A double-sensing integrated weld tracking sensor and correction method

Abstract

This invention provides a dual-sensor integrated weld seam tracking sensor method that simultaneously employs both structured light and monocular vision principles to detect the weld seam position. The structured light sensor obtains accurate weld seam position information, while the monocular vision sensor obtains the weld seam's deflection angle, thereby enabling real-time correction of lead-ahead errors. Furthermore, the two sensors share a single lens and camera, which not only significantly simplifies system composition, reduces costs, minimizes sensor size, and facilitates application integration, but also improves the synchronization and consistency of the two sensors' detection, enhances anti-interference capabilities, and significantly improves practical performance.

Description

Technical Field

[0001] This invention relates to a weld seam tracking sensor. Specifically, it relates to a high-performance dual-sensor integrated weld seam tracking sensor and a correction method. Background Technology

[0002] With the popularization and affordability of factory automation equipment, welding operations have evolved from traditional manual welding to semi-automatic and automatic welding. Robotic welding processes are also becoming widespread in various fields, primarily the automotive industry, and have become a necessary means of streamlining welding procedures. To achieve accurate and deviation-free robotic welding, consistent welding quality, high efficiency, and adaptability to different welding processes and workpiece structures, it is necessary to monitor the center position of the welding torch, arc, or molten pool in real time during the welding process. This requires ensuring the sensor aligns with the joint center position and can adapt to changes in workpiece position and deformations generated during welding. This is the task of a weld seam tracking sensor.

[0003] Currently, weld seam tracking methods fall into two categories: one is the traditional contact method, which uses a probe or a probe to directly contact the seam and judges the deviation through feedback for adjustment. Due to contact wear, long cycle time, and low efficiency, it is gradually being replaced by non-contact methods. The other category is the non-contact method using optical methods or eddy current emission, in which optical primary mirrors have become the mainstream, specifically including structured light methods and vision methods.

[0004] Structured light method is based on the principle of triangulation to obtain three-dimensional information of the weld. Structured light is projected onto the weld at a certain distance in front of the molten pool, and a camera captures the laser stripe image of the weld bevel. After image processing, the characteristic parameters of the weld are obtained, and combined with the camera imaging parameters, the center position of the weld can be determined. Structured light method, with its advantages of high accuracy, good sensitivity, rich information content, and wide applicability, occupies an increasingly large proportion among various types of weld tracking sensors and is widely used in various welding processes. However, this line structured light detection method also has its own defects and shortcomings: First, the detection position of existing structured light vision sensors is a certain distance from the actual welding position. When there is an angle between the travel track and the weld trajectory, or the track surface is uneven, it can easily cause lead-ahead errors in the welding torch. Second, for butt welds with very small gaps and tightly butt welds without bevels, this line structured light has almost no deformation, making this method almost completely ineffective. To eliminate this type of error, some researchers have reduced the distance between the structured light and the molten pool. However, this increases arc interference, complicates post-processing, and results in poor stability and unsatisfactory performance. Other researchers have used high dynamic range cameras to directly image the molten pool and the weld seam in front of it, but this is costly, and in high-speed welding, the weld seam area that can be captured is relatively short, making it difficult to calculate the deviation in time and guide the welding torch movement. For example, Nanchang University proposed a dual-line structured light sensing method. By detecting two different positions of the weld seam, the weld seam trajectory is obtained, and the weld seam deviation at the welding position is calculated, thereby reducing the lead-ahead error. However, this method has a complex system composition, high cost, and large size, making it impractical. In particular, the distance between the two line lasers is relatively short, resulting in excessive deviation and unsatisfactory practical results.

[0005] The principle of visual methods is to acquire images of the weld area and extract the weld center feature information through processing. Visual sensors acquire a large amount of information, especially for butt welds with very small gaps and tight butt welds without beveling, which have unique advantages. For example, Guangdong University of Technology uses a dual-target imaging device to image the weld and the molten pool, performs image processing on the weld and molten pool images to obtain the coordinates of the weld center and the molten pool center, and uses the molten pool center coordinates to correct the welding torch deviation. However, visual methods are easily affected by welding spatter, fumes, steam, arc flash, and rust, scratches, marks, and oxide scale on the workpiece. Moreover, visual sensor technology is expensive and difficult to apply to ordinary welding production. In addition, since the molten pool itself is not violently pulsating, and the weld at the molten pool is completely melted, the weld information is basically submerged, making it difficult to accurately obtain the weld position information and features from the molten pool image.

[0006] Therefore, existing single-technology methods, whether structured light or vision-based, cannot fully meet the requirements for weld seam tracking. The only feasible technical approach is to employ multiple principles and sensors to acquire more multi-dimensional information. For example, Guangdong Guangye University proposed a "binocular vision weld seam tracking method and system," which uses a structured light camera to capture images of the weld seam area and a molten pool camera to capture images of the molten pool. After processing, the weld seam position is obtained, and the welding torch or laser head is corrected. Shanghai Jiao Tong University proposed a "robot weld seam tracking and molten pool monitoring sensor based on active and passive vision," which uses two cameras, a CCD and a CMOS, and a linear laser, combining the characteristics of active and passive vision. Utilizing binocular vision, it can simultaneously perform weld seam tracking and molten pool monitoring functions. Xinjiang Weiao Technology Co., Ltd. proposed a "laser weld seam tracking sensor based on dual-camera image composite," which uses two cameras, a primary and an auxiliary camera, to aim at the weld seam area for detection, overcoming the shortcomings of traditional single-line structured light methods that cannot adapt to gapless weld seams. Guangxi Anbote Intelligent Technology Co., Ltd. proposed a "binocular vision teaching and structured light weld seam tracking vision sensor." A line laser and a plate-level camera are arranged sequentially from front to back in the center of the sensor cavity, with two industrial cameras positioned on either side of the line laser. It employs both line structured light and binocular stereo vision principles for detection. Xiangtan University proposed a "tri-vision all-position corrugated plate intelligent weld seam tracking sensor," comprising a binocular bidirectional adaptive scanning array vision sensor, a CCD area array intelligent vision system, an accelerometer, and a gyroscope, enabling all-position weld seam detection.

[0007] However, the aforementioned methods currently employ two or more cameras to acquire images of the weld or molten pool separately, processing these images individually to derive weld position information and perform corrections. The disadvantages are complex system composition, high cost, large space requirements, and inconvenience for integration into welding robots. In particular, synchronization errors inevitably exist between the two or more cameras, leading to measurement errors during high-speed dynamic welding processes and under conditions of strong vibration, spatter, smoke, and other dynamic interference, significantly reducing the actual tracking performance. Summary of the Invention

[0008] This invention addresses the shortcomings and deficiencies of existing optical weld seam tracking sensors by proposing a dual-sensor integrated weld seam tracking sensor method. It simultaneously employs both structured light and monocular vision principles to detect the weld seam position. The structured light sensor obtains accurate weld seam position information, while the monocular vision sensor obtains the weld seam's deflection angle, enabling real-time correction of lead-ahead errors. Furthermore, the two sensors share a single lens and camera, significantly simplifying system composition, reducing costs, minimizing sensor size, and facilitating application integration. The shared lens and camera also improve the synchronization and consistency of the two sensors' detection, enhance anti-interference capabilities, and significantly improve practical performance.

[0009] This invention is achieved through the following technical solution:

[0010] The unique feature of the dual-sensor integrated weld seam tracking sensor of the present invention is that the sensor comprises several parts, including a line laser, a surface light source, a filter, a reflector, a lens, a camera, a housing, a socket, a cable, and a controller, wherein:

[0011] The line laser is an industrial-grade semiconductor line laser, located on one side of the lower end inside the sensor, projecting the line laser vertically downwards. The line laser is perpendicular to the workpiece surface and the weld, and is used to present the vertical cross-sectional profile of the weld. The center wavelength of the line laser should be far away from the spectral peak region of the electric arc.

[0012] The surface light source is an industrial-grade monochrome LED floodlight, whose center wavelength is the same as that of the line laser. It is located at the middle position at the lower end of the sensor and projects monochrome illumination light vertically downward to illuminate the weld area.

[0013] The filter is a narrowband filter, whose center wavelength is the same as the center wavelength of the line laser and the color of the surface light source, and its bandwidth should cover the wavelength range of the line laser and the surface light source; the filter is located at the bottom of the sensor and is used to filter out other interfering light.

[0014] The reflector is a flat aluminum-plated reflector located on the other side of the bottom of the sensor. It is used to project the image of the weld area onto the lens to form a reflected light path, which is used to reduce the lateral size of the sensor. The reflector is arranged vertically, which is beneficial to reducing the lateral size of the sensor.

[0015] The lens is an industrial-grade lens, mounted on the camera and facing the reflector, responsible for imaging the image of the weld area reflected by the reflector onto the image plane of the camera.

[0016] The camera is an industrial-grade area array camera, located at the upper part of the sensor, and is responsible for converting the image of the weld area into a digital signal.

[0017] The outer shell is made of a highly conductive and heat-dissipating metal material, which is used to support and fix all internal components and provide reliable protection and shielding.

[0018] The socket is a high-reliability, long-life industrial-grade socket, internally connected to a line laser, a surface light source, and a camera;

[0019] The cable is a high-reliability shielded twisted-pair cable used to connect the socket to the controller, provide power to the sensor and send control commands, and transmit the image signals obtained by the sensor to the controller.

[0020] The controller is a micro industrial control computer, which is responsible for sending control commands to the sensors, controlling the line laser and the surface light source to alternately turn on and off, and simultaneously acquiring laser stripe images and weld area images; at the same time, it processes the image signals from the sensors to finally obtain the weld position information.

[0021] The working process of the dual-sensor integrated weld seam tracking sensor of the present invention is as follows: Under the unified control of the controller, the line laser is first turned on to emit a line laser beam to the weld seam and form laser stripes on the weld seam surface. The camera acquires the image of the laser stripes and transmits it to the controller for processing to obtain the weld seam position information. Then, the line laser is turned off and the surface light source is turned on to illuminate the weld seam area. The camera acquires the image of the weld seam area and transmits it to the controller for processing to obtain the weld seam deflection angle information. When the weld seam deflects, the aforementioned weld seam position information is corrected to overcome the lead error, thereby achieving accurate weld seam correction.

[0022] This invention proposes a correction method based on the above-mentioned dual-sensor integrated weld seam tracking sensor, as follows:

[0023] (1) Turn on the line laser, emit a line laser towards the weld and form laser stripes on the surface of the weld, and obtain the image of the laser stripes;

[0024] (2) Process the laser stripe image to obtain weld location information;

[0025] (3) Turn off the line laser, turn on the area light source to illuminate the weld area, and obtain an image of the weld area;

[0026] (4) Process the image of the weld area to obtain the deflection angle information of the weld;

[0027] (5) When the weld seam is deviated, the weld seam position information mentioned above is corrected according to the distance between the laser stripe and the welding gun, thereby obtaining the true and accurate weld seam position information.

[0028] (6) The final weld position information is transmitted to the welding robot controller to correct the weld position deviation and finally realize online correction. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the dual-sensor integrated weld seam tracking sensor of the present invention;

[0030] Figure 2 This is a schematic diagram illustrating the principle of the weld seam positioning algorithm of the present invention;

[0031] Figure 3 This is a schematic diagram illustrating the principle of the weld angle measurement algorithm of the present invention;

[0032] Figure 4 This is a schematic diagram illustrating the principle of the weld position compensation and correction algorithm of the present invention;

[0033] In the diagram, 1-line laser, 2-surface light source, 3-filter, 4-reflector, 5-lens, 6-camera, 7-housing, 8-socket, 9-cable, 10-controller, 11-robot, 12-welding torch, 13-workpiece. Detailed Implementation

[0034] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings. These embodiments are implemented based on the technical solutions of the present invention and provide detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments.

[0035] The special feature of the dual-sensor integrated weld seam tracking sensor of the present invention is that the sensor includes several parts such as a line laser 1, a surface light source 2, a filter 3, a reflector 4, a lens 5, a camera 6, a housing 7, a socket 8, a cable 9, and a controller 10. Figure 1 As shown, where:

[0036] The line laser 1 is an industrial-grade semiconductor line laser, located on one side of the lower end inside the sensor, projecting the line laser vertically downwards. The line laser is perpendicular to the surface of the workpiece 13 and the weld, and is used to present the vertical cross-sectional profile of the weld. The center wavelength of the line laser should be far away from the spectral peak region of the electric arc, for example, 650nm red light can be selected.

[0037] The surface light source 2 is an industrial-grade monochrome LED floodlight, whose center wavelength is basically the same as the center wavelength of the line laser 1, for example, 650nm; the surface light source 2 is located in the middle of the lower end of the sensor, and projects monochrome illumination light vertically downward to illuminate the weld area.

[0038] The filter 3 is a narrowband filter with a center wavelength that matches the center wavelength of the line laser 1 and the color of the surface light source 2, for example, 650nm. The bandwidth of the filter 3 should cover the wavelength range of the line laser 1 and the surface light source 2. Assuming the wavelength fluctuation range of the line laser 1 is ±5nm and the wavelength range of the surface light source 2 is ±10nm, the bandwidth of the filter should be ±10nm. The filter 3 is located at the bottom of the sensor and is used to filter out other interfering light.

[0039] The reflector 4 is a flat aluminum-plated reflector located on the other side of the bottom of the sensor. It is used to project the image of the weld area onto the lens 5 to form a reflected light path, which is used to reduce the lateral size of the sensor. The reflector 4 is arranged vertically, which is beneficial to reducing the lateral size of the sensor.

[0040] The lens 5 is an industrial-grade lens, which is mounted on the camera 6 and faces the reflector 4. It is responsible for imaging the image of the weld area reflected by the reflector 4 onto the image plane of the camera 6.

[0041] The camera 6 is an industrial-grade area array camera, located at the upper part inside the sensor, and is responsible for converting the image of the weld area into a digital signal.

[0042] The outer shell 7 is made of a highly conductive and heat-dissipating metal material, such as stainless steel, aluminum alloy, or brass, to support and fix all internal components and provide reliable protection and shielding.

[0043] The socket 8 is a highly reliable, long-life industrial-grade socket, which is internally connected to the line laser 1, the surface light source 2, and the camera 6.

[0044] The cable 9 is a high-reliability shielded twisted-pair cable used to connect the socket 8 to the controller 10, provide power to the sensor and send control commands, and transmit the image signals obtained by the sensor to the controller 10.

[0045] The controller 10 is a micro industrial control computer responsible for sending control commands to the sensors, controlling the line laser 1 and the surface light source 2 to alternately turn on and off, and simultaneously acquiring laser stripe images and weld area images; at the same time, it processes the image signals from the sensors to finally obtain the weld position information.

[0046] The working process of the dual-sensor integrated weld seam tracking sensor of the present invention is as follows: Under the unified control of the controller 10, the line laser 1 is first turned on to emit a line laser to the weld seam and form laser stripes on the weld seam surface. The camera 6 acquires the image of the laser stripes and transmits it to the controller 10 for processing to obtain the weld seam position information. Then, the line laser 1 is turned off and the surface light source 2 is turned on to illuminate the weld seam area. The camera acquires the image of the weld seam area and transmits it to the controller 10 for processing to obtain the weld seam deflection angle information. When the weld seam deflects, the aforementioned weld seam position information is corrected to overcome the lead error, thereby achieving accurate weld seam correction.

[0047] This invention proposes a correction method based on the above-mentioned dual-sensor integrated weld seam tracking sensor, as follows:

[0048] (1) Turn on the line laser 1, emit a line laser to the weld and form laser stripes on the surface of the weld, and obtain the image of the laser stripes;

[0049] (2) Process the laser stripe image to obtain weld location information; the specific positioning algorithm principle is as follows: Figure 2 As shown, the x-axis is the direction perpendicular to the weld in the plane of workpiece 13 surface, and the z-axis is the height direction perpendicular to the weld section; by extracting the center line position of the weld break contour, the position of the axis of symmetry is calculated, thereby obtaining the weld positioning result Δ;

[0050] (3) Turn off the line laser 1, turn on the area light source 2 to illuminate the weld area, and obtain an image of the weld area;

[0051] (4) Process the image of the weld area to obtain the weld deflection angle information; the specific angle measurement algorithm principle is as follows: Figure 3 As shown, the y-axis represents the direction of the weld seam within the plane of workpiece 13 (i.e., the direction of welding movement); by fitting a straight line to the local weld seam image, the deflection angle result θ can be directly obtained;

[0052] (5) When the weld seam deviates at an angle (θ≠0), the weld seam position information is corrected based on the distance (y direction) between the laser stripe (and line laser 1) and the welding torch 12, thereby obtaining accurate weld seam position information; the principle of the position compensation correction algorithm is as follows: Figure 4 As shown, assuming the distance between the laser stripe and the welding torch 12 is d, the leading error caused by this deflection angle θ is: C = d * tan θ; based on this error value C, the aforementioned weld positioning result Δ can be corrected, and the corrected weld positioning result is: Δ′= Δ– C;

[0053] (6) The final weld position information Δ′ is transmitted to the controller of the welding robot 11 to correct the weld position deviation and finally realize online correction.

[0054] Compared with the prior art, the beneficial effects of the dual-sensor integrated weld seam tracking sensor and correction method of the present invention are:

[0055] (1) The dual-sensor integrated weld seam tracking sensor of the present invention uses both line structured light and monocular vision principles for weld seam detection and tracking, giving full play to the advantages of the two methods and making up for their shortcomings, thus playing the role and effect of maximizing strengths and minimizing weaknesses.

[0056] (2) The present invention uses a monocular vision method to obtain the deflection angle information of the weld in real time, and compensates and corrects the resulting lead error accordingly, effectively overcoming the defects and shortcomings of the traditional line structured light method, and the effect is very significant.

[0057] (3) The two measurement methods of the dual-sensor integrated weld seam tracking sensor of the present invention can share a set of lens and camera, which can not only greatly simplify the system composition, reduce costs, reduce sensor size, and facilitate application integration;

[0058] (4) The two measurement methods of the dual-sensor integrated weld seam tracking sensor of the present invention share the same set of shared lens and camera, which is also conducive to improving the synchronization and consistency of the two sensors, stronger anti-interference ability, and significantly improved practical effect.

[0059] Therefore, it can be seen that the technical solution of the present invention has significant advantages and substantial progress compared with the traditional single-line structured light method and single vision method for weld seam tracking sensors.

Claims

1. A dual-sensor integrated weld seam tracking sensor, characterized in that: The sensor comprises several parts, including a line laser, a surface light source, a filter, a reflector, a lens, a camera, a housing, a socket, a cable, and a controller. The line laser is an industrial-grade semiconductor line laser, located on one side of the lower end inside the sensor, projecting the line laser vertically downwards. The line laser is perpendicular to the workpiece surface and the weld, and is used to present the vertical cross-sectional profile of the weld. The center wavelength of the line laser should be far away from the spectral peak region of the electric arc; The surface light source is an industrial-grade monochrome LED floodlight, whose center wavelength is basically the same as that of a line laser. It is located in the middle of the lower end of the sensor and projects monochrome illumination light vertically downward to illuminate the weld area. The filter is a narrowband filter, whose center wavelength is the same as the center wavelength of the line laser and the center wavelength of the surface light source, and its bandwidth should cover the wavelength range of the line laser and the surface light source; the filter is located at the bottom of the sensor and is used to filter out other interfering light. The reflector is a flat aluminum-plated reflector located on the other side of the bottom of the sensor. It is used to project the image of the weld area onto the lens to form a reflected light path, which is used to reduce the lateral size of the sensor. The reflector is arranged vertically, which is beneficial to reducing the lateral size of the sensor. The lens is an industrial-grade lens, mounted on the camera and facing the reflector, responsible for imaging the image of the weld area reflected by the reflector onto the image plane of the camera. The camera is an industrial-grade area array camera, located at the upper part of the sensor, and is responsible for converting the image of the weld area into a digital signal. The outer shell is made of a highly conductive and heat-dissipating metal material, which is used to support and fix all internal components and provide reliable protection and shielding. The socket is a high-reliability, long-life industrial-grade socket, internally connected to a line laser, a surface light source, and a camera; The cable is a high-reliability shielded twisted-pair cable used to connect the socket to the controller, provide power to the sensor and send control commands, and transmit the image signals obtained by the sensor to the controller. The controller is a micro industrial control computer, which is responsible for sending control commands to the sensors, controlling the line laser and the surface light source to alternately turn on and off, and simultaneously acquiring laser stripe images and weld area images; at the same time, it processes the image signals from the sensors to finally obtain the weld position information.

2. A method for correcting deviations based on the dual-sensor integrated weld seam tracking sensor as described in claim 1, specifically as follows: (1) Turn on the line laser, emit a line laser to the weld and form laser stripes on the surface of the weld, and obtain the image of the laser stripes; (2) Process the laser stripe image to obtain weld location information; (3) Turn off the line laser, turn on the area light source to illuminate the weld area, and obtain an image of the weld area; (4) Process the image of the weld area to obtain the deflection angle information of the weld; (5) When the weld seam deviates from the angle, the weld seam position information mentioned above is corrected according to the distance between the laser stripe and the welding gun, thereby obtaining the true and accurate weld seam position information. (6) The final weld position information is transmitted to the welding robot controller to correct the weld position deviation and finally realize online correction.

Images