. Background Art

[0002] Most steel plants use a crane to lift the mud gun to the vicinity of the gun mud machine, and then manually carry the mud into the gun mud machine piece by piece. Because there is a lot of dust and metal powder near the gun mud machine, the temperature is around 50 degrees all year round, and the density and weight of a single gun mud are high. The gun mud operation position is one of the worst and most difficult positions in blast furnace operation conditions. Therefore, the demand for unmanned gun mud operation positions is very urgent.

[0003] At present, a small number of sites have begun to use robots to assist in manual filling of cannon mud. Since cannon mud is an irregular and plastic block solid, it is often difficult for robots to select suitable grabbing points from the cannon mud pile, resulting in the inability to fully realize the unmanned process. In order to solve the problem of cannon mud grabbing points, some companies choose to add a set of cannon mud processing equipment in the blast furnace iron-making plant to shape the cannon mud into regular blocks of fixed size and store them in a buffer device, which is then grabbed by the robot in a fixed posture. This method greatly increases the cost, and due to the large space required by the cannon mud processing device, it is often impossible to install on site. How to improve the efficiency of robots grabbing cannon mud has become a problem that needs to be solved urgently by technicians in this field. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method for a blast furnace taphole robot to select an optimal grasping point in view of the defects in the prior art.

[0005] The technical solution adopted by the present invention to solve the technical problem is: a method for a blast furnace taphole mud robot to select an optimal grasping point, comprising the following steps:

[0006] 1) Acquire three-dimensional point cloud data of a cannon mud stacking area, wherein the cannon mud stacking area includes cannon mud, a cannon mud stacking container, and an area covered by a robot arm in the cannon mud area;

[0007] Among them, the area covered by the robot arm in the cannon mud area is: the space where the robot arm moves when the robot grabs the cannon mud directly above the cannon mud;

[0008] 2) Extract the blasthole point cloud in the 3D point cloud data, calculate the highest point, and slice the point cloud starting from the highest point;

[0009] 3) Perform point cloud segmentation on slice data;

[0010] 4) Select the optimal point cloud of cannon mud according to the segmented point cloud;

[0011] Perform plane fitting on the segmented point cloud and select the optimal gun mud point cloud on the segmented point cloud fitting plane;

[0012] 5) Calculate the center of the optimal selected mud point cloud as the robot grasping center point.

[0013] According to the above scheme, the step 2) also includes 2) extracting all point cloud data within the area covered by the robot arm of the cannon mud area from the three-dimensional point cloud data to perform obstacle recognition;

[0014] The extracted point cloud is segmented according to a fixed neighborhood. After removing the point cloud with a number less than the set threshold, the remaining point cloud is connected and recognized to confirm that there is no collision in the robot's activity area, otherwise an alarm signal is output.

[0015] According to the above scheme, in step 2), the point cloud is sliced ​​according to the height of the fixture.

[0016] According to the above scheme, in the step 3), a clustering method is used to classify the point cloud, including using Euclidean clustering and kmeans clustering algorithms to perform preliminary segmentation of the point cloud. After obtaining the basic block-based artillery mud point cloud, a watershed algorithm is used to perform edge recognition on the preliminary block-based artillery mud point cloud to segment the precise block-based artillery mud point cloud.

[0017] According to the above scheme, in step 2), the cannon mud point cloud in the three-dimensional point cloud data is extracted and the highest point is calculated as follows:

[0018] Extract all the point cloud data of the slice, including identifying the cannon mud container, removing the container point cloud, filtering outliers and interference points on the remaining cannon mud point cloud, extracting the remaining point cloud as the cannon mud point cloud, and searching for the highest point coordinates.

[0019] According to the above scheme, in step 4), the segmented point cloud is plane-fitted, and the optimal gun mud point cloud is selected on the segmented point cloud fitting plane; the details are as follows:

[0020] 4.1) Perform plane fitting on each separated cannon clay point cloud, select point clouds with plane angles less than a threshold, and sort the area of ​​the plane fitting of each cannon clay point cloud from large to small;

[0021] 4.2) The first K blocks of cannon mud point clouds in the area sorting and the point clouds without specific interference in the point cloud neighborhood are taken as candidates for grabbing cannon mud point clouds. The evaluation objective function is set for the candidate point clouds according to the area, plane inclination angle and height to determine the optimal grabbing cannon mud;

[0022] The point cloud neighborhood has no specific interference, which means that the selected point cloud has no point higher than the point cloud within the range centered on the robotic arm fixture disk;

[0023] If there is no candidate for grabbing the cannon mud point cloud, then for the segmented point cloud, all the points in the neighborhood are traversed to obtain all the local highest points in the cannon mud point cloud, and the plane is segmented and fitted within the neighborhood of the local highest point. Steps 4.1) and 4.2) are repeated to obtain the optimal cannon mud point cloud.

[0024] According to the above scheme, in step 4.1), a sampling consistency algorithm is used for plane fitting.

[0025] The present invention also provides an electronic device, comprising: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the method described in any one of the above schemes.

[0026] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the method described in any one of the above-mentioned schemes when executed by a processor.

[0027] The beneficial effects produced by the present invention are:

[0028] The present invention provides a method for a blast furnace mud adding gun robot to select the optimal method for grabbing gun mud, which can effectively solve the problem that the gun mud robot cannot determine the position of irregular gun mud during the automatic grabbing process, improve the grabbing efficiency, and at the same time avoid the problem that the robot arm cannot avoid gun mud and other obstacles during the grabbing process, resulting in collision and damage to equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0030] Figure 1 It is a method flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0032] like Figure 1 As shown, a method for a blast furnace taphole robot to select an optimal grasping point comprises the following steps:

[0033] 1) Scanning the cannon mud stacking area with a sensing device to obtain three-dimensional point cloud data of the cannon mud stacking area, wherein the cannon mud stacking area includes cannon mud, a cannon mud stacking container, and an area covered by a robot arm in the cannon mud area;

[0034] Perception devices, including laser scanners, structured light cameras, and other devices that can acquire 3D point clouds in a non-contact manner;

[0035] The stacking container of the taphole mud is the taphole mud stacking area, which refers to an open or closed area that can be grasped by the robot;

[0036] The area covered by the robot arm in the taphole mud area is: the space where the robot arm moves when the robot grabs the taphole mud directly above the taphole mud;

[0037] 2) Extract the blasthole point cloud in the 3D point cloud data, calculate the highest point, and slice the point cloud starting from the highest point;

[0038] Step 2) also includes 2) extracting all point cloud data within the area covered by the robot arm of the cannon mud area from the three-dimensional point cloud data to perform obstacle recognition;

[0039] The extracted point cloud is segmented according to a fixed neighborhood. After removing the point cloud with a number less than the set threshold, the remaining point cloud is connected and recognized to confirm that there is no collision in the robot's activity area, otherwise an alarm signal is output.

[0040] Obstacle identification refers to the analysis of point clouds within a certain height range directly above the cannon mud stacking area, where the height range is set to 0.8 meters to 2 meters according to the actual situation on site. Connected domain analysis is performed on all point clouds in this interval. When the area or length of the point cloud data of a connected interval is greater than the set threshold, the connected domain calculation is performed on the connected area and the cannon mud point cloud. If it can be connected to the cannon mud point cloud as a whole, and according to the principle of object stability, it is determined to be stacked stably and non-independent, it is judged as a non-obstacle. If it is identified as a rod-shaped or spherical object, or an independent and unconnected object, it is judged as an obstacle, and an alarm signal is output to the control system to end the whole process.

[0041] Extract the cannon mud point cloud, calculate the highest point, and start from the highest point to slice the point cloud according to the fixture height, extract all the point cloud data of the slice, including identifying the cannon mud container, removing the container point cloud, filtering outliers and interference points on the remaining cannon mud point cloud, extracting the remaining point cloud as the cannon mud point cloud, and searching for the coordinates of the highest point.

[0042] Cut the point cloud of fixed height from the highest point downward as the original point for the first optimal point calculation;

[0043] Among them, the slicing is performed according to the height of the fixture, that is, according to the length of the fixture and the fact that the fixture does not collide with the mechanical equipment around the fixture during the descent process. Here, the slicing height is set to 17CM;

[0044] Usually, the cannon mud is piled inside a cylindrical, rectangular or cube container. By limiting the coordinate range of the point cloud in three dimensions, the point cloud within the range is identified and extracted for a specific shape, and the container point cloud can be eliminated.

[0045] 3) Perform point cloud segmentation on slice data;

[0046] First, a clustering method is used to classify the point cloud, and then, in combination with an edge recognition method, part of the point cloud of each piece of cannon mud is separated; the clustering method used to classify the point cloud includes using Euclidean clustering and kmeans clustering algorithms to perform preliminary segmentation of the point cloud, and after obtaining basic block cannon mud point clouds, a watershed algorithm is used to perform edge recognition on the preliminary block cannon mud point clouds to segment accurate block cannon mud point clouds.

[0047] 4) Select the optimal point cloud of cannon mud according to the segmented point cloud;

[0048] The sampling consistency algorithm is used to perform plane fitting on each piece of the segmented blasthole point cloud, and the optimal blasthole point cloud is selected on the segmented point cloud fitting plane;

[0049] 4.1) Use sampling consistency algorithm to perform plane fitting on each separated cannon clay point cloud, select point clouds with plane angles less than the threshold, and sort the area of ​​the plane fitting of each cannon clay point cloud from large to small;

[0050] In this embodiment, a sampling consistency algorithm is used for fitting, and other algorithms such as least squares method, eigenvalue method, principal component analysis, etc. can also be used for fitting.

[0051] The plane angle threshold needs to be determined based on the fixture. In this embodiment, there is a rotating disc above the robot fixture. When the tilt angle is too large, the disc is likely to collide with the gun mud, causing the robot to stop moving. According to the size of the fixture, the plane angle threshold can be set to 40 degrees.

[0052] Sort the areas of the point cloud fitting planes, project the fitting planes onto the ground, extract the edge contours, and calculate the area of ​​the projected irregular shapes by gridding or integration.

[0053] 4.2) The first K blocks of cannon mud point clouds sorted by area and without specific interference in the neighborhood of the point cloud are taken as candidates for grabbing cannon mud point clouds. The evaluation objective function is set for the candidate point clouds according to the area, plane inclination angle and highest point height to determine the optimal grabbing cannon mud;

[0054] There is no specific interference in the neighborhood of the point cloud, which means that within the range of the selected point cloud centered on the robotic arm fixture disk, there is no point higher than the point cloud;

[0055] In this embodiment, the weight of the height in the evaluation objective function is set to be greater than the weights of the area and the plane inclination angle;

[0056] After evaluating the objective function calculation, the candidate point cloud with the largest score after sorting is extracted from the fixed neighborhood, and the obtained point cloud is subjected to outlier removal and clustering. If the number of clustered point clouds is greater than the set threshold, it is marked as an undesirable point, otherwise it is marked as a desirable point and confirmed as the optimal grabbing mud.

[0057] If there are no candidates for grabbing the cannon mud point cloud, you can stop grabbing the cannon mud, or for the segmented point cloud, traverse all points in the neighborhood to obtain all local highest points in the cannon mud point cloud, segment and fit the plane within the neighborhood of the local highest point, repeat steps 4.1) and 4.2), and obtain the optimal cannon mud point cloud.

[0058] 5) Calculate the center of the optimal selected mud point cloud as the robot grasping center point.

[0059] The center of the plane or centroid is calculated as follows:

[0060] Assume that there are n points in the three-dimensional point cloud of the gun mud point cloud, each point can be represented by the coordinates (x i ,y i , z i ) means that the center (center of mass) G(X G , Y G , Z G )for

[0061]

[0062] The calculated result is the optimal cannon mud grabbing point. Through calibration and calculation with the robot arm coordinate system, it can be converted into the robot's grabbing coordinates to control the robot to grab the cannon mud.

[0063] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims

1. A method for selecting the optimal gripping point of a blast furnace taphole robot, characterized in that: The following steps are involved: 1) Acquire three-dimensional point cloud data of a cannon mud stacking area, wherein the cannon mud stacking area includes cannon mud, a cannon mud stacking container, and an area covered by a robot arm in the cannon mud area; Among them, the area covered by the robot arm in the cannon mud area is: the space where the robot arm moves when the robot grabs the cannon mud directly above the cannon mud; 2) Extract the blasthole point cloud in the 3D point cloud data, calculate the highest point, and slice the point cloud starting from the highest point; 3) Perform point cloud segmentation on slice data; 4) Select the optimal taphole mud point cloud based on the segmented point cloud; Perform plane fitting on the segmented point cloud and select the optimal gun mud point cloud on the segmented point cloud fitting plane; 5) Calculate the center of the optimal selected mud point cloud as the robot grasping center point.

2. The method for selecting the optimal grasping point of the blast furnace taphole mud robot according to claim 1, characterized in that: The step 2) further includes 2) extracting all point cloud data within the area covered by the robot arm of the cannon mud area from the three-dimensional point cloud data to perform obstacle recognition; The extracted point cloud is segmented according to a fixed neighborhood. After removing the point cloud with a number less than the set threshold, the remaining point cloud is connected and recognized to confirm that there is no collision in the robot's activity area, otherwise an alarm signal is output.

3. The method for selecting the optimal grasping point of the blast furnace taphole mud robot according to claim 1, characterized in that: In the step 2), the point cloud is sliced ​​according to the fixture height.

4. The method for selecting the optimal grasping point of a blast furnace taphole mud robot according to claim 1, characterized in that: In the step 2), the cannon mud point cloud in the three-dimensional point cloud data is extracted and the highest point is calculated as follows: Extract all the point cloud data of the slice, including identifying the cannon mud container, removing the container point cloud, filtering outliers and interference points on the remaining cannon mud point cloud, extracting the remaining point cloud as the cannon mud point cloud, and searching for the highest point coordinates.

5. The method for selecting the optimal grasping point of a blast furnace taphole mud robot according to claim 1, characterized in that: In the step 3), the slice data is segmented into point clouds as follows: a clustering method is used to classify the point clouds, and the point clouds are preliminarily segmented to obtain basic block-based artillery mud point clouds. Then, a watershed algorithm is used to perform edge recognition on the preliminarily segmented artillery mud point clouds to segment accurate block-based artillery mud point clouds.

6. The method for selecting the optimal grasping point of a blast furnace taphole mud robot according to claim 1, characterized in that: In the step 4), plane fitting is performed on the segmented point cloud, and the optimal gun mud point cloud is selected on the segmented point cloud fitting plane; the details are as follows: 4.1) Perform plane fitting on each separated cannon clay point cloud, select point clouds with plane angles less than a threshold, and sort the area of ​​the plane fitting of each cannon clay point cloud from large to small; 4.2) The first K blocks of cannon mud point clouds in the area sorting and the point clouds without specific interference in the point cloud neighborhood are taken as candidates for grabbing cannon mud point clouds. The evaluation objective function is set for the candidate point clouds according to the area, plane inclination angle and height to determine the optimal grabbing cannon mud; The point cloud neighborhood range having no specific interference means that within the range of the selected point cloud with the robot arm fixture disk as the center, there is no point higher than the point cloud.

7. The method for selecting the optimal grasping point of a blast furnace taphole mud robot according to claim 5, characterized in that: In the step 4), if there is no candidate for capturing the point cloud of the cannon mud, then for the segmented point cloud, all the points in the neighborhood are traversed to obtain all the local highest points in the cannon mud point cloud, and the plane is segmented and fitted within the neighborhood of the local highest point, and steps 4.1) and 4.2) are repeated to obtain the optimal cannon mud point cloud.

8. The method for selecting the optimal grasping point of a blast furnace taphole mud robot according to claim 5, characterized in that: In step 4.1), a sampling consistency algorithm is used to perform plane fitting.

9. An electronic device, characterized in that: include: one or more processors; as well as a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors are caused to perform the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images