CN104942404A - Dual-wavelength binocular vision seam tracking method and tracking system - Google Patents

Abstract

The invention discloses a dual-wavelength binocular vision welding seam tracking method, which includes the steps of image acquisition, data processing, and welding seam tracking. The invention also discloses a tracking system for realizing the method. The system uses near-infrared And structured light dual-wavelength binocular vision sensing system, simultaneously measure the weld area images of different wavelengths of the molten pool and weld and transmit them to the micro-industrial control computer, use the multi-information fusion algorithm and the three-dimensional reconstruction algorithm of the weld image to accurately measure the welding seam The seam position; the micro-industrial control computer applies the Kalman filter algorithm to optimally estimate the seam tracking deviation state according to the seam position detection results, and drives the servo motor to move through the servo drive to control the three-axis motion table to generate corresponding motion and control the welding seam. Torch or laser head can be used to rectify the deviation, so as to realize the precise tracking of the welding seam. The system can overcome strong light, spatter and electromagnetic interference at the welding site, and improve the accuracy and reliability of welding seam tracking.

Description

Dual-wavelength binocular vision welding seam tracking method and tracking system

technical field

The invention belongs to the field of welding technology, and in particular relates to a dual-wavelength binocular vision welding seam tracking method and tracking system.

Background technique

Welding is an important processing technology in the manufacturing industry, which has many characteristics such as harsh working conditions, heavy workload and high quality requirements. Arc welding and laser welding are more commonly used welding methods in the welding industry. Using arc and laser beam as the controlled object to realize automatic welding control is an important means of welding automation. Among them, accurate seam tracking is the premise to ensure the welding quality, that is, the laser beam or arc must be controlled throughout the welding process so that it is always aligned with the seam, otherwise it will cause waste. To this end, it is necessary to accurately and automatically detect the position of the weld seam and realize automatic tracking.

Since welding is a complex thermal processing technology, the workpiece will be thermally deformed during the welding process, and there will be strong radiation, arc light, smoke, spatter and other interference during the welding process, so that the weld position can be realized during the welding process. Accurate detection is quite difficult. At the same time, the weld path caused by factors such as welding device mechanism error, fixture assembly error, and weldment thermal deformation during welding is actually a three-dimensional curve weld, which involves a relatively complicated three-dimensional curve tracking problem. In addition, the weld gap is small, there is no groove, and the coordinates are not on the same plane, so automatic identification and measurement are extremely difficult.

At present, the acquisition of weld position information at home and abroad mainly focuses on the following methods: (1) structured light visual sensing method. The structured light visual sensing method scans the three-dimensional curved weld seam at the front end of the molten pool by emitting structured light from a laser source. Since the weld seam generally has a height difference with the steel plates on both sides, the contour line of the weld seam will be formed under the irradiation of structured light, showing 3D information of the weld seam. Most of the seam trackers currently used in industry work on the structured light vision sensing method. However, this method has its insurmountable defects: for the butt welding of flat plates of equal thickness, it can only effectively detect welds with a gap greater than 0.15mm. For welds with a gap less than 0.15mm, it is usually necessary to make a micro-groove on the surface of the butt weld before welding, so that the structured light can be deformed here. But this undoubtedly increases the processing cost and reduces the welding production efficiency. For welds with tight butt joints and no grooves, structured light hardly produces deformation. Therefore, it is difficult for traditional single-wavelength structured light measurement methods to accurately identify and measure such welds. Coupled with strong arc light and radiation interference, it is more likely to cause misjudgment of weld detection. (2) Infrared sensing method. This method is mostly used for weld seam identification of arc welding or laser welding with sensor on the back of the weldment. A certain temperature field is formed in and around the welding molten pool and accompanied by infrared radiation. The infrared camera is used to directly shoot the molten pool to obtain infrared thermal images. Quantitative analysis of the infrared thermal image of the arc welding area can obtain the quantitative information of the arc deviation from the weld seam. Due to the drastic melt pool and temperature changes, it is difficult to obtain a well-defined infrared image, and the infrared sensor is susceptible to environmental interference, so this method has problems such as low accuracy. (3) Direct image sensing method. This method uses the camera to directly shoot the molten pool, analyzes the gray distribution through image processing, and infers the deviation information of the weld center. Since the method directly acquires the melt pool image, the leading error is eliminated to a great extent. However, due to the drastic change of the molten pool and the weld at the molten pool has been melted, the information of the weld is basically obliterated, so it is difficult to fundamentally obtain the characteristics and laws of the weld deviation. (4) Other methods. For example, using a differential transformer as a sensor to detect vertical and horizontal deviations; according to the phenomenon that the weld gap has an impact on the propagation of acoustic emission waves, the acoustic emission-microprocessor-controlled weld detection system is used; The V-shaped weld seam detection at the mouth; the eddy current method is used to detect the weld seam, but the left and right deviation position of the weld seam cannot be accurately judged; the ultrasonic sensing method requires the close contact between the ultrasonic transducer and the welded plate, which greatly limits the ultrasonic sensor in the detection Applications for weld seams. The above methods have limitations in detecting small gap welds.

Contents of the invention

The main purpose of the present invention is to overcome the deficiencies of the above-mentioned existing seam tracking methods, and provide a seam tracking with dual-wavelength visual sensing and three-axis servo drive with high detection accuracy, reliable operation, easy to use, and certain versatility system.

In order to achieve the above object, the present invention provides a dual-wavelength binocular visual seam tracking method, the method comprises the following steps:

a. Image acquisition: The micro-industrial control computer issues instructions to make the structured light laser emit structured light across the weld seam, and at the same time issue instructions to start the near-infrared camera and the structured light camera to work, and coordinate the near-infrared camera and the structured light camera through the welding process synchronously The image acquisition card collects the infrared image of the molten pool area and the structured light image of the welding seam at the front of the molten pool, and transmits the obtained dual-wavelength image to the micro industrial control computer;

b. Data processing: apply the multi-information fusion algorithm to process the infrared image of the molten pool area and the structured light image of the welding seam at the front end of the molten pool, and calculate the accurate welding seam position and tracking correction amount;

c. Weld seam tracking: apply Kalman filter to optimally estimate the state of seam tracking deviation, find the relationship between filter estimation error and measurement error, and drive the servo motor to move through the servo drive to control the three-axis motion table to produce corresponding motion. Accurate tracking of weld seams is achieved.

In the above scheme, the near-infrared light captured by the near-infrared camera has a wavelength range of 960-990 nm; the visible structured light captured by the structured light camera has a wavelength range of 640-660 nm.

In order to realize the method, the present invention also discloses a dual-wavelength binocular visual seam tracking system, which includes a dual-wavelength binocular visual detection system, a three-axis servo-driven seam tracking system and a micro-industrial control machine. The binocular vision detection system includes a near-infrared camera, a structured light camera, a structured light laser, and a sensor mounting plate. The structured light camera is installed on the left front end of the sensor mounting plate, and the near-infrared camera is installed on the right front end of the sensor mounting plate. The structured light laser is located between the structured light camera and the near-infrared camera, and is installed at the middle end of the sensor mounting plate; the three-axis servo-driven seam tracking system includes an X-axis mount, a Y-axis mount, and a Z-axis mount. A welding table is installed on the upper surface of the axis fixing frame, and the welding parts are placed on the welding table. The X-axis servo motor is installed on the right end of the X-axis fixing frame. The middle and rear ends of the fixed frame are connected, the Z-axis servo motor is installed on the top of the Z-axis fixed frame, the Y-axis fixed frame is connected to the right side of the Z-axis fixed frame by bolts, and the Y-axis servo motor is installed on the Y-axis At the rear end of the fixed frame, the sensor mounting plate is installed on the front end of the Y-axis fixed frame, and the back of the sensor mounting plate is provided with reinforcing ribs, and the welding torch or laser head mounting plate is installed on the right front end of the Y-axis fixed frame through hexagon socket bolts. The welding torch or laser head is installed directly below the welding torch or laser head mounting plate.

In the above scheme, a long hole is opened on the right side surface of the Z-axis fixing frame, and the Z-axis servo motor drives the Y-axis fixing frame to move up and down in the long hole, and the welding torch can be adjusted by controlling the up and down position of the Y-axis fixing frame. Adjustment of the height from the welded part or the defocus amount during laser welding.

In the above scheme, a long hole is opened on the left side surface of the Y-axis fixing frame, and the Y-axis servo motor drives the Y-axis fixing frame to move back and forth in the long hole. By controlling the front and rear positions of the Y-axis fixing frame, the welding torch can Or laser head to achieve deviation correction.

In the above solution, the X-axis servo motor drives the welded parts placed on the welding table to move left and right, so as to realize the feeding of the welded parts.

In the above scheme, near-infrared light (960-990nm) and visible structured light (640-660nm) are used to obtain the shape of the molten pool and the information of the front weld seam of the molten pool respectively, and the information fusion technology of dual-wavelength binocular vision sensing is used to realize Seam position identification. On the basis of welding seam position measurement, establish a welding torch or laser head servo motion control system, according to the welding seam position coordinates, control the position of the welding torch or laser head through the servo drive system and apply Kalman filter control to make it always align and track welding seam, to achieve seam tracking for arc welding or laser welding.

In the above scheme, with butt joints, lap joints, fillet welds, etc. as the application objects, using dual-wavelength binocular vision sensor calibration, dual-band filter characteristics, weld image enhancement, and molten pool morphological feature extraction technologies, in harsh Accurately detect the position of the weld seam under the industrial welding site. When performing lap welding, fillet welding and butt welding of flat plates with large gaps, the welding seam position is detected mainly based on the visible deformable structured light emitted by the structured light laser to the weld seam and the structured light camera. When performing tight butt welding, the structured light is no longer deformed. At this time, the weld position is mainly obtained based on the heat distribution characteristics in the near-infrared molten pool image captured by the near-infrared camera. By fusing structured light weld images of different bands and near-infrared molten pool images, the measurement and tracking of the weld position is realized.

In the above solution, for the molten pool area under near-infrared light radiation and the weld seam area under visible structured light radiation, the characteristics of dual-wavelength images are used to enhance the ability to resist arc light interference. The micro-industrial control computer calculates the measurement coordinates of the binocular vision sensor, performs three-dimensional reconstruction of the weld image, and establishes a weld position measurement system.

In the above scheme, the three-axis movement of the welding torch or laser head and the seam tracking control system are established. On this basis, the Kalman filter is used for state optimal estimation, accurate prediction of the seam offset and offset direction, and elimination of process noise. and random interference effects of measurement noise. According to the offset, offset direction and offset speed between the weld and the arc or laser head, it can accurately predict the status of the weld and form a weld tracker.

Compared with the existing seam tracking method, the technical solution adopted by the present invention has the following beneficial effects:

1) The dual-wavelength binocular vision sensing method of near-infrared and structured light is used to obtain images of the weld area at different wavelengths of the molten pool and weld at the same time, and accurately measure the three-dimensional coordinates of the weld. This method can overcome strong arc light, radiation, spatter and electromagnetic interference at the welding site, and improve the fault tolerance and reliability of the system;

2) Integrate the visual tracker and servo drive, establish the three-axis movement of the welding torch or laser head and the seam tracking control model, apply the Kalman filter algorithm, eliminate the influence of the welding seam image noise on the measurement accuracy, and the welding seam and welding torch or The state optimal estimation of the deviation amount, the deviation direction and the deviation speed between the laser heads is carried out. For arc welding, implements predictive steering control of the welding torch that determines the position of the arc. For laser welding, a predictive guideline control of the laser head that determines the position of the laser beam is implemented.

Description of drawings

Fig. 1 is a front view of the three-dimensional structure of the present invention.

Fig. 2 is a rear view of the three-dimensional structure of the present invention.

Fig. 3 is a working schematic diagram of the present invention.

In the figure: 1- Z-axis servo motor, 2- Z-axis fixing frame, 3- sensor mounting plate, 4- structured light camera, 5- structured light laser, 6- welding parts, 7- X-axis fixing frame, 8- Y Axis servo motor, 9- Y-axis fixing frame, 10- welding torch or laser head mounting plate, 11- near infrared camera, 12- welding torch or laser head, 13- X-axis servo motor, 14- welding seam, 15- welding Workbench, 16- Rib.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. All other embodiments obtained by ordinary persons in the art without creative efforts belong to the protection scope of the present invention.

Embodiment 1: Dual-wavelength binocular vision welding seam tracking method.

Referring to FIG. 3 , FIG. 3 is a working diagram of the dual-wavelength binocular vision welding seam tracking method of the present invention. The method comprises the steps of:

a. Image acquisition: The micro-industrial control computer issues instructions to make the structured light laser emit structured light across the weld seam, and at the same time issue instructions to start the near-infrared camera and the structured light camera to work, and coordinate the near-infrared camera and the structured light camera through the welding process synchronously The image acquisition card collects the infrared image of the molten pool area and the structured light image of the welding seam at the front of the molten pool, and transmits the obtained dual-wavelength image to the micro industrial control computer;

b. Data processing: apply the multi-information fusion algorithm to process the infrared image of the molten pool area and the structured light image of the welding seam at the front end of the molten pool, and calculate the accurate welding seam position and tracking correction amount;

c. Weld seam tracking: apply Kalman filter to optimally estimate the state of seam tracking deviation, find the relationship between filter estimation error and measurement error, and drive the servo motor to move through the servo drive to control the three-axis motion table to produce corresponding motion. Accurate tracking of weld seams is achieved.

In the above scheme, the present invention takes butt joints, lap joints, and fillet welds as application objects, through dual-wavelength binocular vision sensor calibration, dual-band filter characteristic analysis, weld image enhancement, molten pool morphological feature extraction, Multi-information fusion and Kalman filter technology can accurately detect the position of the weld seam and realize weld seam tracking under harsh industrial welding conditions.

In the above scheme, the present invention aims at the molten pool area under near-infrared radiation and the weld area under visible structured light radiation, simultaneously measures the weld area images of different wavelengths of the molten pool and the weld, and extracts the features of the dual-wavelength image , fused with the obtained feature information to accurately calculate the three-dimensional coordinates of the weld. The micro-industrial control computer calculates the measurement coordinates of the binocular vision sensor, performs three-dimensional reconstruction of the weld image, and establishes a weld position measurement system. Apply the Kalman filter algorithm to eliminate the influence of welding seam image noise on the measurement accuracy, and perform state optimal estimation of the deviation, deviation direction and deviation speed between the welding seam and the arc or laser beam, and realize the welding torch or laser head Predictive steering control of position.

In the above scheme, the near-infrared light captured by the near-infrared camera has a wavelength range of 960-990 nm; the visible structured light captured by the structured light camera has a wavelength range of 640-660 nm.

Embodiment 2: A dual-wavelength binocular visual seam tracking system.

Referring to Figures 1 and 2, a dual-wavelength binocular vision seam tracking system includes a dual-wavelength binocular vision detection system, a three-axis servo-driven seam tracking system and a miniature industrial control computer. The dual-wavelength binocular vision The detection system includes a near-infrared camera 11, a structured light camera 4, a structured light laser 5 and a sensor mounting plate 3, the structured light camera 4 is installed on the left front end of the sensor mounting plate 3, and the near-infrared camera 11 is installed on the sensor mounting plate 3 Right front end, the structured light laser 5 is located between the structured light camera 4 and the near-infrared camera 11, and is installed on the middle end of the sensor mounting plate 3; the three-axis servo-driven seam tracking system includes an X-axis fixing frame 7, a Y-axis fixing frame 9 and Z-axis fixed frame 2, the upper surface of the X-axis fixed frame 7 is equipped with a welding table 15, the weldment 6 is placed on the welding table 15, and the X-axis servo motor 13 is installed on the X-axis The right end of the fixed frame 7, the lower front end of the Z-axis fixed frame 2 is connected with the middle and rear end of the X-axis fixed frame 7, the Z-axis servo motor 1 is installed on the top of the Z-axis fixed frame 2, and the Y-axis is fixed The frame 9 is connected to the right side of the Z-axis fixed frame 2 by bolts, the Y-axis servo motor 8 is installed on the rear end of the Y-axis fixed frame 9, the sensor mounting plate 3 is installed on the front end of the Y-axis fixed frame 9, and the sensor is installed Ribs 16 are installed on the back of the plate 3, and the welding torch or laser head mounting plate 10 is installed on the right front end of the Y-axis fixing frame 9 through an inner hexagonal bolt, and the welding torch or laser head 12 is installed on the welding torch or laser head mounting plate 10 directly below the .

In the above scheme, the right side surface of the Z-axis fixing frame 2 has a long hole, and the Z-axis servo motor 1 drives the Y-axis fixing frame 9 to move up and down in the long hole. By controlling the up and down position of the Y-axis fixing frame 9 The height of the welding torch from the weldment 6 or the adjustment of the defocus amount during laser welding can be adjusted.

In the above scheme, there is a long hole on the left side surface of the Y-axis fixed frame 9, and the Y-axis servo motor 8 drives the Y-axis fixed frame 9 to move back and forth in the long hole. By controlling the front and rear positions of the Y-axis fixed frame 9 Correction of the welding torch or laser head 12 can be realized.

In the above solution, the X-axis servo motor 1 drives the welding piece 6 placed on the welding table 15 to move left and right, so as to realize the feeding of the welding piece 6 .

In the above scheme, the three-axis linkage servo motion platform includes X-axis fixed frame 7, X-axis table 15, Y-axis fixed frame 9, Z-axis fixed frame 2, X-axis servo motor 13, Y-axis servo motor 8 and Z-axis Servo motor 1; the dual-wavelength binocular vision sensing device includes a sensor mounting plate 3 and a structured light camera 4, a structured light laser 5 and a near-infrared camera 11 mounted thereon, and the structured light camera 4 is used to obtain the For the image of the structured light, the near-infrared camera 11 is used to obtain the dynamic image of the molten pool during the welding process. The sensor mounting plate 3 is fixed and can be linked with the Y-axis fixing frame 9 and the welding torch or laser head mounting plate 10 . A reinforcing rib 16 is connected between the sensor mounting plate 3 and the Y-axis fixing frame 9 to ensure the rigidity of the device. The welding torch or laser head mounting plate 10 is installed on the Y-axis fixed frame 9 through the inner hexagonal bolt, and the freely replaceable welding torch or laser head 12 is installed on the welding torch or laser head mounting plate 10 through the inner hexagonal bolt, according to Characteristics such as welding material, thickness, welding torch or laser head 12 can select laser welding head or welding torch. According to the horizontal distance from the welding torch or laser head 12, the near-infrared camera 11, the structured light laser 5 and the structured light camera 4 are installed in turn. When installing, the axis of the near-infrared camera 11 forms an angle with the vertical direction to be able to obtain Image of the melt pool region. Similarly, the structured light camera 4 and the structured light laser 5 are also installed with their axis at an angle to the vertical direction so as to be able to acquire the structured light image, and the structured light image has a leading amount in the welding direction relative to the image of the molten pool area.

In the above scheme, the working principle of the system is as follows: near-infrared camera 11, structured light camera 4, structured light laser 5, welding torch or laser head 12 and servo driver are respectively connected to the micro-industrial control machine. The X-axis fixed frame 7 is used for welding feed, the Y-axis fixed frame 9 is used for deviation correction of the welding torch or laser head 12, and the Z-axis fixed frame 2 is used for adjusting the height of the welding torch from the weldment 6 or the defocus during laser welding Quantity adjustment. When using the present invention for seam tracking, the weldment 6 is placed on the welding workbench 15, and the micro-industrial controller coordinates and starts the near-infrared camera 11, structured light camera 4, structured light laser 5, welding torch or laser head 12 and The servo drive works. The dual-band filter system is used to filter out stray light, collect near-infrared light, structured light and surrounding area information. The micro-industrial control computer adjusts the dynamic range of the weld seam and molten pool images, and uses the image three-dimensional reconstruction method to establish the weld seam position measurement equation. According to the measurement information of the welding seam position, the three-axis servo drive technology is integrated to establish the motion control model of the welding torch or laser head. Through the fusion technology of welding seam visual information and three-axis linkage control system of welding torch or laser head, the tracking control of welding seam is realized. Establish a welding seam tracking control system that integrates visual sensing and welding torch spatial displacement, uses near-infrared light and structured light dual-wavelength binocular vision sensors to obtain images of molten pools and weld seams, and uses three-axis ( X, Y, Z) displacement control method. The micro-industrial control machine applies Kalman filter to optimally estimate the deviation of the welding seam, and timely feeds back the control parameters for the deviation correction movement of the welding torch or laser head 12, so as to realize the predictive correction control of the movement of the welding torch or laser head 12. The system can overcome strong light, radiation, spatter and electromagnetic interference at the welding site, and improve the fault tolerance and reliability of the system.

The above description is only a preferred embodiment of the present invention, but the present invention should not be limited to the content disclosed in the embodiment and accompanying drawings, so any equivalent or modification that does not depart from the disclosed spirit of the present invention can be done. Fall into the protection scope of the present invention.

Claims (8)

1. a dual wavelength binocular vision welding seam tracking method, is characterized in that, the method comprises the steps:

A) IMAQ: miniature Industrial Control Computer sends instruction makes structured light laser instrument (5) launch the structured light being across weld seam, send instruction simultaneously and start near-infrared video camera (11), structured light video camera (4) work, in welding process, synchronous coordination near-infrared video camera (11) and structured light video camera (4) gather the structure light image of molten bath zone infrared image and front end, molten bath weld seam respectively by image pick-up card, and are transferred in miniature Industrial Control Computer by the double-wavelength images of acquisition;

B) data processing: the structure light image of application Multi-information acquisition algorithm to molten bath zone infrared image and front end, molten bath weld seam processes, calculates position while welding and tracking correction amount accurately;

C) weld joint tracking: application card Kalman Filtering butt welded seam tracing deviation state carries out optimal estimation, find the relation between filtering evaluated error and measure error, moved by servo driver drives servomotor thus control three-axis moving workbench and produce corresponding motion, realize the accurate tracking of weld seam.

2. dual wavelength binocular vision welding seam tracking method according to claim 1, is characterized in that: the near-infrared wavelength scope that described near-infrared video camera (11) is absorbed is 960-990nm; The visual structure optical wavelength range that described structured light video camera (4) is absorbed is 640-660nm.

3. a dual wavelength binocular vision seam tracking system, comprise dual wavelength binocular vision detection system, three axle servo-drive seam tracking systems and miniature Industrial Control Computer, it is characterized in that: described dual wavelength binocular vision detection system comprises near-infrared video camera (11), structured light video camera (4), structured light laser instrument (5) and sensor installing plate (3), described structured light video camera (4) is arranged on sensor installing plate (3) left front end, described near-infrared video camera (11) is arranged on sensor installing plate (3) right front ends, described structured light laser instrument (5) is positioned in the middle of structured light video camera (4) and near-infrared video camera (11), be arranged on sensor installing plate (3) middle-end, described three axle servo-drive seam tracking systems comprise X-axis fixed mount (7), Y-axis fixed mount (9) and Z axis fixed mount (2), described X-axis fixed mount (7) upper surface is provided with welding bench (15), weldment (6) is placed on described welding bench (15), X-axis servomotor (13) is arranged on the right-hand member of described X-axis fixed mount (7), the lower front end of described Z axis fixed mount (2) is connected with the middle rear end of X-axis fixed mount (7), Z axis servomotor (1) is arranged on the top of described Z axis fixed mount (2), described Y-axis fixed mount (9) is connected to Z axis fixed mount (2) right side, Y-axis servomotor (8) is arranged on the rear end of described Y-axis fixed mount (9), sensor installing plate (3) is arranged on the front end of Y-axis fixed mount (9), described sensor installing plate (3) back side is provided with reinforcement (16), welding torch or laser head installing plate (10) are arranged on the right front ends of Y-axis fixed mount (9), welding torch or laser head (12) are arranged on immediately below described welding torch or laser head installing plate (10).

4. dual wavelength binocular vision seam tracking system according to claim 3, is characterized in that: described Z axis fixed mount (2) right lateral surface has elongated hole, and Z axis servomotor (1) drives Y-axis fixed mount (9) to move up and down in described elongated hole.

5. dual wavelength binocular vision seam tracking system according to claim 3, is characterized in that: described Y-axis fixed mount (9) left-hand face has elongated hole, and Y-axis servomotor (8) drives Y-axis fixed mount (9) to seesaw in described elongated hole.

6. dual wavelength binocular vision seam tracking system according to claim 3, is characterized in that: weldment (6) side-to-side movement that the driving of described X-axis servomotor (13) is placed on welding bench (15).

7. dual wavelength binocular vision seam tracking system according to claim 3, is characterized in that: described dual wavelength binocular vision detection system is connected with miniature Industrial Control Computer respectively by bus with three axle servo-drive seam tracking systems.

8. dual wavelength binocular vision seam tracking system according to claim 3, is characterized in that: the control system that described dual wavelength binocular vision detection system forms with three axle servo-drive seam tracking systems and miniature Industrial Control Computer is closed-loop control system.