Intelligent overhaul grinding system and method for medium and heavy plate surface with efficient multi-task collaborative allocation

Abstract

The present invention belongs to the technical field of medium and heavy plate defect grinding, and discloses an intelligent surface inspection and grinding system and method for medium and heavy plates with multi-task efficient collaborative allocation. The system includes a medium and heavy plate defect detection system, a task allocation system, and a robot grinding system. The medium and heavy plate defect detection system obtains the position and depth information of the defects. The task allocation system receives the defect information in real time, and through dividing the working area of the steel plate and the constraint of preventing robot collision, uses a consensus-based greedy algorithm to allocate the tasks to the robot grinding system in real time. The robot grinding system starts the steel plate positioning and grinding program according to the defect information, and reinspects the grinding area after grinding is completed. The present invention realizes the intelligent surface inspection and grinding of medium and heavy plates with multi-task efficient collaborative allocation, meets the on-line grinding of robots on the finishing line of medium and heavy plates, replaces manual inspection and grinding, and improves the automation level of the production line.

Description

Technical Field

[0001] The present invention relates to the technical field of defect grinding of medium and heavy plates, and particularly relates to an intelligent inspection and grinding system and method for the surface of medium and heavy plates with efficient multi-task collaborative allocation. Background Art

[0002] Traditional surface inspection and grinding of medium and heavy plates rely on manual operation, lacking effective collaborative division of labor. Workers need to independently complete multiple tasks, such as inspection, defect identification, and grinding. The lack of coordination and clear division of labor leads to frequent task switching, increasing labor intensity and fatigue, and reducing efficiency. Poor information transmission and low collaborative efficiency easily result in missed defects or uneven grinding. Uneven task allocation affects the work progress, thus affecting the overall production efficiency and the stability of surface quality. In addition, due to the harsh working environment, a large amount of dust, high temperature, and noise are generated during the grinding process. Workers are exposed to this environment for a long time and face greater health and safety risks.

[0003] In view of the above problems, the present invention provides an intelligent inspection and grinding system and method for the surface of medium and heavy plates with efficient multi-task collaborative allocation. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides an intelligent inspection and grinding system and method for the surface of medium and heavy plates with efficient multi-task collaborative allocation. Through intelligent scheduling and precise task allocation, the system optimizes the work process, improves the collaborative efficiency, and avoids the uncertainties in manual operation. It reduces worker fatigue and operation errors, reduces health risks, and significantly improves production efficiency and product quality, ensuring that each steel plate meets the ideal quality standard before leaving the factory and enhancing the overall efficiency of the production line.

[0005] The present invention discloses an intelligent inspection and grinding system for the surface of medium and heavy plates with efficient multi-task collaborative allocation. The system includes: a medium and heavy plate defect detection system, a task allocation system, and a robot grinding system;

[0006] The medium and heavy plate defect detection system is composed of multiple juxtaposed cameras for image acquisition and multiple line laser three-dimensional scanning components, and is used for online scanning the surface of the steel plate to obtain defect position information and depth information, and transmitting the obtained defect information to the task allocation system;

[0007] The task allocation system is used for receiving the defect information on the surface of the steel plate in real time, and through the constraints of dividing the working area of the steel plate and preventing robot collisions, using a consensus-based greedy algorithm to allocate tasks to the robot grinding system in real time;

[0008] The described robot grinding system is used to obtain the positioning coordinates of the steel plate through a laser displacement sensor, start the corresponding grinding process according to the defect information sent by the medium-thick plate defect detection system, and use a 3D structured light camera to re-inspect the grinding effect of the grinding area after grinding is completed.

[0009] Preferably, the constraints to prevent robot collisions include:

[0010] The steel plate is divided into multiple grinding areas, including independent areas and consensus areas. Among them, the independent area is an area that only one robot can grind by itself, and the consensus area is an interval where multiple robots can grind together;

[0011] The obtained defect information is allocated to the grinding area where it is located;

[0012] According to the working range of the collaborative robot, an independent task list and a consensus task list are set for the robot to prevent interference and collision of the robot during grinding of the steel plate;

[0013] Calculate the motion envelope of the robot to prevent the motion envelopes of the collaborative robots from overlapping.

[0014] Preferably, the process of using the consensus-based greedy algorithm to allocate tasks to the robot grinding system in real time includes:

[0015] Each robot calculates its own fitness for executing a certain task, and determines the probability of task selection according to the fitness;

[0016] When entering the consensus area, the robot needs to ensure that only one robot is responsible for generating the task list of the current area;

[0017] Through the locking mechanism, the robot checks and updates the consensus task list to ensure that only one robot can create or update the task list of the current area;

[0018] In the consensus area, the robot selects tasks according to the greedy strategy and selects the task closest to itself.

[0019] Preferably, obtaining the positioning coordinates of the steel plate through a laser displacement sensor includes:

[0020] Using the laser displacement sensor to obtain the two-point coordinates of the two ends of the long edge of the steel plate by running the robot positioning instruction , and the coordinate of one point on the wide edge ;

[0021] Using the three-point method to calculate the relative coordinates of the steel plate, and setting the steel plate workpiece coordinates of the robot through the obtained intersection coordinates and the angle with the horizontal axis Set the steel plate workpiece coordinates of the robot.

[0022] Preferably, the process of using a 3D structured light camera to re-inspect the grinding effect of the grinding area after grinding includes:

[0023] Use a 3D structured light camera to re-inspect the grinding area and obtain the point cloud data after grinding;

[0024] According to the point cloud data, calculate the depth after grinding, that is, the re-inspection depth;

[0025] Use the RANSAC algorithm (Random Sample Consensus algorithm) to perform plane fitting on the point cloud data to determine the plane reference of the steel plate and the plane reference after grinding , and then after adjusting the two planes to be parallel, calculate the distance between the two planes , ensuring the error between the pre-grinding depth and the actual grinding depth.

[0026] The present invention also provides an intelligent maintenance grinding method for the surface of medium-thick plates with efficient multi-task collaborative allocation. The method is implemented by using any one of the systems described above. The method includes:

[0027] S1: The medium-thick plate defect detection system collects image defects through a camera, and the line laser three-dimensional scanning component scans the steel plate online to provide defect depth information;

[0028] S2: After the steel plate passes through the optoelectronic switch, the roller table stops, and the steel plate positioning of the robot grinding system is started to obtain the pose coordinates of the steel plate;

[0029] S3: Use the task allocation system to obtain the defect depth and corresponding position information, and through the constraints of dividing the working area of the steel plate and preventing robot collisions, use the consensus-based greedy algorithm to allocate the tasks to the robot grinding system in real time;

[0030] S4: The robot grinding system starts the corresponding grinding program by receiving the defect task;

[0031] S5: Collect the grinding chips through the chip collection device to prevent blocking the grinding marks and polluting the production environment;

[0032] S6: According to the grinding requirements, use a 3D structured light camera to detect the grinding area to ensure that the grinding depth reaches the preset requirements;

[0033] S7: Re-inspect the grinding area using a 3D structured light camera;

[0034] S8: For the remaining defect problems after grinding, use a 3D structured light camera to re-inspect the depth of the defects in the grinding area, and perform the next grinding and re-inspection again until the defects are ground and meet the preset requirements;

[0035] S9: According to the requirements of the production line, coordinate the control of multiple robots for grinding operations to achieve efficient grinding of medium and heavy plates.

[0036] Preferably, in the above-mentioned S2, the method for the steel plate to stop the roller table after passing through the photoelectric switch, start the steel plate positioning, and obtain the pose coordinates of the steel plate includes:

[0037] Place the photoelectric switch under the roller table. After the steel plate passes through the photoelectric switch, the feedback signal stops the roller table, and the robot receives the start of the roller table to perform the corresponding steel plate positioning;

[0038] Run the robot positioning command through the laser displacement sensor to obtain the two-point coordinates of the two ends of the long edge of the steel plate , and the coordinate of one point on the wide edge ;

[0039] Use the three-point method to calculate the relative coordinates of the steel plate. The formula is:

[0040] Calculate the slope and the intercept :

[0041] ;

[0042] ;

[0043] Calculate the slope and the intercept :

[0044] ;

[0045] ;

[0046] Calculate the intersection coordinates of the long side and the wide side :

[0047] ;

[0048] ;

[0049] Calculate the angle with the horizontal axis :

[0050] ;

[0051] Set the steel plate workpiece coordinates of the robot through the obtained intersection coordinates and the angle with the horizontal axis.

[0052] Preferably, in the above-mentioned S3, dividing the working area by the steel plate includes:

[0053] Expand the length L and width W of the defect according to the depth D of the defect; the expansion formula is:

[0054] ;

[0055] ;

[0056] where, is the actual grinding length, is the actual grinding width;

[0057] Determine the reference point at the lower right corner of the working area by aligning the grinding position of the tool with the defect area; the coordinates of the reference point are calculated as:

[0058] ;

[0059] ;

[0060] where, is the reference point coordinate, is the reference point coordinate, is the starting point of grinding coordinate, is the starting point of grinding coordinate, is the total length of the grinding equipment;

[0061] Combine the length and width of the tool with the expanded defect area to calculate the final length and width of the working area:

[0062] ;

[0063] ;

[0064] where, is the final length of the working area, C W is the final width of the working area, is the total width of the grinding equipment;

[0065] Assume the rotation angle is , and use the rotation matrix to calculate the coordinates of the four vertices of the rotated working area:

[0066] ;

[0067] ;

[0068] where, is the vertex coordinate before rotation, is the vertex coordinate after rotation;

[0069] After rotation, take the minimum and maximum values of the X and Y coordinates of the four vertices respectively to determine the boundary of the working area:

[0070] ;

[0071] Among them, is the minimum boundary of the working area, is the maximum boundary of the working area, is the minimum boundary of the working area, is the maximum boundary of the working area.

[0072] Preferably, in step S3, the method of using a consensus-based greedy algorithm to allocate tasks to the robotic grinding system in real time includes:

[0073] Each robot calculates its own fitness for executing a certain task, and determines the probability of task selection according to the fitness; the probability of task selection is:

[0074] ;

[0075] Among them, is the task selection probability, is the sum of the fitness scores of robot i for all tasks, is the fitness of robot i to complete the task, that is, the score;

[0076] When entering the consensus area, the robot needs to ensure that only one robot is responsible for generating the task list of the current area;

[0077] Through the locking mechanism, the robot checks and updates the consensus task list to ensure that only one robot can create or update the task list of the current area;

[0078] Within the consensus area, the robot selects tasks according to the greedy strategy and selects the task closest to itself.

[0079] Preferably, in step S6, the method of using a 3D structured light camera to re-inspect the grinding area includes:

[0080] Use a 3D structured light camera to re-inspect the grinding area to obtain the point cloud data after grinding;

[0081] According to the point cloud data, calculate the depth after grinding, that is, the re-inspection depth, and detect the depth and surface smoothness of the grinding area to ensure that the depth error does not exceed 0.05 mm; the specific method is:

[0082] Use the RANSAC algorithm to perform plane fitting on the point cloud data to determine the plane reference of the steel plate and the plane reference after grinding , the formula is as follows:

[0083] Randomly select the minimum number of points from the point cloud data to fit the model. For plane fitting, select three points , , to determine a plane;

[0084] Calculate the normal vector :

[0085] ;

[0086] After expansion, the normal vector of the plane is :

[0087] ;

[0088] ;

[0089] ;

[0090] Substitute a known point into the plane equation to calculate the constant term in the plane equation :

[0091] ;

[0092] Calculate the number of inliers. Substitute all points into the plane equation and calculate the distance from the inliers to the plane;

[0093] If the distance from a certain point to the plane is less than the preset threshold , then the corresponding point is considered an inlier;

[0094] The distance formula is as follows:

[0095] ;

[0096] Among them, is the distance from the point to the plane, is a point in the point cloud. Repeat the iteration until the model with the most inliers is found or the iteration count reaches the upper limit. The model with the most inliers is the best-fitting plane model;

[0097] By calculating the rotation matrix, rotate the normal vector of the grinding plane to the normal vector of the reference plane;

[0098] The rotation axis is the cross product of the two normal vectors:

[0099] ;

[0100] Among them, is the rotation axis. If the cross product result is the zero vector, the two normal vectors are already parallel and no rotation is required;

[0101] Calculate the angle θ between the two normal vectors:

[0102] ;

[0103] Use the Rodrigues rotation formula to calculate the rotation matrix according to the rotation axis and rotation angle :

[0104] ;

[0105] where, is the identity matrix, is the skew-symmetric matrix of the rotation axis, defined as follows:

[0106] ;

[0107] where, is the component of the rotation axis ;

[0108] After adjusting the two planes to be parallel, calculate the distance between the two planes :

[0109] ;

[0110] where, is the common normal vector of the two planes, and are the constant terms in the equations of the two planes respectively, ensuring that the error does not exceed 0.05.

[0111] Compared with the prior art, the beneficial effects of the present invention are:

[0112] The intelligent surface maintenance grinding system for medium and heavy plates with multi-task high-efficiency collaborative allocation of the present invention realizes the full-process intelligent operation from defect detection to grinding by integrating defect collection, steel plate positioning, task real-time allocation, grinding re-inspection and grinding technology.

[0113] The task allocation system assigns defects to the robot grinding system for grinding according to the defect depth, position information, anti-interference method and consensus-based greedy algorithm, and realizes the standardization and consistency of the grinding process through the preset steel plate positioning, grinding program and grinding re-inspection process.

[0114] The ground rail equipped with the robotic grinding system enables the robot to move flexibly instead of being fixed in one position, and it can automatically adjust its position according to the length of the steel plate for grinding. With this design, the robot can cover the entire surface of the steel plate to ensure large-scale grinding. Regardless of the change in the size of the steel plate, the robot can adjust its working position in real time to achieve efficient and continuous grinding operations, significantly improving the automation level and grinding accuracy of the production line. At the same time, it optimizes the efficiency of the grinding process and reduces manual intervention.

[0115] The chip collection equipment equipped with the robotic grinding system effectively collects the grinding chips generated during the grinding process, reducing the risk of workers inhaling harmful dust, thus significantly reducing the labor intensity and the health risk of workers. The system can achieve continuous production without frequent shutdowns due to manual operations, significantly improving the overall efficiency of the production line.

[0116] Through the integration with other automated equipment on the production line, full automation management and intelligent monitoring are achieved, enabling timely detection and resolution of problems during the production process to ensure the stability and efficiency of the production process, thereby significantly enhancing the market competitiveness of products and the overall effectiveness of the production line. Brief Description of the Drawings

[0117] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0118] Figure 1 It is the overall logic flow chart of an intelligent inspection and grinding method for the surface of medium and heavy plates with multi-task high-efficiency collaborative allocation in the embodiments of the present invention;

[0119] Figure 2 It is the task allocation logic flow chart in the embodiments of the present invention;

[0120] Figure 3 It is the area division logic flow chart in the embodiments of the present invention;

[0121] Figure 4 It is a schematic diagram of a grinding robot in the embodiments of the present invention;

[0122] Figure 5 It is the overall system structure diagram of the intelligent inspection and grinding system for the surface of medium and heavy plates in the embodiments of the present invention.

[0123] In the figure: 10, medium-thick plate to be measured; 20, roller table; 30, camera; 40, line laser three-dimensional scanning component; 50, robot control cabinet; 60, robot walking ground rail; 70, grinding robot; 80, optoelectronic switch; 90, chip box; 100, chip collection port; 110, grinding wheel; 120, electric spindle; 130, laser displacement sensor; 140, force control compensator. Specific implementation mode

[0124] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0125] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should have the ordinary meaning understood by those of ordinary skill in the art within the field to which the present disclosure belongs. The "first", "second" and similar terms used in the embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before the term cover the elements or objects listed after the term and their equivalents, without excluding other elements or objects. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left" and "right" are only used to represent relative position relationships. When the absolute position of the object to be described changes, the relative position relationship may also change accordingly.

[0126] To make the above objects, features and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation modes.

[0127] Embodiment 1

[0128] As Figure 5 shown, the embodiment of the present invention provides a multi-task efficient collaborative distribution intelligent surface inspection and grinding system for medium-thick plates, and the system includes: a medium-thick plate defect detection system, a task distribution system and a robot grinding system;

[0129] The medium-thick plate defect detection system is composed of a plurality of juxtaposed cameras 30 for collecting images and a plurality of line laser three-dimensional scanning components 40, and is used for online scanning the surface of the steel plate to obtain defect position information and depth information, and transmitting the obtained defect information to the task distribution system;

[0130] A task allocation system is used to receive defect information on the steel plate surface in real time. Through the constraints of dividing the working area of the steel plate and preventing robot collisions, a consensus-based greedy algorithm is used to allocate tasks to the robot grinding system in real time;

[0131] A robot grinding system is used to obtain the positioning coordinates of the steel plate through a laser displacement sensor 130. According to the defect information sent by the medium-thick plate defect detection system, it starts the corresponding grinding process. After grinding is completed, a 3D structured light camera is used to re-inspect the grinding effect of the grinding area.

[0132] In this embodiment, as Figure 2 shown, the constraints for preventing robot collisions include:

[0133] The steel plate is divided into multiple grinding areas, including independent areas and consensus areas. Among them, the independent area is an area that only one robot can grind by itself, and the consensus area is an interval where multiple robots can grind together;

[0134] The obtained defect information is allocated to the grinding area where it is located;

[0135] According to the working intervals of the collaborative robots, independent task lists and consensus task lists are set for the robots to prevent interference and collisions when the robots are grinding the steel plate;

[0136] Calculate the motion envelope of the robot to prevent the motion envelopes of the collaborative robots from overlapping.

[0137] In this embodiment, as Figure 3 shown, the process of using a consensus-based greedy algorithm to allocate tasks to the robot grinding system in real time includes:

[0138] Calculate the task selection probability: In order to allocate tasks within the consensus area, each robot calculates its own fitness for executing a certain task (such as distance, current workload, etc.) and determines the probability of task selection based on this;

[0139] When entering the consensus area, the robot needs to ensure that only one robot is responsible for generating the task list for the current area;

[0140] Through a locking mechanism, the robot checks and updates the consensus task list to ensure that only one robot can create or update the task list for the current area, thereby preventing multiple robots from repeatedly selecting the same task;

[0141] Within the consensus area, the robot selects tasks according to the greedy strategy, choosing the task closest to itself; the specific steps are as follows:

[0142] Lock the task selection process: Before each task selection, the robot locks the entire task selection and execution process to prevent other robots from simultaneously selecting the same task;

[0143] Task selection: Select the task with the shortest distance and that has not been assigned. For the working area of the selected task, check whether it overlaps with the current working areas of other robots to ensure that task assignment does not cause interference.

[0144] Task execution: Execute the selected task. After completion, mark the task as assigned and unlock it. After the task is completed, update the task status to the global consensus task list through the consensus mechanism.

[0145] In the independent area, all tasks are selected according to the greedy algorithm and the working areas of the robots are assigned.

[0146] In the consensus area, if all tasks cannot be successfully assigned (interference conflicts occur), these unassigned tasks will be added to the global conflict task list and wait for re-assignment.

[0147] The robot can regularly check the conflict task list, make greedy selections and assignments for the unfinished conflict tasks until all tasks are completed.

[0148] In this embodiment, obtaining the positioning coordinates of the steel plate by the laser displacement sensor 130 includes:

[0149] Using the laser displacement sensor 130 to obtain the two-point coordinates of the two ends of the long side of the steel plate by running the robot positioning instruction , and the one-point coordinate of the wide side edge ;

[0150] Using the three-point method to calculate the relative coordinates of the steel plate, and setting the steel plate workpiece coordinates of the robot through the obtained intersection coordinates and the angle with the horizontal axis .

[0151] In this embodiment, as Figure 4 shown, the robot grinding system includes a robot, an electric spindle 120, a force control compensator 140, a 3D structured light camera, a laser displacement sensor 130, a robot walking ground rail 60 and a chip collecting device;

[0152] The robot uses the laser displacement sensor 130 to measure and calculate the coordinate pose of the steel plate reached.

[0153] The robot is used to set the corresponding grinding program for the received depth and corresponding position information of the defect.

[0154] The robot is also used to move laterally through the robot walking ground rail 60 and adjust the position according to the change of the steel plate length, so as to realize the extension of the grinding area and the precise grinding of the entire range of the steel plate.

[0155] The chip collection device is used to collect the grinding chips on the stainless - steel surface;

[0156] The 3D structured - light camera is used to re - inspect the grinding area to verify whether the grinding meets the requirements;

[0157] The electric spindle 120 is used to provide the rotational power for the grinding tool and adjust the grinding speed and force;

[0158] The force - control compensator 140 is used to adjust the downward pressure of the grinding head in real - time to ensure a constant pressure during the grinding process;

[0159] The force - control compensator 140 is also used to monitor the downward pressure during grinding in real - time and feed it back to the robot;

[0160] The robot is also used to dynamically adjust the speed of the electric spindle 120 and the downward pressure of the force - control compensator 140 according to the feedback information.

[0161] In this embodiment, the process of using the 3D structured - light camera to re - inspect the grinding effect of the grinding area after grinding includes:

[0162] Using the 3D structured - light camera to re - inspect the grinding area to obtain the point - cloud data after grinding;

[0163] According to the point - cloud data, calculate the grinding depth after grinding, that is, the re - inspection depth;

[0164] Use the RANSAC algorithm to perform plane fitting on the point - cloud data to determine the plane reference of the steel plate and the plane reference after grinding , and then adjust the two planes to be parallel and calculate the distance between the two planes , ensuring the error between the pre - grinding depth and the actual grinding depth.

[0165] The system of this embodiment is used to implement the intelligent surface inspection and grinding method for medium - thick plates with efficient multi - task collaborative allocation corresponding to any one of the following embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be elaborated here.

[0166] It should be noted that a system for intelligent surface inspection and grinding of medium - thick plates with efficient multi - task collaborative allocation is embodied in the form of functional units. The term "module" here can be implemented in the form of software and / or hardware, and no specific limitation is made in this regard.

[0167] For example, a "module" can be a software program, a hardware circuit, or a combination of both that implements the above functions. The hardware circuit may include an application specific integrated circuit (ASIC), an electronic circuit, a processor (such as a shared processor, a proprietary processor, or a group of processors, etc.) for executing one or more software or firmware programs, a memory, a merged logic circuit, and / or other suitable components that support the described functions.

[0168] Embodiment 2

[0169] Based on the same inventive concept, corresponding to any of the above-described embodiment systems, as Figure 1 shown, the present invention also provides a method for intelligent inspection and grinding of the surface of medium and heavy plates with efficient multi-task collaborative allocation. The method is implemented by applying any of the above-described systems, and the method includes:

[0170] S1: The medium and heavy plate defect detection system collects image defects through a camera, and the line laser three-dimensional scanning component 40 scans the steel plate online to provide defect depth information;

[0171] S2: After the steel plate passes through the optoelectronic switch 80, the roller path 20 stops, and the steel plate positioning of the robot grinding system is started to obtain the pose coordinates of the steel plate;

[0172] S3: Using the task allocation system to obtain the defect depth and the corresponding position information, through the constraints of dividing the working area of the steel plate and preventing robot collisions, the task is allocated to the robot grinding system in real time using a consensus-based greedy algorithm;

[0173] S4: The robot grinding system starts the corresponding grinding program by receiving the defect task;

[0174] S5: The grinding chips are collected through the chip collection device to prevent blocking the grinding marks and polluting the production environment;

[0175] S6: According to the grinding requirements, the 3D structured light camera is used to detect the grinding area to ensure that the grinding depth reaches the preset requirements;

[0176] S7: Re-inspection of the grinding area using the 3D structured light camera;

[0177] S8: For the defect problems still existing after grinding, the depth of the defects in the grinding area is re-inspected using the 3D structured light camera, and the next grinding and re-inspection are carried out again until the defects are ground and meet the preset requirements;

[0178] S9: According to the requirements of the production line, multiple robots are coordinated to perform grinding operations to achieve efficient grinding of medium and heavy plates.

[0179] In this embodiment, in S2, after the steel plate passes through the photoelectric switch 80, the roller path 20 stops, and the method for starting the steel plate positioning and obtaining the pose coordinates of the steel plate includes:

[0180] Place the photoelectric switch 80 below the roller path 20. After the steel plate passes through the photoelectric switch 80, the feedback signal stops the roller path 20, and the robot receives the start of the corresponding steel plate positioning of the roller path 20;

[0181] Run the robot positioning instruction through the laser displacement sensor 130 to obtain the two-point coordinates of the two ends of the long edge of the steel plate , and the coordinate of one point on the wide edge ;

[0182] Use the three-point method to calculate the relative coordinates of the steel plate. The formula is:

[0183] Calculate the slope of the long side and the intercept :

[0184] ;

[0185] ;

[0186] Calculate the slope of the wide side and the intercept :

[0187] ;

[0188] ;

[0189] Calculate the intersection coordinates of the long side and the wide side :

[0190] ;

[0191] ;

[0192] Calculate the angle with the horizontal axis :

[0193] ;

[0194] Set the steel plate workpiece coordinates of the robot through the obtained intersection coordinates and the angle with the horizontal axis.

[0195] In this embodiment, in S3, dividing the working area by the steel plate includes:

[0196] Divide the steel plate into multiple grinding areas, including independent areas and consensus areas;

[0197] The independent area is the area that only one robot can polish by itself;

[0198] The consensus area is the interval where multiple robots can polish together;

[0199] Allocate the obtained defect information to the grinding area at its location;

[0200] For the working area of the collaborative robot, set an independent task list and a consensus task list for the robot;

[0201] To prevent the robots from interfering and colliding during the grinding of the steel plate, calculate the motion envelope of the robots to prevent the motion envelopes of the collaborative robots from overlapping; The specific method is as follows:

[0202] According to the depth D of the defect, extend the length L and width W of the defect; The extension formula is:

[0203] ;

[0204] ;

[0205] Among them, is the actual grinding length, is the actual grinding width;

[0206] By aligning the grinding position of the tool with the defect area, determine the reference point at the lower right corner of the working area; The coordinates of the reference point are calculated as:

[0207] ;

[0208] ;

[0209] Among them, is the reference point coordinate, is the reference point coordinate, is the starting point of grinding coordinate, is the starting point of grinding coordinate, is the total length of the grinding equipment;

[0210] Combine the length and width of the tool with the extended defect area to calculate the final length and width of the working area:

[0211] ;

[0212] ;

[0213] Among them, is the final length of the working area, CW is the final width of the working area, is the total width of the grinding equipment;

[0214] Assume the rotation angle is , and use the rotation matrix to calculate the coordinates of the four vertices of the rotated working area:

[0215] ;

[0216] ;

[0217] Among them, are the vertex coordinates before rotation, are the vertex coordinates after rotation;

[0218] After rotation, take the minimum and maximum values of the X and Y coordinates of the four vertices respectively to determine the boundary of the working area:

[0219] ;

[0220] Among them, is the minimum boundary of the working area, is the maximum boundary of the working area, is the minimum boundary of the working area, is the maximum boundary of the working area.

[0221] In this embodiment, in step S3, the method of using the consensus-based greedy algorithm to allocate tasks to the robot grinding system in real time includes:

[0222] Each robot calculates its own fitness for executing a certain task, and determines the probability of task selection according to the fitness; the probability of task selection is:

[0223] ;

[0224] Among them, is the task selection probability, is the sum of the fitness scores of robot i for all tasks, is the fitness of robot i to complete the task, that is, the score;

[0225] When entering the consensus area, the robot needs to ensure that only one robot is responsible for generating the task list of the current area;

[0226] Through the locking mechanism, the robot checks and updates the consensus task list to ensure that only one robot can create or update the task list of the current area;

[0227] Within the consensus area, the robot selects tasks according to the greedy strategy and selects the task closest to itself.

[0228] Among them, within the consensus area, the robot selects tasks according to the greedy strategy and chooses the task closest to itself; the specific steps are as follows:

[0229] Lock the task selection process: Before each task selection, the robot locks the entire task selection and execution process to prevent other robots from simultaneously selecting the same task;

[0230] Task selection: Select the nearest unassigned task. For the working area of the selected task, check whether it overlaps with the current working areas of other robots to ensure that task allocation does not cause interference;

[0231] Task execution: Execute the selected task. After completion, mark the task as assigned and unlock it; after the task is completed, update the task status to the global consensus task list through the consensus mechanism;

[0232] In the independent area, all tasks are selected according to the greedy algorithm and the working areas of the robots are allocated;

[0233] In the consensus area, if all tasks cannot be successfully allocated (interference conflicts occur), these unallocated tasks will be added to the global conflict task list and wait for reallocation;

[0234] The robot can regularly check the conflict task list and perform greedy selection and allocation on the unfinished conflict tasks until all tasks are completed.

[0235] In this embodiment, in S6, the method for re-inspecting the grinding area using a 3D structured light camera includes:

[0236] Use a 3D structured light camera to re-inspect the grinding area and obtain the point cloud data after grinding;

[0237] According to the point cloud data, calculate the depth after grinding, that is, the re-inspection depth, and detect the depth and surface smoothness of the grinding area to ensure that the depth error does not exceed 0.05 mm; the specific method is as follows:

[0238] Use the RANSAC algorithm to perform plane fitting on the point cloud data to determine the plane reference of the steel plate and the plane reference after grinding , the formula is as follows:

[0239] Randomly select the minimum number of points from the point cloud data to fit the model. For plane fitting, select three points , , to determine a plane;

[0240] Calculate the normal vector :

[0241] ;

[0242] After expansion, the normal vector of the plane is obtained :

[0243] ;

[0244] ;

[0245] ;

[0246] Substitute a known point into the plane equation to calculate the constant term in the plane equation :

[0247] ;

[0248] Calculate the number of inliers. Substitute all points into the plane equation and calculate the distance from the inliers to the plane;

[0249] If the distance from a certain point to the plane is less than the preset threshold , then the corresponding point is considered an inlier;

[0250] The distance formula is as follows:

[0251] ;

[0252] Among them, is the distance from the point to the plane, is a point in the point cloud. Repeat the iteration until the model with the most inliers is found or the iteration count reaches the upper limit. The model with the most inliers is the best-fitting plane model;

[0253] The normal vector of the grinding plane is inconsistent with that of the reference plane, and the grinding plane needs to be adjusted by rotation. The rotation matrix can be calculated through the following steps to rotate the normal vector of the grinding plane to the normal vector of the reference plane;

[0254] The rotation axis is the cross product of the two normal vectors:

[0255] ;

[0256] Among them, is the rotation axis. If the cross product result is a zero vector, the two normal vectors are already parallel and no rotation is required;

[0257] Calculate the angle θ between the two normal vectors:

[0258] ;

[0259] Calculate the rotation matrix using the Rodrigues rotation formula according to the rotation axis and rotation angle :

[0260] ;

[0261] wherein, is the identity matrix, is the skew-symmetric matrix of the rotation axis, defined as follows:

[0262] ;

[0263] wherein, is the rotation axis components of;

[0264] After adjusting the two planes to be parallel, calculate the distance between the two planes :

[0265] ;

[0266] wherein, is the common normal vector of the two planes, and are the constant terms in the equations of the two planes respectively, ensuring that the error does not exceed 0.05.

[0267] It should be noted that the method of the embodiments of the present disclosure can be executed by a single device, such as a computer or a server, etc. The method of this embodiment can also be applied to a distributed scenario, and multiple devices cooperate with each other to complete it. In this case of a distributed scenario, one of the multiple devices can only execute one or more steps of the method of the embodiments of the present disclosure, and these multiple devices will interact with each other to complete the described method.

[0268] It should be noted that some embodiments of the present disclosure have been described above. Other embodiments are within the scope of the appended invention content. In some cases, it should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention. The actions or steps recorded in the invention content can be executed in a different order from that in the above embodiments and still achieve the desired result. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0269] Embodiments of the present disclosure are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended disclosure of the invention. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present disclosure shall be included within the protection scope of the present disclosure.

Claims

1. A multi-task efficient collaborative distribution medium and thick plate surface intelligent grinding system, characterized by: The system includes: a medium and thick plate defect detection system, a task allocation system and a robot grinding system; The medium and thick plate defect detection system is composed of multiple parallel cameras for collecting images and multiple line laser three-dimensional scanning components, which are used to scan the surface of the steel plate online, obtain defect location information and depth information, and transmit the obtained defect information to the task allocation system; The task allocation system is used to receive the defect information of the steel plate surface in real time, divide the working area by the steel plate and prevent the robot from colliding, and use the consensus-based greedy algorithm to allocate the task to the robot grinding system in real time; The robot grinding system is used to obtain the positioning coordinates of the steel plate through the laser displacement sensor, start the corresponding grinding process according to the defect information sent by the medium and thick plate defect detection system, and use the 3D structured light camera to re-check the grinding effect of the grinding area after the grinding is completed; The constraints that prevent the robot from colliding are: Divide the steel plate into multiple grinding areas, including independent areas and consensus areas, wherein the independent area is an area that can be ground by only one robot, and the consensus area is an area that can be ground by multiple robots together; Assign the acquired defect information to the grinding area at the location; According to the collaborative robot's working range, set an independent task list and a consensus task list for the robot to prevent interference and collision when the robot is grinding steel plates; Calculate the robot's motion envelope to prevent the motion envelopes of collaborative robots from overlapping.

2. The system according to claim 1, characterized in that The process of allocating tasks to the robotic grinding system in real time using a consensus-based greedy algorithm includes: Each robot calculates its fitness to perform a certain task and determines the probability of task selection based on the fitness; When entering a consensus area, the robot needs to ensure that only one robot is responsible for generating the task list for the current area; Through the locking mechanism, the robot checks and updates the consensus task list to ensure that only one robot can create or update the task list of the current area; In the consensus area, the robot selects tasks according to the greedy strategy and chooses the task closest to itself.

3. The system according to claim 1, characterized in that The positioning coordinates of the steel plate obtained by the laser displacement sensor include: Use the laser displacement sensor to obtain the coordinates of the two ends of the long side of the steel plate by running the robot positioning command , and the coordinates of a point on the wide side ; The relative coordinates of the steel plate are calculated using the three-point method, and the intersection coordinates are obtained Angle with horizontal axis Set the robot's steel plate workpiece coordinates.

4. The system according to claim 1, characterized in that After grinding is completed, the process of using a 3D structured light camera to recheck the grinding effect of the grinding area includes: Use a 3D structured light camera to re-inspect the grinding area and obtain point cloud data after grinding; According to the point cloud data, the depth after grinding, i.e. the re-inspection depth, is calculated; Use the RANSAC algorithm to perform plane fitting on the point cloud data to determine the plane reference of the steel plate And the polished plane reference , and then adjust the two planes to be parallel, and calculate the distance between the two planes , to ensure the error between the pre-grinding depth and the actual grinding depth.

5. A method for intelligent surface maintenance and grinding of medium and thick plates with efficient collaborative allocation of multiple tasks, the method is implemented by the system according to any one of claims 1 to 4, characterized in that: The method comprises: S1: The medium and thick plate defect detection system collects image defects through the camera, and the line laser 3D scanning component scans the steel plate online to provide defect depth information; S2: After the steel plate passes through the photoelectric switch, the roller stops, the steel plate positioning of the robot grinding system is started, and the position coordinates of the steel plate are obtained; S3: The task allocation system is used to obtain the defect depth and corresponding location information, and the work area is divided by steel plates and the robot collision prevention constraints are used to allocate tasks to the robot grinding system in real time using a consensus-based greedy algorithm; S4: The robot grinding system starts the corresponding grinding program by receiving the defect task; S5: Grinding chips are collected by chip collection equipment to prevent them from covering the grinding marks and polluting the production environment; S6: Based on the grinding requirements, the grinding area is inspected using a 3D structured light camera to ensure that the grinding depth meets the preset requirements; S7: Re-inspect the grinding area using a 3D structured light camera; S8: If there are still defects after grinding, use a 3D structured light camera to re-check the depth of the defect in the grinding area, and perform the next grinding and re-inspection again until the defect grinding is completed and meets the preset requirements; S9: According to the needs of the production line, multiple robots are collaboratively controlled to perform grinding operations to achieve efficient grinding of medium and thick plates.

6. The method according to claim 5, characterized in that In S2, after the steel plate passes through the photoelectric switch, the roller is stopped, the steel plate positioning is started, and the method for obtaining the steel plate position coordinates includes: Place the photoelectric switch under the roller. After the steel plate passes through the photoelectric switch, the feedback signal stops the roller. The robot receives the corresponding steel plate positioning when the roller starts. The laser displacement sensor is used to run the robot positioning command to obtain the coordinates of the two ends of the long side of the steel plate. , and the coordinates of a point on the wide side ; The three-point method is used to calculate the relative coordinates of the steel plate. The formula is: Calculate the slope of the long side and the intercept : ； ； Calculate the slope of the broadside and the intercept : ； ； Calculate the coordinates of the intersection of the long side and the wide side : ； ； Calculate the angle with the horizontal axis : ； By obtaining the intersection coordinates Angle with horizontal axis Set the robot's steel plate workpiece coordinates.

7. The method according to claim 5, characterized in that In S3, dividing the working area by steel plates includes: According to the depth D of the defect, the length L and width W of the defect are expanded; the expansion formula is: ； ； in, is the actual grinding length, is the actual grinding width; The lower right corner reference point of the working area is determined by aligning the grinding position of the tool with the defect area; the coordinates of the reference point are calculated as: ； ； in, Is the reference point coordinate, Is the reference point coordinate, The starting point for grinding coordinate, The starting point for grinding coordinate, is the total length of the grinding equipment; Combining the length and width of the tool with the expanded defect area, calculate the final length and width of the working area: ； ； in, is the final length of the working area, C W is the final width of the working area, is the total width of the grinding equipment; Assume the rotation angle is , use the rotation matrix to calculate the coordinates of the four vertices of the work area after rotation: ； ； in, are the vertex coordinates before rotation, are the coordinates of the vertex after rotation; After rotation, take the minimum and maximum values ​​of the X and Y coordinates of the four vertices to determine the boundaries of the working area: ； in, is the minimum boundary of the working area, is the maximum boundary of the working area, is the minimum boundary of the working area, is the maximum boundary of the working area.

8. The method according to claim 5, characterized in that In S3, the method of allocating tasks to the robot grinding system in real time using a consensus-based greedy algorithm includes: Each robot calculates its fitness to perform a certain task, and determines the probability of task selection based on the fitness; the probability of task selection is: ； in, is the task selection probability, is the sum of the fitness scores of robot i for all tasks, is the fitness or score of robot i in completing the task; When entering a consensus area, the robot needs to ensure that only one robot is responsible for generating the task list for the current area; Through the locking mechanism, the robot checks and updates the consensus task list to ensure that only one robot can create or update the task list of the current area; In the consensus area, the robot selects tasks according to the greedy strategy and chooses the task closest to itself.

9. The method according to claim 5, characterized in that In S6, the method of re-inspecting the grinding area using a 3D structured light camera includes: Use a 3D structured light camera to re-inspect the grinding area and obtain point cloud data after grinding; According to the point cloud data, calculate the depth after grinding, i.e. the re-inspection depth, and detect the depth and surface smoothness of the grinding area to ensure that the depth error does not exceed 0.05mm; the specific method is: Use the RANSAC algorithm to perform plane fitting on the point cloud data to determine the plane reference of the steel plate And the polished plane reference , the formula is as follows: Randomly select the minimum number of points from the point cloud data to fit the model. For plane fitting, select three points. , , Determine a plane; Calculating the normal vector : ； After unfolding, the normal vector of the plane is : ； ； ； A known point Substitute into the plane equation and calculate the constant term in the plane equation : ； Calculate the number of interior points, substitute all points into the plane equation, and calculate the distance from the interior points to the plane; If the distance from a point to the plane is less than the preset threshold , then the corresponding points are considered to be interior points; The distance formula is as follows: ； in, is the distance from the point to the plane, is a point in the point cloud. Iterate repeatedly until the model with the most inliers is found or the upper limit of the number of iterations is reached. The model with the most inliers is the best fitting plane model. By calculating the rotation matrix, the normal vector of the polishing plane Normal vector rotated to the base plane superior; The rotation axis is the cross product of the two normal vectors: ； in, is the rotation axis. If the cross product result is a zero vector, the two normal vectors are already parallel and no rotation is required. Calculate the angle θ between the two normal vectors: ； Use the Rodrigues rotation formula to calculate the rotation matrix based on the rotation axis and rotation angle : ； in, is the identity matrix, is the antisymmetric matrix of the rotation axis and is defined as follows: ； in, The rotation axis The amount of Calculate the distance between the two planes after aligning them to be parallel : ； in, is the common normal vector of the two planes, and are the constants in the equations of the two planes, ensuring that the error does not exceed 0.05.

Images