Vision-based profile steel geometric defect automatic inspection robot and positioning deviation correction method

Abstract

The invention discloses a vision-based profile steel geometric defect automatic inspection robot positioning and deviation rectifying method, and relates to the technical field of profile steel detection. The system comprises a moving module, an operation module and a scanning module. The scanning module comprises a three-dimensional camera and is used for scanning a to-be-detected component; the operation module is a robot arm, carries the scanning module and sends the scanning module to a proper position for scanning; and the moving module comprises a reference platform and a moving platform and is used for carrying the operation module and the scanning module to run to a specified position for scanning operation, monitoring the advancing direction of the moving platform in real time and controlling and correcting the direction of the moving platform according to a feedback result. The device is large in detection working range, high in precision, high in automation degree and low in cost, and can meet the detection requirements of profile steel structure processing factories.

Description

Technical Field

[0001] The invention belongs to the technical field of steel section detection equipment, and more specifically, relates to a vision-based automatic inspection robot for steel section geometric defects and a positioning and deviation correction method. Background Art

[0002] Steel structures are widely used in purlins, wall beams, roof panels, and wall panels of steel structures due to their light weight and easy assembly. Since the mechanical properties of steel components are extremely sensitive to the initial geometric defects generated during their manufacturing, transportation, storage, and use, and ultimately affect their failure mode and stability, it is very important to detect the initial defects of steel components, and obtaining an accurate steel structure defect model has become a topic of concern.

[0003] At present, the mainstream defect detection method uses tools such as vernier calipers, theodolites, and levels for manual detection. This detection method is not only inefficient, but also has high labor costs and poor accuracy. In recent years, the use of 3D laser scanning technology for steel defect detection has also become very popular. In order to ensure accuracy, it is necessary to manually position the components. Although accuracy is guaranteed, it still requires a lot of labor costs and the efficiency is still low. In addition, the high value of the equipment makes it difficult to promote the technology.

[0004] In order to improve the defects of mainstream methods, some technologies that use vehicle robots to detect geometric defects in steel have emerged in recent years. These technologies generally use SLAM technology or wireless sensor positioning technology for positioning and correction. These two new technologies can meet the accuracy requirements of geometric defect detection in steel and meet automation requirements, but the equipment costs are relatively high. The latter requires the overall transformation of the steel processing factory and the establishment of signal base stations to complete it, which is not conducive to the market promotion of the technology.

[0005] For large-scale steel component processing plants, detecting the initial geometric defects of the steel materials they process is an important task. However, neither the traditional manual inspection nor the emerging handheld 3D laser scanning technology and the existing robotic steel component scanning technology can simultaneously meet the needs of a large working range, high precision and low cost, and the degree of automation is relatively low. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide a vision-based automatic inspection robot for geometric defects of steel sections and a positioning and correction method, which has a large detection working range, high precision, high degree of automation, and low cost, and can meet the detection requirements of steel structure processing plants.

[0007] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0008] A vision-based automatic inspection robot for geometric defects of steel sections comprises a mobile module, an operating module and a scanning module; the scanning module comprises a three-dimensional camera for scanning the components to be inspected; the operating module is a robot arm, which carries the scanning module and sends the scanning module to a suitable position for scanning; the mobile module is used to carry the operating module and send the operating module to a designated position; the mobile module can perform positioning and direction correction during movement.

[0009] Preferably, the mobile module includes a reference platform and a mobile platform, the reference platform includes a reference platform bracket, a first laser rangefinder, and a second laser rangefinder, the first laser rangefinder and the second laser rangefinder are installed on the reference platform bracket, the light beams of the two laser rangefinders are parallel to each other, and the visible light beam emitted by each laser rangefinder is concentric with the light beam of the emitted ranging laser; during measurement, the reference platform bracket is fixedly connected next to the component to be inspected; the mobile platform includes a photosensitive plate, a first visual camera, a second visual camera, and a vehicle model robot, the first visual camera and the second visual camera are located on both sides of the photosensitive plate, and scale lines are provided on the photosensitive plate, the lasers emitted by the first laser rangefinder and the second laser rangefinder can be irradiated on the photosensitive plate, and the photosensitive plate, the first visual camera, the second visual camera, and the robot arm are all fixedly connected to the vehicle model robot and move with the vehicle model robot.

[0010] A vision-based positioning and deviation correction method for an automatic inspection robot for geometric defects of steel sections. The positioning method of the automatic inspection robot is as follows:

[0011] The first laser rangefinder and the second laser rangefinder are on the same horizontal line, with their lens origin as the origin, and establish coordinate systems P1 (x1, y1, z1) respectively. T and P2(x2, y2, z2) T ; P1 and P2 coordinate systems are both right-handed systems, with the z1 axis and z2 axis pointing vertically upward, and the y1 axis and y2 axis coincide with the visible lasers emitted by the first laser rangefinder and the second laser rangefinder; since the working environment is smooth, the first laser rangefinder and the second laser rangefinder are installed on the same horizontal line, z1=z2=0; the distance between the origins of the two coordinate systems is known, which is l′, and the conversion relationship between P1 and P2 can be obtained:

[0012]

[0013] There is a standard scale on the photosensitive plate to calculate the position of the vehicle robot; the scale line on the photosensitive plate close to the edge of the first visual camera and the light spot are taken as the origin to establish the mobile platform coordinate system P m (x m ,y m , z m ) T , the coordinate system is along the moving direction of the mobile platform as ym Axis, vertical upward direction is z m Axis, also uses the right-hand coordinate system to express coordinates and postures; when the working condition is flat, z1=z2=z m =0; when the vehicle robot is on the correct path, the y1 axis, y2 axis and y m The axes always remain parallel;

[0014] When it is necessary to determine the precise positioning of the reference platform i at a certain moment, the visible lasers emitted by the first laser rangefinder and the second laser rangefinder leave two visible light spots on the photosensitive plate. The visible light spot left by the laser emitted by the first laser rangefinder is the first light spot, and the visible light spot left by the laser emitted by the second laser rangefinder is the second light spot, which can be obtained by visually observing the first light spot and the second light spot at the scale position of the photosensitive plate through the first visual camera and the second visual camera; because the working environment is smooth, the first light spot and the second light spot are at the same vertical height on the photosensitive plate, and the coordinates are (u1, v1) and (u2, v1) respectively; further, it is known that the coordinates of the first light spot in the reference coordinate system P1 at this time are (0, l1, z1), and the coordinates of the second light spot in the reference coordinate system P2 are (0, l2, z2). According to formula (1), it can be deduced that the transformed coordinates of the second light spot in the reference coordinate system are (l', l2, z1);

[0015] We can get x at this time m The deflection angle of the axis relative to the x1 axis

[0016]

[0017] When the relative deflection angle is known, the mobile platform coordinate system P can be determined m The coordinates of the origin in the P1 coordinate system are (x0, y0, 0), and P is derived m Conversion relationship with P1;

[0018]

[0019]

[0020]

[0021]

[0022]

[0023] Among them, R is the rotation transformation matrix, and T is the translation transformation matrix;

[0024] After positioning the car, it is possible to determine whether the car is moving in the wrong direction. Assuming that the working environment of the mobile platform is smooth, the target coordinates on the photosensitive plate of the first light spot and the second light spot are (u 10 , v1) and (u 20 , v1), when the first visual camera and the second visual camera judge u through computer vision at time s 10 ≠u1 or u 20 ≠u2, it is necessary to correct the route, that is, to correct the deviation;

[0025] Assume that the mobile platform is in the P1 coordinate system, and its velocities on the x-axis, y-axis, and counterclockwise around the z-axis are dx1, dy1 respectively.

[0026] The deviation correction method of the automatic inspection robot includes the following steps:

[0027] S1. Positioning: First, the mobile platform is paused and positioned according to the above-mentioned automatic inspection robot positioning method, so as to determine the reference platform tracking coordinate system P m Deflection angle relative to coordinate system P1 And the coordinate system P m The coordinates of the origin in the coordinate system P1 (x m1s ,y m1s , z m1s );

[0028] S2. Determine the target point: While the system is positioning the mobile platform, it will retrieve and return the last normal shooting site number i, thereby determining site N i+1 is the target site and obtains its coordinates in the coordinate system P1;

[0029] S3. Determine the error: according to the target position N i+1 Coordinates can determine the correction task target x 10 ,y 10 , are the correct coordinates and deflection angles of the mobile platform at the target point, respectively, and the error values ​​Δx, Δy can be obtained.

[0030] Δx=x m1s -x 10 (8)

[0031] Δy=y m1s -y 10 (9)

[0032]

[0033] because but

[0034]

[0035] S4. Determine the speed of the mobile platform: set the speed control rate k of the three degrees of freedom respectively x , k y ,

[0036]

[0037] S5, deviation correction: dx1, dy1, Converted into the three-degree-of-freedom velocity dx of the mobile platform's servo coordinate system m ,dy m , And control the mobile platform to move to the target point;

[0038]

[0039]

[0040]

[0041] S6. Confirm and shoot: Move to the scheduled site N i+1 After that, positioning should be performed again. When the error values ​​are all 0, you can continue shooting and scanning normally.

[0042] The beneficial effects of adopting the above technical solution are:

[0043] 1. The scanning robot and depth camera combination of the present invention scans the geometric defect information of steel structure, which has the application characteristics of low cost, high precision, high degree of automation, and no need for point sticking. It can meet the wide range of application needs of large-scale steel structure manufacturers and improve their production efficiency.

[0044] 2. The positioning and correction method proposed in the present invention can accurately position the robot car in space, and further provide good coarse registration conditions for subsequent point cloud stitching. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the overall structure of the robot;

[0046] Figure 2 It is a schematic diagram of route planning and site settings;

[0047] Figure 3 It is a schematic diagram of the working scene;

[0048] Figure 4 It is a schematic diagram of point cloud stitching;

[0049] Figure 5 It is a schematic diagram of the work process;

[0050] Figure 6 It is a schematic diagram of the coordinates of the mobile module;

[0051] Figure 7 It is the principle diagram of positioning and deviation correction of mobile module;

[0052] Figure 8 It is a working schematic diagram of the moving module of the component to be tested;

[0053] In the figure: 101, reference platform bracket; 102, first laser rangefinder; 103, second laser rangefinder; 111, photosensitive plate; 112, first visual camera; 113, second visual camera; 114, vehicle model robot; 115, robot arm; 116, three-dimensional camera. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0055] like Figure 1 As shown, the automatic inspection robot includes a mobile module, an operation module, and a scanning module. The scanning module includes a three-dimensional camera 116 to scan the component to be inspected. The three-dimensional camera 116 saves the point cloud model of the component under each viewing angle, and belongs to the point cloud stitching algorithm. The point cloud stitching algorithm performs rough registration of the point cloud model under multiple viewing angles through the saved ranging laser distance value and the robot arm coordinate inverse solution value, and introduces the ICP algorithm to perform point cloud fine registration to generate a component point cloud model that meets the accuracy requirements.

[0056] The mobile module includes a reference platform and a mobile platform. The reference platform includes a reference platform bracket 101, a first laser rangefinder 102, and a second laser rangefinder 103. The first laser rangefinder 102 and the second laser rangefinder 103 are mounted on the reference platform bracket 101. The light beams of the two laser rangefinders are parallel to each other, and the visible light beam emitted by each laser rangefinder is concentric with the light beam of the emitted ranging laser. During measurement, the reference platform bracket 101 is fixedly connected next to the component to be inspected. The mobile platform includes a photosensitive plate 111, a first visual camera 112, a second visual camera 113, and a vehicle robot 114. The vehicle robot 114 is an omnidirectional mobile vehicle robot. The first visual camera 112 and the second visual camera 113 are located on both sides of the photosensitive plate 111. Scale lines are provided on the photosensitive plate 111, and the lasers emitted by the first laser rangefinder 102 and the second laser rangefinder 103 irradiate the photosensitive plate 111 leaving light spots. The visible light spot left by the laser emitted by the first laser rangefinder 102 is the first light spot, and the visible light spot left by the laser emitted by the second laser rangefinder 103 is the second light spot. The photosensitive plate 111, the first visual camera 112, the second visual camera 113, and the robot arm 115 are all fixedly connected to the vehicle robot 114 and move with the vehicle robot 114. The mobile module can perform positioning and direction correction during the movement.

[0057] The operation module is composed of a robot arm 115. The robot arm 115 carries the scanning module and sends the scanning module to a suitable position for scanning. The robot arm 115 is mounted on the vehicle robot 114, and the vehicle robot 114 sends the operation module to a designated position. When the mobile platform stops at the expected scanning point, the robot arm 115 can send the three-dimensional camera 116 to a suitable position for scanning according to the actual conditions such as the shape of the component and the scanning requirements, and return and save its coordinate information.

[0058] The working principle of the entire automatic inspection robot system is as follows:

[0059] Take the inspection robot scanning a point cloud model of a steel structure to be tested as an example. Figure 2 As shown in the figure, to ensure that the designed robot system can effectively carry out scanning along the length of the component, an accurate straight line path needs to be planned based on the length L of the component to be measured and the scanning coverage of the scanning module at a single station before scanning begins. n scanning stations are reasonably arranged on this path to achieve station-by-station accurate scanning of the designed robot system.

[0060] After the scan is started, the mobile module moves along the predetermined path and stops briefly at each preset scanning station. Figure 3 As shown, when the mobile module arrives at the designated site N iWhen the robot stops moving, it adjusts its posture and accurately positions the scanning module to the appropriate scanning position for the component. The system automatically collects point cloud data of the component in the vicinity of the site, and returns and records the spatial coordinate information of the scanning module.

[0061] like Figure 4 As shown, the mobile platform is at site N i With site N i+1 The two scanning ranges have overlapping viewing angles. Save N according to the above workflow. i With N i+1 After shooting the point cloud model, the algorithm can convert N i With N i+1 Shoot point cloud models and stitch and align them. Further stitch and align from N1 to N n The point cloud model can obtain a complete model that meets the requirements for component structure geometric defect detection.

[0062] During the movement of the automatic inspection robot, the travel route will be offset because the travel system of the vehicle robot 114 is not fully constrained. In order to ensure the scanning accuracy, it is necessary to correct its travel route through feedback positioning of the global position. The subsequent point cloud stitching and registration work also requires accurate coordinate values ​​of the vehicle robot 114 in space. Based on this need, a feedback control method is designed to automatically return coordinates and correct the travel route of the vehicle robot 114.

[0063] The positioning method of the automatic inspection robot is as follows:

[0064] like Figure 6 , Figure 7 As shown, the first laser rangefinder 102 and the second laser rangefinder 103 are placed on the same horizontal line, with their lens origin as the origin. Coordinate systems P1 (x1, y1, z1) are established respectively. T and P2(x2,y2,z2) T . The P1 and P2 coordinate systems are both right-handed systems, with the z1 axis and the z2 axis pointing vertically upward, and the y1 axis and the y2 axis coincide with the visible lasers emitted by the first laser rangefinder 102 and the second laser rangefinder 103. Since the designed system working environment has no fluctuations, the first laser rangefinder 102 and the second laser rangefinder 103 are installed on the same horizontal line, and z1=z2=0. The distance between the origins of the two coordinate systems is known, which is l′, and the conversion relationship between P1 and P2 can be obtained:

[0065]

[0066] like Figure 6As shown, the photosensitive plate 111 has a standard scale to calculate the position of the vehicle robot 114. The edge scale line and the light spot on the photosensitive plate 111 close to the first visual camera 112 are taken as the origin to establish the mobile platform coordinate system P m (x m ,y m , z m ) T , the coordinate system is along the moving direction of the mobile platform as y m Axis, vertical upward direction is z m Axis, also uses the right-hand coordinate system to express coordinates and posture. When the working condition is flat, z1 = z2 = z m = 0. When the vehicle robot 114 is on the correct path, the y1 axis, y2 axis and y m The axes always remain parallel.

[0067] When it is necessary to determine the precise positioning of the reference platform i at a certain moment, it can be obtained by visually observing the first light spot and the second light spot on the scale position of the photosensitive plate 111 through the first visual camera 112 and the second visual camera 113. Because the working environment is smooth, the first light spot and the second light spot have the same vertical height on the photosensitive plate 111, which are (u1, v1) and (u2, v1) respectively. Further, it is known that the coordinates of the first light spot in the reference coordinate system P1 are (0, l1, z1), and the coordinates of the second light spot in the reference coordinate system P2 are (0, l2, z2). According to formula (1), it can be deduced that the transformed coordinates of the second light spot in the reference coordinate system are (l', l2, z1).

[0068] We can get x at this time m The deflection angle of the axis relative to the x1 axis

[0069]

[0070] When the relative deflection angle is known, the mobile platform coordinate system P can be determined m The coordinates of the origin in the P1 coordinate system are (x0, y0, 0), and P is derived m Conversion relationship with P1.

[0071]

[0072]

[0073]

[0074]

[0075]

[0076] After positioning the car, it is possible to determine whether the car is moving in the wrong direction.

[0077] Assuming that the working environment of the mobile platform is smooth, the target coordinates of the first light spot and the second light spot on the photosensitive plate 111 are (u 10 , v1) and (u 20 , u1). At time s, the first visual camera 112 and the second visual camera 113 determine u through computer vision. 10 ≠u1 or u 20 ≠u2, it is necessary to correct the route, that is, to correct the deviation.

[0078] Assume that the mobile platform is in the P1 coordinate system, and its velocities on the x-axis, y-axis, and counterclockwise around the z-axis are dx1, dy1 respectively.

[0079] The correction work is divided into the following steps:

[0080] S1. Positioning: First, make the mobile platform pause. According to the method introduced in the previous section, determine the reference platform tracking coordinate system P m Deflection angle relative to coordinate system P1 And the coordinate system P m The coordinates of the origin in the coordinate system P1 (x m1s ,y m1s , z m1s ).

[0081] S2. Determine the target point: While the system is positioning the mobile platform, it will retrieve and return the last normal shooting site number i, thereby determining site N i+1 is the target site and obtains its coordinates in the coordinate system P1.

[0082] S3. Determine the error: according to the target position N i+1 Coordinates can determine the correction task target x 10 ,y 10 , are the correct coordinates and deflection angles of the mobile platform at the target point, respectively, and the error values ​​Δx, Δy can be obtained.

[0083] Δx=x m1s -x 10 (8)

[0084] Δy=y m1s -y 10 (9)

[0085]

[0086] because but

[0087]

[0088] S4. Determine the speed of the mobile platform: set the speed control rate k of the three degrees of freedom respectively x , k y ,

[0089]

[0090] S5, deviation correction: dx1, dy1, Converted into the three-degree-of-freedom velocity dx of the mobile platform's servo coordinate system m ,dy m , And control the mobile platform to move to the target point.

[0091]

[0092]

[0093]

[0094] S6. Confirm and shoot: Move to the scheduled site N i+1 After that, positioning should be performed again. When the error values ​​are all 0, you can continue shooting and scanning normally.

[0095] When multiple steel structures need to be inspected for geometric defects, multiple reference platforms can be set up for point cloud coordinate positioning and automated inspection. Figure 8 As shown, take the case where there are two components to be measured as an example. First, preset the lengths of component one and component two to be measured and scan component one. The outbound control method is consistent with the above description, and reference platform one controls the mobile platform to move along the outbound journey. After reaching the preset length of component one to be measured, it automatically deflects β in the direction of component two to be measured and continues to move. After traveling an appropriate distance, it deflects -β in the original deflection direction, that is, it reaches the return point through ② cutting and deflects the mobile platform to the original direction of travel. At this time, the first visual camera 112 and the second visual camera 113 can detect the visible light spot left by reference platform two on the photosensitive plate 111, and can detect the position coordinates of the reference platform relative to reference platform two. Using the above algorithm, the mobile platform is controlled to retreat to the starting position of component two to be measured and start scanning component two.

[0096] The automatic inspection robot for steel structures designed in the present invention, through its mobile module, meets the requirements of low-cost, large-scale, high-precision positioning and geometric defect detection in large-scale steel structure processing plants, is beneficial to the healthy development of my country's steel structure industry, and has also contributed to my country's civil engineering safety to a certain extent.

[0097] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and inventive concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A vision-based automatic inspection robot for geometric defects of steel sections, characterized in that: It includes a moving module, an operating module, and a scanning module; the scanning module includes a three-dimensional camera to scan the component to be inspected; the operating module is a robot arm, which carries the scanning module and sends the scanning module to a suitable position for scanning; the moving module is used to carry the operating module and send the operating module to a designated position; the moving module can perform positioning and direction correction during movement.

2. The vision-based automatic inspection robot for geometric defects of steel sections according to claim 1 is characterized in that: The mobile module includes a reference platform and a mobile platform. The reference platform includes a reference platform bracket, a first laser rangefinder, and a second laser rangefinder. The first laser rangefinder and the second laser rangefinder are installed on the reference platform bracket. The light beams of the two laser rangefinders are parallel to each other. The visible light beam emitted by each laser rangefinder is concentric with the light beam of the emitted ranging laser. During measurement, the reference platform bracket is fixedly connected next to the component to be inspected. The mobile platform includes a photosensitive plate, a first visual camera, a second visual camera, and a vehicle model robot. The first visual camera and the second visual camera are located on both sides of the photosensitive plate. Scale lines are provided on the photosensitive plate. The lasers emitted by the first laser rangefinder and the second laser rangefinder can be irradiated on the photosensitive plate. The photosensitive plate, the first visual camera, the second visual camera, and the robot arm are all fixedly connected to the vehicle model robot and move with the vehicle model robot.

3. A vision-based steel geometric defect automatic inspection robot positioning and correction method, characterized in that: The positioning method of the automatic inspection robot is as follows: The first laser rangefinder and the second laser rangefinder are on the same horizontal line, with their lens origin as the origin, and establish coordinate systems P1 (x1, y1, z1) respectively. T and P2(x2, y2, z2) T ; P1 and P2 coordinate systems are both right-handed systems, with the z1 axis and z2 axis pointing vertically upward, and the y1 axis and y2 axis coincide with the visible lasers emitted by the first laser rangefinder and the second laser rangefinder; since the working environment is smooth, the first laser rangefinder and the second laser rangefinder are installed on the same horizontal line, z1=z2=0; the distance between the origins of the two coordinate systems is known, which is l′, and the conversion relationship between P1 and P2 can be obtained: There is a standard scale on the photosensitive plate to calculate the position of the vehicle robot; the scale line on the photosensitive plate close to the edge of the first visual camera and the light spot are taken as the origin to establish the mobile platform coordinate system P m (x m ,y m , z m ) T , the coordinate system is along the moving direction of the mobile platform as y m Axis, vertical upward direction is z m Axis, also uses the right-hand coordinate system to express coordinates and postures; when the working condition is flat, z1=z2=z m =0; when the vehicle robot is on the correct path, the y1 axis, y2 axis and y m The axes always remain parallel; When it is necessary to determine the precise positioning of the reference platform i at a certain moment, the visible lasers emitted by the first laser rangefinder and the second laser rangefinder leave two visible light spots on the photosensitive plate. The visible light spot left by the laser emitted by the first laser rangefinder is the first light spot, and the visible light spot left by the laser emitted by the second laser rangefinder is the second light spot, which can be obtained by visually observing the first light spot and the second light spot at the scale position of the photosensitive plate through the first visual camera and the second visual camera; because the working environment is smooth, the first light spot and the second light spot are at the same vertical height on the photosensitive plate, and the coordinates are (u1, b1) and (u2, b1) respectively; further, it is known that the coordinates of the first light spot in the reference coordinate system P1 at this time are (0, l1, z1), and the coordinates of the second light spot in the reference coordinate system P2 are (0, l2, z2). According to formula (1), it can be deduced that the transformed coordinates of the second light spot in the reference coordinate system are (l', l2, z1); We can get x at this time m The deflection angle of the axis relative to the x1 axis When the relative deflection angle is known, the mobile platform coordinate system P can be determined m The coordinates of the origin in the P1 coordinate system are (x0, y0, 0), and P is derived m Conversion relationship with P1; Among them, R is the rotation transformation matrix, and T is the translation transformation matrix; After positioning the car, it is possible to determine whether the car is moving in the wrong direction. Assuming that the working environment of the mobile platform is smooth, the target coordinates on the photosensitive plate of the first light spot and the second light spot are (u 10 , v1) and (u 20 , v1), when the first visual camera and the second visual camera judge u through computer vision at time s 10 ≠u1 or v 20 ≠u2, it is necessary to correct the route, that is, to correct the deviation; Assume that the mobile platform is in the P1 coordinate system, and its velocities on the x-axis, y-axis, and counterclockwise around the z-axis are dx1, dy1 respectively. The deviation correction method of the automatic inspection robot includes the following steps: S1. Positioning: First, the mobile platform is paused and positioned according to the above-mentioned automatic inspection robot positioning method, so as to determine the reference platform tracking coordinate system P m Deflection angle relative to coordinate system P1 And the coordinate system P m The coordinates of the origin in the coordinate system P1 (x m1 s,y m1 s, z m1s ); S2. Determine the target point: While the system is positioning the mobile platform, it will retrieve and return the last normal shooting site number i, thereby determining site N i+1 is the target site and obtains its coordinates in the coordinate system P1; S3. Determine the error: according to the target position N i+1 Coordinates, can determine the correction task target x 10 ,y 10 , are the correct coordinates and deflection angles of the mobile platform at the target point, respectively, and the error values ​​Δx, Δy can be obtained. Δx=x m1s -x 10 (8) Δy=y 1ms -y 10 (9) because but S4. Determine the speed of the mobile platform: set the speed control rate k of the three degrees of freedom respectively x , k y , S5, deviation correction: dx1, dy1, Converted into the three-degree-of-freedom velocity dx of the mobile platform's servo coordinate system m ,dy m , And control the mobile platform to move to the target point; S6. Confirm and shoot: Move to the scheduled site N i+1 After that, perform positioning again. When the error values ​​are all 0, continue shooting and scanning normally.

Images