Automatic welding method for large rebar mesh based on AI vision

Abstract

An object of the present invention is to provide an automatic welding method for large-scale steel bar meshes based on AI vision. 【Solution means】 It includes a work platform and an industrial computer. The work platform includes a guide rail and a pedestal that moves along the guide rail. A support arm is rotatably provided on the pedestal, a robot is slidably provided on the support arm, a welding torch and a second vision device are provided on the robot, a first vision device is provided on the support arm, and the work platform further includes an operation table provided under the robot. It also includes a three-dimensional template construction step, a steel bar mesh laying step, a first rough positioning step, a precise positioning step, a turning-over step, and a second rough positioning step and a precise positioning step. Compared with the prior art, the present invention completes the rough positioning of the steel bar mesh through the first vision device, then improves the positioning accuracy of the steel bar mesh through the second vision device, meets the requirements of robot welding, has a high welding speed and high efficiency for the steel bar mesh. Especially when welding large-scale steel bar meshes, the welding speed is significantly faster than that of a welding robot using manual teaching.

Description

[Technical Field]

[0001] The present invention relates to the technical field of rebar mesh welding, and in particular to an automatic welding method for large rebar meshes based on AI vision. [Background technology]

[0002] Prefabricated rebar mesh structures are widely used in the construction industry. Currently, prefabricated rebar mesh structures are typically hand-tied or welded at the intersections of the rebar mesh. However, manual welding has limitations, including unstable welding quality, low construction efficiency, high welding costs, and an inability to work under extreme conditions such as high temperatures and pressures. Mechanized automatic welding technology is currently widely used in various fields, including automotive, aerospace, and industrial manufacturing. Meeting the welding process requirements typically requires high-precision positioning of the workpiece, combined with multiple path teaching operations for the robotic welding torch. However, due to the bending of the steel bars themselves and the inherent elasticity of the bending bars, errors occur in the bent steel bars. This inevitably leads to irregularities in the joined steel bar mesh, uneven gaps between the steel bars, and unevenness between the steel bars. These factors result in arc start failure, arc interruption, collisions, and ultimately failure of the automated welding. Summary of the Invention [Problem to be solved by the invention]

[0003] The present invention provides an automatic welding method for large rebar meshes based on AI vision to solve the problems of irregularities and large gaps in large rebar meshes, which make automatic welding difficult. [Means for solving the problem]

[0004] The present invention provides an automatic welding method for large-scale rebar meshes based on AI vision, comprising: a work platform; and an industrial computer; the work platform comprises a guide rail and a base that moves along the guide rail; a support arm is rotatably mounted on the base; a robot is slidably mounted on the support arm; a welding torch and a second visual device are mounted on the robot; a first visual device is mounted on the support arm; the work platform further comprises an operation table mounted under the robot; and the method also includes the following steps: 3D template construction step: constructing a 3D model template of the working platform, the 3D model template including the coordinate positions of the robot, the coordinate positions of the operation table, the coordinate positions of the guide rail, the coordinate positions of the base, the coordinate positions of the support arm and the coordinate positions of the first visual device, a first rebar mesh model and a second rebar mesh model pre-constructed on the operation table, the second rebar mesh model is a 180-degree flip of the first rebar mesh model, and the first rebar mesh model and the second rebar mesh model both include coordinate positions of a plurality of joint sections, the industrial computer respectively rearranges the joint sections of the first rebar mesh model and the second rebar mesh model, the joint sections have weld seams, the length of the weld seams is shorter than the length of the joint sections, and the joint sections and the weld seams correspond one-to-one; Rebar mesh laying step: Lay multiple steel bars on the operating table to form a rebar mesh, and secure adjacent steel bars with clips on the operating table. a first rough positioning step: the base moves along the guide rail to move the support arm from one end of the rebar mesh to the other end; while the support arm is moving, the first vision device takes continuous images of the rebar mesh to generate a first real 3D model; and compares the joint section of the first rebar mesh model with the joint section of the first real 3D model to obtain an offset amount of the joint section of the first rebar mesh model in the first real 3D model; Precision positioning step: according to the offset amount and the 3D model template, the robot moves to the first joint section, the second vision device collects images of the first joint section to obtain a 3D model of the first joint section, the industrial computer obtains the position coordinates of the weld seam A in the first joint section through the 3D model of the first joint section, and then adjusts the starting position of the weld seam A through the preset gap value, the welding torch starts an arc on one steel bar at the starting point of the weld seam A, and then moves to the middle position D between the two steel bars while swinging, and completes the welding of the first joint section from D as the starting point along the extension direction of the steel bars while swinging; By repeating the precision positioning steps, the robot completes the welding of the remaining joint sections. Flip step: Loosen the clips and flip the rebar mesh 180 degrees so that the back side of the rebar mesh is facing up. a second rough positioning step: the base moves along the guide rail to move the support arm from one end of the rebar mesh to the other end; while the support arm is moving, the first vision device takes continuous images of the rebar mesh to generate a second real 3D model; and compares the joint section of the second rebar mesh model with the joint section of the second real 3D model to obtain an offset amount of the joint section of the second rebar mesh model in the second real 3D model; The welding of the back side of the rebar mesh is completed through a precision positioning step.

[0005] Preferably, the specific steps of adjusting the start position of the weld seam A through the preset gap value are as follows: if the gap at the start of the weld seam A is smaller than the preset value, the start of the weld seam A becomes the current position; if the gap at the start of the weld seam A is larger than the preset value, the weld seam A moves L mm in the narrower direction from the start position; if the gap at the start of the weld seam A after the movement is smaller than the preset value, the start of the weld seam A becomes the current position; if it is larger than the preset value, the above process is repeated until the gap at the start of the weld seam A becomes smaller than the preset value.

[0006] Preferably, in the step of starting the arc on one steel bar at the start of the weld seam A, the welding torch obtains the gap Wmm at the start of the weld seam A through the 3D model of the first joint section, and offsets the welding torch at the middle position of the start of the weld seam A by half Wmm, and then further offsets it by Hmm, after which the welding wire on the welding torch comes into contact with one steel bar, where H represents the depth of the thread on the steel bar.

[0007] Preferably, in the precision positioning process, the welding torch adjusts its posture according to the coordinate positions of the two steel bars in the 3D model of the first joint section so that the welding torch is always perpendicular to the plane on which the two steel bars are located.

[0008] Preferably, in the precision positioning process, when the gap of the weld seam is small, the moving speed of the welding torch is fast and the feeding speed of the wire is slow; when the gap of the weld seam is large, the moving speed of the welding torch is slow and the feeding speed of the wire is fast.

[0009] Preferably, in the precision positioning step, the industrial computer calculates the spatial position information of the joint section of the first rebar mesh model based on the offset amount and the 3D model template, and then the robot moves to the first joint section based on the spatial position information of the joint section of the first rebar mesh model; or the industrial computer calculates the spatial position information of the joint section of the second rebar mesh model based on the offset amount and the 3D model template, and then the robot moves to the first joint section based on the spatial position information of the joint section of the second rebar mesh model.

[0010] Preferably, there are two operation tables, which are located on either side of the guide rail, and when the robot is welding the reinforcing steel mesh on one operation table, it is laying or flipping the reinforcing steel mesh on the other operation table.

[0011] Preferably, in the first coarse positioning step, while the support arm is moving, the first vision device collects images of the rebar mesh and also collects images of the operation table, and processes the images collected by the first vision device to obtain a first real 3D model on the operation table model.

[0012] Preferably, the reinforcing steel mesh comprises a plurality of second steel bars and a plurality of straight first steel bars, and one or both ends of the second steel bars are curved.

[0013] Preferably, in the turning-over step, the support arm is rotated so that the support arm does not remain on the operation table, and the reinforcing steel mesh is turned over 180 degrees through the hanging fixture. [Effects of the Invention]

[0014] Compared with the prior art, in this invention, the first visual device takes continuous photographs along the rebar mesh from start to finish to obtain the coordinate position of the rebar mesh. Although the accuracy of the rebar mesh is low, the speed of obtaining the rebar mesh position is high and the efficiency is high. Then, using the obtained coordinate positions of the rebar mesh, the device moves to the vicinity of each joint section and takes photographs with the second visual device to obtain a 3D model of each joint section. Using the 3D model of the joint section, the start position of the weld seam is adjusted, the welding torch is offset so that the welding wire of the welding torch contacts the steel bars, and two undulation points are set to complete the welding of the joint section. During the welding process, the welding torch posture is adjusted so that the welding torch is always perpendicular to the plane of the two steel bars. This ensures a smooth arc start, prevents arc interruption during the welding process, ensures the stability of the welding wire extension length parameters, ensures the effectiveness of the welding molten pool, and ensures the smooth continuation of the welding process.

[0015] In order to clearly describe the embodiments of the present invention or the technical means in the prior art, the accompanying drawings that need to be used to depict the embodiments or the prior art will be briefly described below. The accompanying drawings described below are only some embodiments of the present invention, and those skilled in the art can obtain other accompanying drawings based on these accompanying drawings without any creative activity. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 1 is a top view of the present invention. [Figure 2] FIG. 1 is a front view of the present invention. [Figure 3] FIG. 2 is a left side view of the present invention. [Figure 4] FIG. 10 is a schematic diagram illustrating a state in which the attitude of the welding torch of the present invention is adjusted. [Figure 5] FIG. 2 is a schematic diagram illustrating the adjustment of the welding seam start point according to the present invention. [Figure 6] 2 is a schematic diagram of the movement path of the welding torch of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0017] In order to clarify the purpose, technical means and advantages of the present invention, the technical means of the present invention will be described clearly and completely below with reference to the accompanying drawings. It goes without saying that the described embodiments are only some of the embodiments of the present invention, and do not represent all of the embodiments. All other implementations obtained by those skilled in the art based on the embodiments of the present invention without any creative activity fall within the scope of protection of the present invention.

[0018] 1 to 3, this embodiment provides an automatic welding method for large rebar mesh based on AI vision, and includes a work platform 100 and an industrial computer. The work platform 100 includes a guide rail 1 and a base 2 that moves along the guide rail 1. A support arm 3 is rotatably mounted on the base 2. A robot 4 is slidably mounted on the support arm 3. A welding torch and a second visual device are mounted on the robot 4. A first visual device is mounted on the support arm 3. The work platform 100 further includes an operation table 5 mounted below the robot 4. A plurality of clips 6 are mounted on the operation table 5. The method also includes the following steps:

[0019] 3D template construction step: construct a 3D model template of the working platform 100, the 3D model template includes the coordinate position of the robot 4, the coordinate position of the operating table 5, the coordinate position of the guide rail 1, the coordinate position of the base 2, the coordinate position of the support arm 3 and the coordinate position of the first visual device, the first rebar mesh model and the second rebar mesh model pre-constructed on the operating table 5, the second rebar mesh model is a 180-degree flip of the first rebar mesh model, and the first rebar mesh model and the second rebar mesh model Each of the reinforcing bar mesh 7 models includes the coordinate positions of multiple joint sections 200. The industrial computer rearranges the joint sections 200 of the first reinforcing bar mesh model and the second reinforcing bar mesh model, respectively. The joint sections 200 have weld seams 300, and the lengths of the weld seams 300 are shorter than the lengths of the joint sections 200. The joint sections 200 and the weld seams 300 correspond one-to-one. Each joint section 200 includes a weld seam 300. Positioning a joint section 200 is equivalent to positioning a weld seam 300. Because the reinforcing bar mesh 7 needs to be welded on one side (front side) and then turned over and welded on the other side (back side), two reinforcing bar mesh 7 models need to be constructed on the operation table 5. The joint section 200 is a closed section where two steel bars 71 are joined within the reinforcing bar mesh 7. The two steel bars 71 within the joint section 200 have some contacting and some not-contacting areas, but the gap is relatively small (as shown in the figure).

[0020] Reinforcement mesh 7 laying step: A plurality of steel bars 71 are laid on the operation table 5 to form the reinforcement mesh 7, and adjacent steel bars 71 are fixed with clips 6 on the operation table 5. The reinforcement mesh 7 needs to be joined together with the steel bars 71, and many errors occur during the joining process, such as the straightness of the steel bars 71 themselves and the elastic modulus characteristics of the steel bars 71. These errors result in differences between the joined reinforcement mesh 7 and the reinforcement mesh 7 in the 3D model template, and these differences can affect the automatic welding of the reinforcement mesh 7 by causing arc start failures, arc interruptions during the welding process, changes in the wire extension length, etc.

[0021] First rough positioning step: the base 2 moves along the starting point of the guide rail 1 to the end point of the guide rail 1, and moves the support arm 3 from one end of the rebar mesh 7 to the other end. While the support arm 3 is moving, the first vision device takes continuous images of the rebar mesh 7 to generate a first real 3D model. During this process, the height position of the support arm 3, i.e., the height position of the first vision device, is determined. The first vision device collects images of the rebar mesh 7 and also collects images of the operation table 5. The images collected by the first vision device are processed to obtain a first real 3D model on the operation table 5 model. The first real 3D model is a rebar mesh. The first reinforcing steel mesh model includes a model of the reinforcing steel mesh 7 and a model of the operation table 5, and the operation table 5 remains constant, thereby determining the coordinate position of the operation table 5. When comparing the joint section 200 of the first reinforcing steel mesh model with the joint section 200 of the first actual 3D model, the operation table 5 is used as a common reference point to overlap the operation table 5 of the 3D model template with the operation table 5 of the first actual 3D model, and then an offset amount of the joint section 200 of the first actual 3D model in the first reinforcing steel mesh model is obtained, and then the coordinate position of the joint section 200 of the reinforcing steel mesh 7 is obtained through the offset amount. In this step, the coordinate position of the joint section 200 of the reinforcing steel mesh 7 (actual object) is obtained by moving the first visual device to collect and process images, which is fast and efficient. However, in order to capture the entire reinforcing steel mesh 7 in the width direction, the first visual device is installed high, which results in insufficient photographic accuracy, and as a result, there is a certain error in the coordinate position of the reinforcing steel mesh joint section 200 obtained, making it difficult for the accuracy to meet the welding needs of the robot 4.

[0022] 6 , the welding torch 8 starts an arc at point C on one steel bar 71 at the start of the welding seam 300A, and then swings to a midpoint D between the two steel bars 71. Then, starting from D, it swings to point E along the extension direction of the steel bars 71 to complete the welding of the first joint section 200. In this step, the industrial computer obtains the position coordinates of the weld seam 300A in the first joint section 200. By obtaining the position coordinates of the weld seam 300A, the difference between the actual rebar mesh 7 and the rebar mesh 7 in the 3D model template is eliminated. However, if the gap at the start of the weld seam 300A is too large, arc start and welding may not be possible. Therefore, the start position of the weld seam 300A is adjusted via a preset gap value so that the gap at the start of the weld seam 300A meets the arc start requirements of the welding torch 8, preventing arc start and welding failure and ensuring the smooth welding process. Next, the welding torch 8 starts an arc on one steel bar 71 at the start of the weld seam 300A, which is the first arc start point. During the first arc start, the welding wire 9 of the welding torch 8 comes into contact with the steel bar 71 (ensuring a smooth arc start), and then moves to the midpoint D between the two steel bars 71 during oscillation, D being the second arc start point. From D as the starting point, the welding wire 9 moves along the extension direction of the steel bar 71 during oscillation to complete the welding of the first joint section 200, and the welded metal strip from the first arc start point to D ensures that the welding wire 9 of the welding torch 8 comes into contact with the metal strip during the second arc start (ensuring a smooth arc start).By providing two arc starting points, electric welding allows for 100% arc starting, ensuring the welding process proceeds smoothly.

[0023] By repeating the precision positioning step, the robot 4 completes welding of the remaining joint sections 200. For example, the robot 4 moves to the second joint section 200 according to the offset amount and the 3D model template, the second vision device collects images of the second joint section 200 to obtain a 3D model of the second joint section 200, the industrial computer obtains the position coordinates of the weld seam 300B in the second joint section 200 through the 3D model of the second joint section 200, and then obtains the position of the weld seam 300B by adjusting the starting position of the weld seam 300B according to a preset gap value. The welding torch 8 starts an arc on one steel bar 71 at the starting point of the weld seam 300B, then moves to position D between the two steel bars 71 while swinging, and then moves from D as a starting point along the extension direction of the steel bars 71 while swinging to complete welding of the second joint section 200.

[0024] Flip step: Loosen clip 6 and flip the rebar mesh 7 180 degrees so that the back side of the rebar mesh 7 faces up. Second rough positioning step: the base 2 moves along the guide rail 1 to move the support arm 3 from one end of the reinforcing steel mesh 7 to the other end. While the support arm 3 is moving, the first vision device takes continuous images of the reinforcing steel mesh 7 to generate a second real 3D model. The joint section 200 of the second reinforcing steel mesh model is compared with the joint section 200 of the second real 3D model to obtain an offset amount of the joint section 200 of the second real 3D model in the second reinforcing steel mesh model. The welding of the back side of the reinforcing steel mesh 7 is completed through a precision positioning step.

[0025] The specific steps for adjusting the start position of the weld seam 300A using the preset gap value are as follows: If the gap at the start of the weld seam 300A is smaller than the preset value, for example, if the gap at the start of the weld seam 300A is less than 10 mm, there is no need to offset the weld seam 300A. Referring to FIG. 5, if the gap at the start O of the weld seam 300A is greater than 10 mm, the weld seam 300A is shifted 2 mm in the narrower direction to P (if the weld seam 300A extends from the start O to both ends, one end must be narrow and the other end must be wide; otherwise, the rebar mesh 7 must be re-fixed). After the shift, it is determined whether the gap at the start P of the weld seam 300A is greater than 10 mm. If it is less than 10 mm, the start of the weld seam 300A is P. If it is greater than 10 mm, the above process is repeated until the gap at the start of the weld seam 300A is less than 10 mm.

[0026] In the step of starting the arc with the welding torch 8 on one steel bar at the start of the weld seam 300A, the gap W mm at the start of the weld seam 300A is obtained through the 3D model of the first joint section 200. For example, if the gap at the start of the weld seam 300A is 10 mm, the welding torch 8 is offset 5 mm from the midpoint of the start of the weld seam 300A, and after a further offset of H mm, the welding wire 9 on the welding torch 8 will contact one steel bar 71. There are two steel bars 71 within one joint section 200, and the welding wire 9 on the welding torch 8 will usually contact the steel bar 71 inside the rebar mesh 7. The lowest point determined after scanning with the second vision device is the midpoint of the start of the weld seam 300A. However, due to errors in the rebar mesh 7 joining process, starting the arc from the midpoint of the start of the weld seam 300A is likely to fail. To ensure a successful arc start, the conventional arc start at the intermediate position must be offset so that the welding wire 9 can reliably contact the steel bars 71 and have a stable arc starting point. Because the gap value between the steel bars 71 is dynamic, an offset of 5 mm is first performed to eliminate any gap error in the steel bars 71. However, because the steel bars 71 have threads and the thread height H is generally set to 5 mm, an additional 5 mm offset is required to allow the welding wire 9 to contact the steel bars 71 inside the rebar mesh 7.

[0027] In the precision positioning step, the industrial computer calculates the spatial position information of the joint section 200 of the first rebar mesh model based on the offset amount and the 3D model template, and then the robot 4 moves to the first joint section 200 based on the spatial position information of the joint section 200 of the first rebar mesh model, or the industrial computer calculates the spatial position information of the joint section 200 of the second rebar mesh model based on the offset amount and the 3D model template, and then the robot 4 moves to the first joint section 200 based on the spatial position information of the joint section 200 of the second rebar mesh model.

[0028] 4 , in the precision positioning process, the welding torch 8 adjusts its posture according to the coordinate positions of the two steel bars 71 in the 3D model of the first joint section 200 so that it is always perpendicular to the plane on which the two steel bars 71 are located. After the rebar mesh 7 is joined, there will be height variations between the steel bars 71. If the height difference between the two steel bars 71 exceeds a certain range, the welding torch 8 will hit the taller steel bar 71, causing the welding torch 8 to stop welding. The second vision device obtains a 3D model of each joint section 200 through 3D modeling. The 3D model not only provides the accurate trajectory of the weld seam 300 but also the coordinate positions of the steel bars 71. An algorithm is then used to guide the welding gun 8 to adjust the angle offset in the z direction so that the welding torch 8 is always perpendicular to the plane on which the two steel bars 71 are located. This prevents collisions and ensures that the distances between the end of the welding torch 8 and the two steel bars 71 are consistent, ensuring the stability of the welding wire extension length parameters and the effectiveness of the weld pool.

[0029] In the precision positioning process, the movement speed of the welding torch 8 and the wire feed speed are adjusted according to the size of the gap in the weld seam 300. When the gap in the weld seam 300 is small, the movement speed of the welding torch 8 is fast and the wire feed speed is slow; when the gap in the weld seam 300 is large, the movement speed of the welding torch 8 is slow and the wire feed speed is fast, ensuring that no arc interruption occurs and that the fill amount is consistent. The wire feed speed can be adjusted by the magnitude of the current; the higher the current, the faster the wire feed speed, and the lower the current, the slower the wire feed speed.

[0030] The amplitude of the welding torch 8 is set to a fixed value rather than automatically adjusted according to the width of the weld seam 300. This setting ensures the stability of the welding wire extension length parameter and also guarantees the effectiveness of the welding molten pool. The absence of arc interruption during welding and the guarantee of the fill amount are adjusted by adjusting the movement speed of the welding torch 8 and the wire feed speed.

[0031] In another embodiment of the present invention, there are two operation tables 5, each located on either side of the guide rail 1. When the robot 4 is welding the rebar mesh 7 on one operation table 5, the support arm 3 rotates above the operation table 5, and the other operation table 5 lays or turns over the rebar mesh 7, without interference between the support arm 39 and the robot 4, allowing these processes to proceed smoothly. Rotating the support arm 3 helps to improve the welding efficiency of the robot 4 and makes it easier to lay and turn over the rebar mesh 7.

[0032] The reinforcing bar mesh 7 comprises a plurality of straight first steel bars 71 and a plurality of second steel bars 71, with one or both ends of the second steel bars 71 being curved. Due to the characteristics of the elastic modulus of the steel bars 71, errors always occur after the production of the second steel bars 71, and the laid reinforcing bar mesh 7 does not match the first reinforcing bar mesh model.

[0033] In the flipping step, the support arm 3 is rotated so that it does not remain on the operation table 5, and the reinforcing steel mesh 7 is flipped 180 degrees through the lifting device so that the back side of the reinforcing steel mesh 7 faces upward, and the steel bar 71 can naturally be carried onto the operation table 5 using the lifting device.

[0034] In the flipping step, one side of the mounting base is joined to the operation table 5, and then the reinforcing bar mesh 7 is flipped onto the mounting base using a lifting device, and then the mounting base is moved onto the operation table 5. Specifically, a plurality of positioning square pipes are fixed on the operation table 5, the mounting base is equipped with a plurality of crossbars (the crossbars are also square pipes), one end of the crossbars is inserted into the positioning square pipes, and the lifting device flips the reinforcing bar mesh 7 onto the mounting base, and the crossbars move forward along the positioning square pipes until they straddle the operation table 5, at which point the reinforcing bar mesh 7 is placed on the operation table 5 via the crossbars, and the positioning square pipes are chamfered to make it easier for the reinforcing bar mesh 7 to pass through them, and the height of the crossbars is greater than that of the clips 6 so that the clips 6 do not affect the reinforcing bar mesh 7.

[0035] In the present invention, the first visual device takes continuous photographs along the rebar mesh 7 from start to finish to obtain the coordinate position of the rebar mesh 7. Although the accuracy of the rebar mesh 7 is low, the speed and efficiency of obtaining the position of the rebar mesh 7 are high. Then, using the obtained coordinate positions of the rebar mesh 7, the device moves to the vicinity of each joint section 200, takes photographs with the second visual device to obtain a 3D model of each joint section 200. Using the 3D model of the joint section 200, the start position of the weld seam 300 is adjusted, the welding torch 8 is offset so that the welding wire 9 of the welding torch 8 contacts the steel bars 71, and the posture of the welding torch 8 is adjusted so that the welding torch 8 is always perpendicular to the plane on which the two steel bars 71 are located. This ensures a smooth arc start, prevents arc interruption during the welding process, ensures the stability of the welding wire extension length parameters, ensures the effectiveness of the welding molten pool, and ensures the smooth continuation of the welding process.

[0036] Finally, it should be noted that the above embodiments are merely intended to clarify the technical means of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art will understand that the technical means described in each of the above embodiments can be modified or some of the technical features can be replaced with equivalents. Such modifications or replacements do not deviate from the essence of the corresponding technical means from the spirit and scope of the technical means of each embodiment of the present invention. [Explanation of symbols]

[0037] 100 working platform 1 guide rail 2 pedestal 3 Support Arms 4. Robot 5 Operation Table 6 clips 7 Rebar mesh

Claims

1. An automatic welding method for large rebar meshes based on AI vision, comprising: a work platform; and an industrial computer; the work platform comprises a guide rail and a base that moves along the guide rail; a support arm is rotatably mounted on the base; a robot is slidably mounted on the support arm; a welding torch and a second visual device are mounted on the robot; a first visual device is mounted on the support arm; and the work platform further comprises an operation table mounted below the robot; a 3D template construction step: constructing a 3D model template of the working platform, the 3D model template including the coordinate positions of the robot, the coordinate positions of the operation table, the coordinate positions of the guide rail, the coordinate positions of the base, the coordinate positions of the support arm, and the coordinate positions of the first visual device, a first reinforcing bar mesh model and a second reinforcing bar mesh model pre-constructed on the operation table, the second reinforcing bar mesh model being a 180-degree flip of the first reinforcing bar mesh model, and the first reinforcing bar mesh model and the second reinforcing bar mesh model both including coordinate positions of a plurality of joint sections; the industrial computer respectively rearranges the joint sections of the first reinforcing bar mesh model and the second reinforcing bar mesh model, the joint sections having weld seams, the length of the weld seams being shorter than the length of the joint sections, and the joint sections and the weld seams having a one-to-one correspondence; A reinforcing bar mesh laying step: laying a plurality of steel bars on the operation table to form the reinforcing bar mesh, and fixing the adjacent steel bars with clips on the operation table; a first rough positioning step: the base moves along the guide rail to move the support arm from one end of the reinforcing bar mesh to the other end, and while the support arm is moving, the first vision device takes continuous images of the reinforcing bar mesh to generate a first real 3D model, and compares the joint section of the first reinforcing bar mesh model with the joint section of the first real 3D model to obtain an offset amount of the joint section of the first reinforcing bar mesh model in the first real 3D model; precision positioning step: based on the offset amount and the 3D model template, the robot moves to the first joint section, the second vision device collects images of the first joint section to obtain a 3D model of the first joint section, the industrial computer obtains the position coordinates of the weld seam A in the first joint section through the 3D model of the first joint section, and then adjusts the start position of the weld seam A through a preset gap value, the welding torch starts an arc on one of the steel bars at the start of the weld seam A, and then moves to a middle position D between the two steel bars while swinging, and completes the welding of the first joint section from D as the starting point along the extension direction of the steel bars while swinging; repeating the fine positioning step, the robot completes welding of the remaining joint sections; Flip step: loosen the clips and flip the rebar mesh 180 degrees so that the back side of the rebar mesh faces up; a second rough positioning step: the base moves along the guide rail to move the support arm from one end of the reinforcing bar mesh to the other end, and while the support arm is moving, the first vision device takes continuous images of the reinforcing bar mesh to generate a second actual 3D model, and compares the joint section of the second reinforcing bar mesh model with the joint section of the second actual 3D model to obtain an offset amount of the joint section of the second reinforcing bar mesh model in the second actual 3D model; The welding of the back side of the reinforcing steel mesh is completed through the precision positioning step. A method for automatic welding of large rebar meshes based on AI vision, characterized by including the above steps.

2. 2. The automatic welding method for large rebar meshes based on AI vision according to claim 1, wherein the specific step of adjusting the start position of the weld seam A through the preset gap value is as follows: if the gap at the start point of the weld seam A is smaller than a predetermined value, the start point of the weld seam A becomes a current position; if the gap at the start point of the weld seam A is larger than the predetermined value, the weld seam A moves L mm in a narrower direction from the start position; if the gap at the start point of the weld seam A after the movement is smaller than the predetermined value, the start point of the weld seam A becomes a current position; if it is larger than the predetermined value, the above process is repeated until the gap at the start point of the weld seam A becomes smaller than the predetermined value.

3. 2. The automatic welding method for large rebar meshes based on AI vision according to claim 1, wherein in the step of starting the arc on one of the steel bars at the start of the weld seam A, the gap W mm at the start of the weld seam A is obtained through the 3D model of the first joint section, and the welding torch at the middle position of the start of the weld seam A is offset by half W mm, and then further offset by H mm, after which the welding wire on the welding torch comes into contact with one of the steel bars, where H represents the height of the thread of the steel bar.

4. 2. The method for automatic welding of large rebar meshes based on AI vision as claimed in claim 1, characterized in that in the precision positioning process, the welding torch adjusts its posture according to the coordinate positions of the two steel bars in the 3D model of the first joint section so that the two steel bars are always perpendicular to a plane.

5. The AI vision-based automatic welding method for large rebar meshes according to claim 4, characterized in that in the precision positioning process, when the gap of the weld seam is small, the moving speed of the welding torch is fast and the feeding speed of the wire is slow; and when the gap of the weld seam is large, the moving speed of the welding torch is slow and the feeding speed of the wire is fast.

6. 2. The AI vision-based automatic welding method for large rebar meshes according to claim 1, wherein in the precision positioning step, the industrial computer calculates spatial position information of the joint section of the first rebar mesh model based on the offset amount and the 3D model template, and then the robot moves to the first joint section based on the spatial position information of the joint section of the first rebar mesh model; or the industrial computer calculates spatial position information of the joint section of the second rebar mesh model based on the offset amount and the 3D model template, and then the robot moves to the first joint section based on the spatial position information of the joint section of the second rebar mesh model.

7. The automatic welding method for large reinforcing steel meshes based on AI vision according to claim 1, characterized in that there are two operation tables, the two operation tables are located on both sides of the guide rail, and when the robot is welding the reinforcing steel meshes on one operation table, the other operation table lays or flips over the reinforcing steel meshes.

8. 2. The method for automatic welding of large rebar meshes based on AI vision according to claim 1, characterized in that in the first rough positioning step, while the support arm is moving, the first vision device collects images of the rebar mesh and also collects images of the operation table, and processes the images collected by the first vision device to obtain the first real 3D model on the model of the operation table.

9. The automatic welding method for large reinforcing steel mesh based on AI vision described in claim 1, characterized in that the reinforcing steel mesh comprises a plurality of second steel bars and a plurality of straight first steel bars, and one or both ends of the second steel bars are curved.

10. The automatic welding method for large reinforcing steel meshes based on AI vision as described in claim 7, characterized in that in the flipping step, the support arm is rotated so that the support arm does not remain on the operation table, and the reinforcing steel mesh is flipped 180 degrees through a hanging device.

Images