Part size automatic detection device and method

Abstract

The present invention discloses an automated part size detection device and method, relating to the field of optical detection technology, wherein the device comprises: a linear slide, provided with a slider, the slider being fixed with a first electric angular displacement platform for reciprocating along a first direction, the first electric angular displacement platform being fixed with a second electric angular displacement platform for deflecting around a first direction within a first angular range, the second electric angular displacement platform being fixed with a rotary table for deflecting around a second direction within a second angular range, the rotary table being used to fix the part to be detected and rotate around a third direction, the first direction, the second direction, and the third direction being perpendicular to each other; a laser sensor being adjustable and fixed above the part to be detected, for collecting three-dimensional point cloud data of the part to be detected; and a host computer being electrically connected to control the above structures, obtaining and giving detection results and controlling a sorting robot to sort the part to be detected. The present invention can improve the accuracy of automated part detection and meet the needs of the industrial chain for efficient and intelligent industrial production.

Description

Technical Field

[0001] The present invention belongs to the technical field of optical detection, and relates to an automatic detection device and method for part dimensions. Background Art

[0002] With the continuous development of science and technology, the level of automation and intelligent production in the manufacturing industry continues to increase. Mechanical parts are the foundation of the manufacturing industry. The demand for complex mechanical parts in high-tech fields such as automobiles and aerospace is constantly increasing. However, due to the diverse shapes and lack of regular features of complex mechanical parts, the quality inspection of these parts is crucial to ensure the quality of the parts.

[0003] In related technologies, parts inspection is mainly completed by manual contact measurement methods such as calipers, micrometers, and three-dimensional coordinate measuring machines. This has problems such as low inspection efficiency, high workload, and low precision. Researchers began to try to use three-dimensional scanning technology to detect part precision and proposed their own full-surface non-contact inspection solutions. However, during the part scanning process, because the line laser needs to be perpendicular to the part, it can only scan the target part in a given direction at a time, resulting in low part molding accuracy and missing key feature points. In addition, after the parts are inspected, they need to be sorted. The sorting of transmission parts is done manually by workers, and qualified and unqualified parts are selected from them.

[0004] Due to the inevitable subjective standards of manual sorting, it is difficult to form a unified sorting standard, resulting in low efficiency and poor reliability of parts sorting, which cannot meet the requirements of efficient and accurate production. Summary of the Invention

[0005] To solve the above problems, the first aspect of the present invention provides an automatic part size detection device, including: a linear slide, provided with a slider, the slider is fixed with a first electric angular displacement table, the first electric angular displacement table is fixed with a second electric angular displacement table, and the second electric angular displacement table is fixed with a rotary table, wherein the first electric angular displacement table reciprocates along a first direction, the second electric angular displacement table deflects around the first direction within a first angle range, the rotary table deflects around the second direction within a second angle range, the rotary table is used to fix the part to be detected and rotate around a third direction, the first direction, the second direction and the third direction are perpendicular to each other; a laser sensor, which can be adjusted and fixed above the part to be detected, and is used to collect three-dimensional point cloud data of the part to be detected; a host computer, electrically connected to control the linear slide, the first electric angular displacement table, the second electric angular displacement table, the rotary table and the laser sensor, to obtain three-dimensional data of the part to be detected and give detection results; a sorting robot, electrically connected to the host computer, and used to sort the parts to be detected according to the detection results.

[0006] Preferably, the automatic part size detection device also includes a base and a movable bracket, wherein the linear slide, the movable bracket and the sorting robot are respectively detachably fixed to the top surface of the base; the laser sensor is movably connected to the movable bracket and placed above the rotating table, and the laser sensor moves along the third direction.

[0007] In any of the above schemes, preferably, the base is a precision dual-frequency damping vibration isolation optical platform; the movable bracket includes a support rod arranged along a third direction, one end of the support rod is detachably connected to the base, and two fixed plates are arranged on the support rod at intervals along the third direction, the two fixed plates are respectively rotatably connected to the two ends of the lead screw, the lead screw is threadedly connected to a nut, and the laser sensor is fixed to the nut.

[0008] In any of the above schemes, preferably, the support rod between the two fixed plates is provided with a slide rail along the third direction, and the nut reciprocates along the slide rail; the nut is fixed with an extension plate extending along the first direction, and the laser sensor is connected to the nut through the extension plate.

[0009] In any of the above schemes, preferably, the screw is a ball screw, the two fixed plates are respectively fixed with bearings, and the two ends of the screw are respectively fixed to the inner rings of the bearings and are rotatably connected to the fixed plates.

[0010] In any of the above solutions, preferably, the end of the screw away from the base is connected to a knob, and the slide rail is marked with a scale.

[0011] In any of the above solutions, preferably, the support rod is provided with a plurality of fixing holes along the third direction, and the fixing plate is detachably connected to the support rod along the third direction.

[0012] In any of the above solutions, preferably, a binocular structured light camera is provided at the end of the sorting robot; and the automatic part size detection device further comprises at least one sorting box for accommodating qualified or unqualified parts to be inspected.

[0013] According to a second aspect of the present invention, there is provided a method for automated part size detection, comprising: fixing a part to be detected on a rotating table of an automated part size detection device as described in the first aspect and any one of the preferred embodiments above, and placing the part to be detected directly under a laser sensor by adjusting a slider; adjusting the posture of the part to be detected by a host computer by controlling a first electric angular displacement table, a second electric angular displacement table, and a rotating table; collecting three-dimensional point cloud data of the part to be detected in multiple postures by a laser sensor; constructing a point cloud model of the part to be detected based on the received three-dimensional point cloud data, and providing a detection result by comparing it with a standard model; and controlling a sorting robot to sort the part to be detected based on the detection result.

[0014] Preferably, before constructing the point cloud model of the part to be inspected, the method also includes: projecting the three-dimensional point cloud data to obtain the projection coordinate system and point cloud edge data; comparing the image of the part to be inspected captured by the binocular structured light camera to extract the coordinate information of the missing part; converting the coordinate information of the missing part into the projection coordinate system to determine the three-dimensional point cloud coordinate information of the missing part.

[0015] In any of the above schemes, preferably, the host computer constructs a point cloud model of the part to be inspected based on the received three-dimensional point cloud data, including: supplementing the three-dimensional point cloud coordinate information of the missing parts of the three-dimensional point cloud data to obtain combined three-dimensional point cloud data; using Hausdorff distance to simplify the combined three-dimensional point cloud data; and using principal component analysis to align the simplified combined three-dimensional point cloud data to obtain a point cloud model.

[0016] In any of the above schemes, preferably, obtaining point cloud edge data includes: projecting the three-dimensional point cloud data into a two-dimensional plane; obtaining point cloud edge data based on two-dimensional image edge processing; using two-dimensional image edge extraction, based on the image of the part to be inspected and the point cloud edge data, determining the edge data of the missing part; supplementing the edge data of the missing part of the point cloud edge data to determine the combined point cloud edge data, and the combined three-dimensional point cloud data is obtained by mapping the combined point cloud edge data into a projection coordinate system; wherein, the coordinates of point P on the inspection surface of the part to be inspected with coordinates (x1, y1, z1) are projected onto the plane M: Ax+By+Cz+D=0 as follows:

[0017]

[0018] (x p ,y p , z p ) is the projection point of point P on plane M;

[0019] When C=0 and D=0, plane M is Ax+By=0, and the projection point of point P on plane M is:

[0020]

[0021] In any of the above schemes, preferably, the combined three-dimensional point cloud data is simplified using the Hausdorff distance, including: estimating the principal curvature of all points of the three-dimensional point cloud data using a quadratic parabola fitting method; calculating the Hausdorff distance between a point in the three-dimensional point cloud data and its neighboring points, and taking the maximum value as the Hausdorff value of the point; traversing all points of the three-dimensional point cloud data and calculating the Hausdorff values ​​of all points; dividing the three-dimensional point cloud data into multiple intervals according to the Hausdorff values, and setting a threshold for each interval; traversing all intervals and deleting points of the three-dimensional point cloud data whose Hausdorff values ​​are less than the corresponding threshold.

[0022] In any of the above solutions, preferably, before registering the streamlined combined three-dimensional point cloud data, the method further includes: denoising the point cloud data; and performing fitting and filtering processing based on the principal component combined three-dimensional point cloud data.

[0023] Through the above design, the present invention can achieve at least the following beneficial effects:

[0024] The automatic part size detection device and method of the present invention realizes large-stroke movement of the part to be detected in the first direction by setting a linear slide, increases the operating space for the addition of the robotic arm in the sorting mechanism, and facilitates the implementation of the sorting operation; by adjusting the first electric angular displacement table, the second electric angular displacement table and the rotating table, the purpose of adjusting the part posture is achieved, so that the detection surface is always kept in the vertical direction of the laser sensor light beam, which is convenient for obtaining three-dimensional point cloud data with more complete data, improves the forming efficiency and scanning accuracy of the part to be detected, makes the overall outline of the part to be detected clearer, and is conducive to the judgment and analysis of the error of the part to be detected; at the same time, the setting of the sorting robot can not only obtain the three-dimensional coordinates of the point cloud missing on the part surface, but also automatically sort the parts to be detected, thereby improving the accuracy of the fully automated detection and sorting of parts, and meeting the needs of the industrial chain for efficient and intelligent industrial production. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a structural diagram of an embodiment of an automatic detection device for part dimensions of the present invention;

[0026] Figure 2 This is a schematic diagram of the assembly structure of the linear slide, the first electric angular displacement stage, the second electric angular displacement stage, and the rotary stage;

[0027] Figure 3 It is a structural diagram of an embodiment of a movable bracket;

[0028] Figure 4 This is a schematic diagram of an embodiment of a laser sensor operating state;

[0029] Figure 5 This is a flow chart of an embodiment of a method for automatically detecting part dimensions according to the present invention;

[0030] Figure 6 It is a flow chart of another embodiment of the automatic detection method of part dimensions of the present invention.

[0031] Description of the numbers in the figure:

[0032] 1-Linear slide; 11-Slider; 2-First electric angular displacement stage; 3-Second electric angular displacement stage; 4-Rotary stage; 5-Laser sensor; 6-Upper computer; 7-Sorting robot; 8-Base; 9-Movable bracket; 91-Support rod; 911-Fixing hole; 92-Fixing plate; 93-Lead screw; 94-Nut; 95-Slide rail; 96-Knob; 97-Extension plate; 10-Sorting box. DETAILED DESCRIPTION

[0033] To make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0034] In addition, in the description of the present invention, the terms "longitudinal", "transverse", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the present invention and do not require that the present invention must be constructed and operated in a specific orientation. Therefore, they should not be understood as limitations on the present invention. The terms "connected" and "connected" used in the present invention should be understood in a broad sense. For example, they can be fixed connections or detachable connections; they can be directly connected or indirectly connected through intermediate components. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0035] The first aspect of the embodiments of the present invention aims to solve the problems existing in the above-mentioned related technologies and provide an automated device for part size detection to improve the detection efficiency of part size detection, reduce work intensity, and improve the accuracy of part size detection.

[0036] Figure 1 It is a structural diagram of an embodiment of an automatic detection device for part dimensions of the present invention; Figure 2 This is a schematic diagram of the assembly structure of the linear slide, the first electric angular displacement stage, the second electric angular displacement stage, and the rotary stage; Figure 3 It is a structural diagram of an embodiment of a movable bracket; Figure 4 FIG. 1 is a schematic diagram of an embodiment of the laser sensor operation state. Figures 1 to 4As shown, the automated part size detection device of this embodiment may include a linear slide 1, a laser sensor 5, a host computer 6, and a sorting robot 7. The linear slide 1 may be provided with a slider 11, to which a first electric angular displacement stage 2 may be fixed. The first electric angular displacement stage 2 may be fixed with a second electric angular displacement stage 3, to which a rotary table 4 may be fixed. The first electric angular displacement stage 2 may reciprocate along a first direction, the second electric angular displacement stage 3 may deflect about a first direction within a first angular range, and the rotary table 4 may deflect about a second direction within a second angular range. The rotary table 4 may be used to fix the part to be detected and rotate about a third direction, wherein the first direction, the second direction, and the third direction may be perpendicular to each other. The laser sensor 5 may be adjustable and fixed above the part to be detected, for collecting three-dimensional point cloud data of the part to be detected. The host computer 6 may electrically connect and control the linear slide 1, the first electric angular displacement stage 2, the second electric angular displacement stage 3, the rotary table 4, and the laser sensor 5 to obtain three-dimensional data of the part to be detected and provide detection results. The sorting robot 7 can be electrically connected to the host computer 6, and the sorting robot 7 can be used to sort the parts to be inspected according to the inspection results.

[0037] In this embodiment, the first direction, the second direction, and the third direction are perpendicular to each other and can be equivalent to the X-axis, Y-axis, and Z-axis of the three-dimensional rectangular coordinate system. The linear slide 1 can be a linear guide rail arranged along the X-axis. The slider 11 is slidably connected to the linear slide 1, and the slider 11 can be used to fix the first electric angular displacement stage 2 deflected around the X-axis.

[0038] The first electric angular translation stage 2 is a type of instrumentation used on an optical platform. It is primarily used to position and package coupled optical instruments. It can achieve positioning and precision adjustment functions by moving in different directions. The spacing and stroke of position movement are determined based on the coarse and fine accuracy of the micrometer head. It is a device used to rotate objects within a small angular range, rotating around a fixed point in space. Similar to a linear translation stage, it does not provide linear movement, but instead rotates around a fixed point above the stage's mounting surface. The first electric angular translation stage 2 of this embodiment deflects around the X-axis within a set angular range, which can be ±15°.

[0039] The first electric angular translation stage 2 can include a base with a concave arc surface, on which a drive motor and controller are fixedly mounted. The base can also be engraved with arc scale lines on its upper portion. A sliding stage is provided at the upper end of the base, and the lower end surface of the sliding stage is convexly arc-shaped, which mates with the concave arc surface of the base. During the rotation of the sliding stage, its fixed pointer points to the arc scale lines on the base, allowing the operator to observe the angular value of the rotation of the part to be inspected around the X-axis. The first electric angular translation stage 2 is driven by a precision-ground worm gear, which provides high positioning accuracy. It adopts a curved guide rail, has a strong load capacity, and is equipped with a standard two-phase stepper motor with a self-locking function, ensuring stable and reliable performance.

[0040] In this embodiment, the second electric angular displacement stage 3 can be of the same or similar structure as the first electric angular displacement stage 2, except that the sliding stage of the second electric angular displacement stage 3 deflects around the Y-axis, i.e., the second direction. The base of the second electric angular displacement stage 3 can be detachably fixed to the sliding stage of the first electric angular displacement stage 2, and the second electric angular displacement stage 3 can rotate as a whole around the X-axis as the sliding stage of the first electric angular displacement stage 2 rotates. The linear slide 1 can be a commercially available linear slide, including a linear track on which a slider 11 is provided, and the slider 11 is driven by an electric motor to move along the linear track. The first electric angular displacement stage 2 and the second electric angular displacement stage 3 can be PSAG series precision electric angular displacement stages produced by Zhuoli Hanguang Company, which can rotate + / - 15 degrees around the first direction and around the second direction. They are driven by a precision-ground worm gear, have high positioning accuracy, use arc-shaped V-shaped guide rails, have strong load capacity, and are equipped with a standard two-phase stepper motor with a self-locking function, and have stable and reliable performance.

[0041] Rotary table 4 is removably attached to the sliding table of second electric angular table 3, and the part to be inspected can be removably fixed to rotary table 4. Therefore, the part to be inspected can be deflected about the X-axis by first electric angular table 2, deflected about the Y-axis by second electric angular table 3, and rotated about the Z-axis by rotary table 4, thereby adjusting the position of the part to be inspected. Rotary table 4 can be an ultra-high-precision electric rotary table with a built-in imported ultra-high-precision circular grating, providing ultra-high closed-loop resolution. It uses a precision-ground worm gear drive for ultra-high positioning performance, and uses imported ultra-high-grade cross-roller rings for ultra-high motion performance. It also comes standard with a two-phase stepper motor for stable and reliable performance.

[0042] In this embodiment, the laser sensor 5 can be used to collect three-dimensional point cloud data of the parts to be inspected. It can be a high-speed contour measuring instrument with an ultra-fast speed of 64,000 contours per second. It can simultaneously measure materials with different reflective intensities and has high speed, high precision and high stability.

[0043] like Figure 4As shown, the laser sensor 5 projects a laser beam onto the part to be inspected. The laser beam is reflected and received by the filter. After passing through the lens group, the beam is finally projected onto the surface of the photosensitive film. The laser sensor 5 uses triangulation to measure distance. The laser beam forms a triangle between the laser projector, the part to be inspected, and the photosensitive film. Based on the position of the reflected laser beam, the spatial position and actual physical size of the surface of the part to be inspected illuminated by the laser can be determined. The triangulation distance measurement method has a simple structure and strong practicality. It is often used in medium and short distance measurements. At this time, the advantages of the laser directionality will be highlighted. After the laser beam reaches the surface of the part to be inspected, part of the scattered light returns to the filter and is received. If the surface of the part to be inspected moves, the imaging light spot on the photosensitive film will also move accordingly. The two form a similar triangle relationship. Therefore, the movement of the surface of the part to be inspected can be calculated by measuring the movement of the light spot on the photosensitive film.

[0044] The sorting robot 7 can be equipped with a binocular structured light camera at its end. The sorting robot 7 is capable of multi-degree-of-freedom movement and can be positioned directly in front of or to the side of the linear slide 1. The automated part size detection device of this embodiment can also include at least one sorting box 10, respectively for accommodating qualified and unqualified parts to be inspected. Based on the inspection results, the sorting robot 7 removes the parts to be inspected from the rotating table 4 and places them in the next operating station or sorting box 10. The sorting boxes 10 can also be placed on both sides of the sorting robot 7, with "qualified" and "unqualified" labels affixed to the surface of the sorting boxes 10. In this way, the host computer 6 sends a pass / fail instruction code to the sorting robot 7. When the parts to be inspected are transported from the linear slide 1 to the designated position after a fixed time, they have moved out of the detection range of the laser sensor 5. The timer transmits a command to the host computer 6, which then sends a command to the sorting robot 7. The sorting robot 7 then grabs the parts to be inspected and places them in the designated sorting box 10 through teaching.

[0045] In some embodiments, the automatic part size detection device may further include a base 8 and a movable bracket 9, wherein the linear slide 1, the movable bracket 9 and the sorting robot 7 are respectively detachably fixed to the top surface of the base 8; the laser sensor 5 is movably connected to the movable bracket 9 and is placed above the rotating table 4, and the laser sensor 5 moves along the third direction. The host computer 6 can be connected to the corresponding controllers of the linear slide 1, the first electric angular displacement table 2, the second electric angular displacement table 3, the rotating table 4, the laser sensor 5, the sorting robot 7, etc. by wireless transmission or wired means. If it is a wired connection method, a wire groove can also be opened on the base 8 to ensure that the surface of the base 8 is flat and avoid protrusions. The host computer 6 can be placed on the base 8 or in other convenient places for operation. This embodiment is not specifically limited here.

[0046] In this embodiment, the movable bracket 9 can be fixed on the base 8 at a position close to one end of the linear slide 1, the sorting robot 7 can be fixed on the base 8 at a position close to the other end of the linear slide 1, or the sorting robot 7 can be fixed on one side of the linear slide 1 on the base 8. This embodiment is not specifically limited here. Its purpose is to leave room for movement for the sorting robot 7 according to actual trial conditions, and to leave operating space for placing the parts to be inspected on the rotating table 4.

[0047] In the specific use of the automatic part size detection device of this embodiment, the first electric angular displacement table 2 can be slid along the linear slide 1 to one end position, and then the part to be detected can be placed on the rotating table 4. The first electric angular displacement table 2 is then moved along the linear slide 1 so that the part to be detected on the rotating table 4 is aligned with the laser sensor 5. The host computer 6 controls and adjusts the first electric angular displacement table 2, the second electric angular displacement table 3, the rotating table 4, the laser sensor 5, and the sorting robot 7 to realize the collection of three-dimensional point cloud data of the part to be detected. After the part to be detected is fixed on the rotating table 4 by the clamp, the first electric angular displacement table 2 and the second electric angular displacement table 3 drive the rotating table 4 to rotate about the first direction and the second direction, and the rotating table 4 itself drives the part to be detected to rotate about the third direction, so that the laser sensor 5 can perform a full-range scan of the part to be detected, realizing the collection of its three-dimensional point cloud data. The host computer 6 reconstructs the 3D model of the part to be inspected based on the collected 3D point cloud data, compares it with the CAD standard model pre-stored in the host computer 6, outputs the inspection results, and controls the sorting robot 7 to perform sorting operations based on the inspection results, thereby completing the automated inspection and sorting of the size of the parts to be inspected.

[0048] In order to increase the accuracy of the three-dimensional point cloud model, in some embodiments, a binocular structured light camera can be configured at the end of the sorting robot 7. The binocular camera builds a multi-dimensional visual inspection system, adopts a method of combining two-dimensional images with three-dimensional point cloud depth information, and the host computer 6 projects the point cloud model data obtained by the line laser emitted by the laser sensor 5. Through the two-dimensional image processing method, the host computer 6 extracts the edge of its projection, and then converts the point cloud projection coordinate system to the pixel coordinates of the image of the part to be inspected, while keeping the projection size consistent with the workpiece size in the image. At this time, a binocular structured light camera is used to capture the overall shape image of the part, extract the coordinate information of the missing point part, and convert it to the point cloud projection coordinate system to calculate its depth information, and finally obtain the missing point cloud three-dimensional coordinate information, thereby obtaining a more accurate part model, which is compared with the CAD standard model of the part to be inspected by the host computer 6, and the detection result is output. The sorting robot 7 is controlled to perform sorting operations according to the detection result, thereby completing the automated detection and sorting of the size of the part to be inspected.

[0049] In this embodiment, the position of the part to be inspected in the first direction is adjusted by the linear slide 1, the angle of the part to be inspected around the first direction is adjusted by the first electric angular displacement stage 2, the angle of the part to be inspected around the second direction is adjusted by the second electric angular displacement stage 3, and the angle of the part to be inspected around the third direction is adjusted by the rotary stage 4. The posture of the part to be inspected is adjusted from multiple angles and in all directions, which facilitates the real-time scanning and detection by the laser sensor 5, improves the forming efficiency and scanning accuracy of the part to be inspected, makes the overall outline of the part to be inspected clearer, and is conducive to the judgment and analysis of the error of the part to be inspected. The sorting robot 7 can be a multi-degree-of-freedom robot arm, which does not need to consider the requirements of the fixture for the shape and size of the object, and can meet the needs of most industrial parts. The use of the part size automatic detection device of this embodiment can realize the full automation of part detection and sorting, which meets the requirements of the industrial chain for efficient and intelligent industrial production.

[0050] In some embodiments, the base 8 can be a precision dual-frequency damping vibration isolation optical platform, which is a three-layer sandwich honeycomb structure and is composed of a table, a bracket, a dual-frequency damping vibration isolation mechanism, a height adjustment mechanism, and silent casters with brakes. The interior of the table can be a three-layer sandwich honeycomb structure made of ferromagnetic stainless steel with excellent corrosion resistance; the dual-frequency damping vibration isolation mechanism is located under the table, in the middle of the table and the bracket, which has a vibration isolation effect; the bracket adopts an integral welding process, is in the shape of a long hair, and has a four-support structure. The four brackets are connected by two sections of trusses, which has good rigidity and stability. There are silent casters with height adjustment mechanisms and brakes under the bracket. The bottom of the height adjustment mechanism is oblate under each support leg, which increases the contact area with the ground. By adjusting the upper and lower distances of the height adjustment mechanism, the problem of bracket distortion and deformation caused by uneven ground can be solved. The silent casters are located below the lower truss and are connected to the truss with four bolts, which facilitates movement and handling. This precision dual-frequency damping vibration isolation optical platform provides good rigidity and vibration isolation performance for parts detection and sorting.

[0051] In some embodiments, the movable bracket 9 includes a support rod 91 arranged along the third direction. One end of the support rod 91 is detachably connected to the base 8. Two fixing plates 92 are spaced apart on the support rod 91 along the third direction. The two fixing plates 92 are rotatably connected to the ends of a lead screw 93, which is threadedly connected to a nut 94. The laser sensor 5 is fixed to the nut 94. The nut 94 is supported by the support rod 91, limiting the nut's rotation in the third direction. In this way, during the rotation of the lead screw 93, the nut 94 drives the laser sensor 5 to reciprocate up and down along the third direction, thereby achieving the purpose of adjusting the distance between the laser sensor 5 and the part to be detected.

[0052] Specifically, the support rod 91 between the two fixed plates 92 can be provided with a slide rail 95 extending in the third direction, along which a nut 94 reciprocates. An extension plate 97 extending in the first direction is fixed to the nut 94, and the laser sensor 5 is connected to the nut 94 via the extension plate 97. The lead screw 93 can be a ball screw, with bearings fixed to each of the two fixed plates 92. The ends of the lead screw 93 are fixed to the inner rings of the bearings, rotatably connected to the fixed plates 92. The bearings reduce friction between the lead screw 93 and the fixed plates 92, making the operation of rotating the lead screw 93 smoother.

[0053] In other embodiments, a knob 96 may be connected to the end of the lead screw 93 away from the base 8, and a scale may be marked on the slide rail 95. An operator can operate the knob 96 to rotate the lead screw 93. By rotating the knob 96 in a third direction, the nut 94 can drive the laser sensor 5 to adjust the distance between the laser sensor 5 and the part to be inspected. The scale on the slide rail 95 can accurately determine the specific distance between the part to be inspected and the laser sensor 5. A circular opening can be provided in the center of fixed plate 92, housing a bearing. The ends of lead screw 93 are then tightly fitted with the inner ring of the bearing, with one of the longer ends extending through fixed plate 92 and fixedly connected to knob 96. Nut 94 is connected to extension plate 97, which is arranged along a first direction and on which laser sensor 5 is mounted. Nut 94 can be mounted on a slide rail 95, which can be arranged parallel to lead screw 93 along a third direction to ensure stable operation of extension plate 97 and laser sensor 5. A scale can also be provided around knob 96, proportional to the travel of lead screw 93. This allows manual rotation of knob 96, which in turn drives the lead screw nut and the connected laser sensor 5 along the third direction, to easily determine the adjusted distance of the laser sensor. The travel of lead screw 93 is approximately 150 mm. In actual testing, component heights vary, and the operating distance of laser sensor 5 has a certain range. The design of movable bracket 9 facilitates adjustment of the height of laser sensor 5 according to actual working conditions, ensuring that it remains within the appropriate operating range.

[0054] Furthermore, in some embodiments, the support rod 91 is provided with multiple sets of fixing holes 911 along the third direction, and the fixing plate 92 is detachably connected to the support rod 91 along the third direction. The approximate position of the fixing plate 92 on the support rod 91 can be adjusted according to the specific height of the part to be inspected along the third direction, and then fine-tuned through the cooperation of the lead screw 93 and the nut 94. This can expand the applicability of the automatic part size inspection device of this embodiment to parts of different sizes to be inspected.

[0055] Based on the same or similar design concept, a second aspect of an embodiment of the present invention provides a method for automatically detecting part dimensions. Figure 5FIG. 1 is a flow chart of an embodiment of a method for automatically detecting part dimensions according to the present invention. Figure 5 As shown, the method may include the following steps:

[0056] S1: Fix the part to be inspected on the rotating table 4 of the automatic part size inspection device of any one of the first aspect and preferred embodiments described above, and place the part to be inspected directly under the laser sensor 5 by adjusting the slider 11;

[0057] S2: The host computer 6 controls the first electric angular displacement stage 2, the second electric angular displacement stage 3, and the rotary stage 4 to adjust the posture of the part to be inspected so that the inspection surface faces the laser sensor 5.

[0058] S3: The laser sensor 5 collects three-dimensional point cloud data of the part to be inspected in multiple positions;

[0059] S4: The host computer 6 constructs a point cloud model of the part to be inspected based on the received three-dimensional point cloud data, and compares it with the standard model to give the inspection result;

[0060] S5: The host computer 6 controls the sorting robot 7 to sort the parts to be inspected based on the inspection results.

[0061] The automated part size detection method of this embodiment uses a clamp to fix the part to be inspected on the rotating table 4 of the automated part size detection device involved in any of the aforementioned embodiments, and then transports the part to be inspected to directly below the laser sensor 5 through the linear slide 1, and then adjusts the relative position of the part to be inspected and the laser sensor 5 through the movable bracket 9; adjusts the posture of the part to be inspected by adjusting the first electric angular displacement table 2, the second electric angular displacement table 3 and the rotating table 4, and at the same time uses the laser sensor 5 to complete the collection of three-dimensional point cloud data of the part to be inspected, and then uses the linear slide 1 to move the part to be inspected to a position convenient for the sorting robot 7 to grasp, and uses the sorting robot 7 to complete the sorting of the part to be inspected.

[0062] First, determine the surface to be inspected for the part to be inspected, and use a fixture to place the part to be inspected on the rotating table 4 facing the laser sensor 5. The part to be inspected can be a separate part or a simply assembled component. In order to create a good measurement environment for the part to be inspected, the distance between the laser sensor 5 and the part to be inspected can be adjusted by turning the knob 96. By adjusting the first electric angular displacement table 2, the second electric angular displacement table 3 and the rotating table 4, the position of the part to be inspected facing the laser sensor can be changed, so that the laser sensor 5 is facing the surface required to be measured of the part to be inspected, and a more suitable starting point is selected.

[0063] Then, the host computer 6 sets a program to control the rotation of the first electric angular displacement stage 2, the second electric angular displacement stage 3, and the rotating table 4, driving the movement of the part to be inspected. The laser sensor 5 is used to complete the collection of three-dimensional point cloud data of the part to be inspected set on the rotating table 4. In addition, for the structure on one side of the part to be inspected that is fixed to the rotating table 4, the laser sensor 5 cannot collect the relevant three-dimensional point cloud data on this side during the above-mentioned posture change process. In order to complete the three-dimensional point cloud data of the part to be inspected collected by the laser sensor 5, the part to be inspected must be re-fixed on the rotating table 4, and the side of the part to be inspected that has not been collected with the three-dimensional point cloud data is exposed so that the structure on this side can be collected by the laser sensor 5. The above process is repeated to complete the collection of three-dimensional point cloud data for other surfaces, and finally the collection of the entire part to be inspected is completed. The host computer 6 performs fitting and deduplication operations on the three-dimensional point cloud data collected each time to generate point cloud data for each part of the part to be inspected.

[0064] The host computer 6 uses the Hausdorff distance to simplify the three-dimensional point cloud data of the inspected part, adopts the principal component analysis method to perform point cloud registration, realizes accurate reconstruction of the point cloud data of the inspected part, and constructs its 3D model. The Hausdorff distance is used to simplify the point cloud data. The Hausdorff distance is a definition of the distance between two point sets, which is used to describe the similarity between two motors. The Hausdorff distance between two finite point sets A and B is defined as: D(A,B) = max(d(A,B), d(B,A)), where d(A,B) and d(B,A) represent the one-way Hausdorff distance from set A to set B and from set B to set A, respectively, and are defined as follows:

[0065]

[0066] Among them, ‖ab‖ and ‖ba‖ represent the distance norm. Let the data point set be {p i , i=1,2,3....n}, take any point p, its neighbor point set is {q i, i = 1, 2...k}. The algorithm for calculating the Hausdorff distance of each sampling point and simplifying the data points is as follows: Step 1: Estimate the principal curvature of all points using a quadratic parabola fitting method. Step 2: Calculate the Hausdorff distance between point p and its neighboring points, and take the maximum value as the Hausdorff value of that point. Step 3: Traverse all points using Step 2 to calculate the Hausdorff value of all data points. Step 4: Divide the point cloud into multiple intervals based on the Hausdorff value of the data points, setting a different threshold ε for each interval. Step 5: For any point within a curvature interval whose Hausdorff value is less than the threshold ε, delete it. Step 6: Traverse all intervals to complete the simplification. Using the Hausdorff distance to simplify the point cloud ensures that sufficient points are retained in areas with large curvature changes to highlight the surface features of the model, while retaining a small number of points in areas with small curvature changes to reduce redundancy. This effectively reduces the number of data points while well preserving the detailed feature information of the model. When using principal component analysis to align point clouds, it is necessary to denoise the point cloud data, perform fitting based on the principal component point cloud, and then perform filtering.

[0067] Before registering the streamlined combined three-dimensional point cloud data, in some embodiments, the method for automated part size detection may further include: denoising the point cloud data; and performing fitting and filtering processing based on the principal component combined three-dimensional point cloud data.

[0068] The process of denoising point cloud data includes:

[0069] Obtain the mean value of the point cloud coordinates sliding within a certain radius window (o x , o y , o z ), find its covariance matrix

[0070]

[0071] Among them, (x i ,y i , z i ) are the coordinates of each point in the window. Find the two main directions of the matrix is the main direction of the window point cloud, then the point cloud noise direction for:

[0072]

[0073] Fit the local plane in the main direction and find the value of each point (x i ,y i , z i ) and the plane along the noise direction distance

[0074] Each point is displaced in the noise direction, and the displacement distance d is:

[0075]

[0076] Here, α is a coefficient. This step can remove the noise caused by large outliers.

[0077] The fitting process based on the principal component point cloud includes:

[0078] The equation for further fitting the point cloud is:

[0079]

[0080] Where i, j = 0, 1, 2, ..., n, a ij is the coefficient.

[0081] The error function is defined as:

[0082]

[0083] Substitute all points and find the coefficient by minimizing the overall error:

[0084]

[0085] The optimal a can be obtained ij .

[0086] The above formula is used to process each window point cloud. When the window traverses all point clouds, the overall point cloud surface is obtained.

[0087] The filtering processing formula is:

[0088]

[0089] in, is the distance from the point to the surface, x, y, z are coordinate values, σ x,y,z is the standard deviation within the window, and its calculation formula is:

[0090]

[0091] The window size is (2M+1)×(2M+1), and the M value calculation formula is:

[0092] M=k×f

[0093] Among them, K is a given coefficient, f is the fluctuation of the point cloud within the range, that is, the mean of the absolute value of the distance from each point to the surface,

[0094] f=mean(d i )

[0095] Among them, d i The principal component analysis method is the absolute value of the distance from each point to the surface. It can eliminate large noises and effectively retain the sharpness, edge features and detail features in the model.

[0096] Figure 6 FIG. 1 is a flow chart of another embodiment of the method for automatically detecting part dimensions according to the present invention. Figure 6 As shown, in some embodiments, before constructing the point cloud model of the part to be inspected, the method further includes:

[0097] S31: Projecting the three-dimensional point cloud data to obtain a projection coordinate system and point cloud edge data;

[0098] S32: extracting coordinate information of the missing part by comparing the image of the part to be inspected acquired by the binocular structured light camera;

[0099] S33: Convert the coordinate information of the missing part into a projection coordinate system to determine the three-dimensional point cloud coordinate information of the missing part.

[0100] In some embodiments, the host computer 6 constructs a point cloud model of the part to be inspected based on the received three-dimensional point cloud data, including: supplementing the three-dimensional point cloud coordinate information of the missing parts of the three-dimensional point cloud data to obtain combined three-dimensional point cloud data; using Hausdorff distance to simplify the combined three-dimensional point cloud data; and using principal component analysis to align the simplified combined three-dimensional point cloud data to obtain a point cloud model.

[0101] In some embodiments, obtaining point cloud edge data includes: projecting the three-dimensional point cloud data into a two-dimensional plane; obtaining point cloud edge data based on two-dimensional image edge processing; using two-dimensional image edge extraction, based on the image of the part to be inspected and the point cloud edge data, determining the edge data of the missing portion; supplementing the edge data of the missing portion of the point cloud edge data to determine the combined point cloud edge data, wherein the combined three-dimensional point cloud data is obtained by mapping the combined point cloud edge data into a projected coordinate system; wherein the coordinates of point P with coordinates (x1, y1, z1) on the inspection surface of the part to be inspected are projected onto the plane M: Ax+By+Cz+D=0 as follows:

[0102]

[0103] (x p ,y p , z p ) is the projection point of point P on plane M;

[0104] When C=0 and D=0, plane M is Ax+By=0, and the projection point of point P on plane M is:

[0105]

[0106] In some embodiments, the combined three-dimensional point cloud data is simplified using the Hausdorff distance, including: estimating the principal curvature of all points of the three-dimensional point cloud data using a quadratic parabola fitting method; calculating the Hausdorff distance between a point in the three-dimensional point cloud data and its neighboring points, and taking the maximum value as the Hausdorff value of the point; traversing all points of the three-dimensional point cloud data and calculating the Hausdorff values ​​of all points; dividing the three-dimensional point cloud data into multiple intervals according to the Hausdorff values, and setting a threshold for each interval; traversing all intervals and deleting points of the three-dimensional point cloud data whose Hausdorff values ​​are less than the corresponding threshold.

[0107] The linear slide 1 is controlled to drive the parts to be detected set on the rotating table 4 to move out of the position facing the laser sensor 5, and the sorting operation is performed with the help of the sorting robot 7. The sorting robot 7 has a teaching function. The program is edited on the teach pendant. The sorting robot 7 executes the program code taught by the teach pendant. A clamp is installed at the front end, and the grabbing and placing actions of the parts can be completed through fixed-point teaching. The sorting robot 7 receives the script instruction string sent by the host computer through a specific programming interface, and runs the received script instruction to complete the sorting action of the parts. The host computer 6 establishes a TCP / IP connection with the controller of the sorting robot 7 and obtains the status information of the sorting robot 7.

[0108] Due to the diverse shapes of the parts to be inspected, the 3D point cloud model generated by the line laser sensor 5 alone may contain missing points at certain angles. To increase the accuracy of the 3D point cloud model, a binocular structured light camera is installed at the end of the sorting robot 7. This binocular structured light camera forms a multi-dimensional visual inspection system. The host computer 6 first projects the point cloud model data acquired by the laser sensor 5, extracts the edges of the projection through 2D image processing, and then converts the point cloud projection coordinate system to the pixel coordinates of the image of the part to be inspected, while keeping the projection size consistent with the workpiece size in the image. At this time, the binocular structured light camera is used to capture the overall shape image of the part, extract the coordinate information of the missing points, and convert them to the point cloud projection coordinate system to calculate their depth information, ultimately obtaining the missing 3D coordinate information of the point cloud.

[0109] In combination with the automated part size detection device of any of the aforementioned embodiments, the automated part size detection method of this embodiment can utilize the linear slide 1 to achieve a large range of motion in the first direction, thereby increasing the operating space for the sorting robot 7 and facilitating the sorting operation. By adjusting the first electric angular displacement stage 2, the second electric angular displacement stage 3, and the rotary stage 4, the position of the part to be inspected is adjusted, ensuring that the measured portion of the part to be inspected always remains perpendicular to the light beam, facilitating the acquisition of more complete three-dimensional point cloud data, improving the molding efficiency and scanning accuracy of the part to be inspected, and making the overall outline of the part to be inspected clearer, which facilitates the judgment and analysis of errors in the part to be inspected. At the same time, due to the diverse shapes of parts, the three-dimensional point cloud model generated by a line laser scanner alone may have missing point clouds at certain angles. In order to increase the accuracy of the three-dimensional point cloud model, a binocular structured light camera is installed at the end of the sorting robot 7. The binocular structured light camera builds a multi-dimensional visual inspection system, which uses a two-dimensional image combined with three-dimensional point cloud depth information to obtain the three-dimensional coordinates of the missing point clouds on the part surface. There is no need to consider the requirements of the fixture for the shape and size of the object. It can meet the needs of most industrial parts to be inspected, and can realize full automation of parts detection and sorting, which meets the needs of the industrial chain for efficient and intelligent industrial production.

[0110] It is understood from common technical knowledge that the present invention may be implemented by other embodiments that do not deviate from its essence or essential features. Therefore, the embodiments disclosed above are merely illustrative in all respects and are not intended to be exclusive. All modifications within the scope of the present invention or equivalent to the scope of the present invention are intended to be encompassed by the present invention.

Claims

1. A device for automatically detecting part dimensions, characterized in that: include: A linear slide (1) is provided with a slider (11), the slider (11) is fixed with a first electric angular displacement platform (2), the first electric angular displacement platform (2) is fixed with a second electric angular displacement platform (3), and the second electric angular displacement platform (3) is fixed with a rotary platform (4), wherein the first electric angular displacement platform (2) reciprocates along a first direction through the slider (11), the first electric angular displacement platform (2) comprises a base with a concave arc surface, the upper end of the base is provided with a sliding platform, the lower end surface of the sliding platform is a convex arc shape matching the concave arc surface, and the sliding platform drives its fixed pointer to point to the arc scale line on the base during rotation, the first electric angular displacement platform (2) deflects around the first direction within a first angle range, the second electric angular displacement platform (3) deflects around the second direction within a second angle range, and the rotary platform (4) is used to fix the part to be detected and rotate around a third direction, and the first direction, the second direction and the third direction are perpendicular to each other; A laser sensor (5) is adjustable and fixed above the part to be detected, and is used to collect three-dimensional point cloud data of the part to be detected; A host computer (6) electrically connected to control the linear slide (1), the first electric angular displacement table (2), the second electric angular displacement table (3), the rotary table (4), and the laser sensor (5) to obtain three-dimensional data of the part to be inspected and provide inspection results; A sorting robot (7), electrically connected to the host computer (6), for sorting the parts to be inspected according to the inspection results; The end of the sorting robot (7) is equipped with a binocular structured light camera; The host computer (6) projects the point cloud model data obtained by the line laser emitted by the laser sensor (5), extracts the edge of the projection by a two-dimensional image processing method, and then converts the point cloud projection coordinate system into the pixel coordinates of the image of the part to be detected, while keeping the projection size consistent with the workpiece size in the image, uses the binocular structured light camera to shoot the overall shape image of the part to be detected, extracts the coordinate information of the missing point part, converts it into the point cloud projection coordinate system to obtain its depth information, and finally obtains the missing point cloud three-dimensional coordinate information to obtain the part model, compares the part model with the CAD standard model of the part to be detected of the host computer (6), and outputs the detection result.

2. The automatic part size detection device according to claim 1, characterized in that: The automatic part size detection device further comprises a base (8) and a movable bracket (9), wherein the linear slide (1), the movable bracket (9) and the sorting manipulator (7) are respectively detachably fixed to the top surface of the base (8); The laser sensor (5) is movably connected to the movable bracket (9) and is placed above the rotating platform (4), and the laser sensor (5) moves along a third direction.

3. The automatic part size detection device according to claim 2, characterized in that: The base (8) is a precision dual-frequency damping vibration isolation optical platform; The movable bracket (9) includes a support rod (91) arranged along the third direction, one end of the support rod (91) is detachably connected to the base (8), and two fixing plates (92) are spaced apart on the support rod (91) along the third direction, the two fixing plates (92) are rotatably connected to the two ends of a lead screw (93), the lead screw is threadedly connected to a nut (94), and the laser sensor (5) is fixed to the nut (94).

4. The automatic part size detection device according to claim 3, characterized in that: The support rod (91) between the two fixing plates (92) is provided with a slide rail (95) along the third direction, and the nut (94) reciprocates along the slide rail (95); An extension plate (97) extending along the first direction is fixed to the nut (94), and the laser sensor (5) is connected to the nut (94) via the extension plate (97).

5. A method for automatic inspection of part dimensions, characterized in that: The steps include: Fixing the part to be inspected on the rotating table (4) of the automatic part size inspection device according to any one of claims 1 to 4, and placing the part to be inspected directly below the laser sensor (5) by adjusting the slider (11); The first electric angular displacement stage (2), the second electric angular displacement stage (3) and the rotating stage (4) are controlled by the host computer (6) to adjust the posture of the part to be inspected; The laser sensor (5) collects three-dimensional point cloud data of the part to be detected in multiple positions; The host computer (6) constructs a point cloud model of the part to be inspected based on the received three-dimensional point cloud data, and compares it with a standard model to provide a detection result; The host computer (6) controls the sorting robot (7) to sort the parts to be inspected based on the inspection results; Before constructing the point cloud model of the part to be inspected, the method further includes: Projecting the three-dimensional point cloud data to obtain a projection coordinate system and point cloud edge data; Comparing the image of the part to be inspected with the binocular structured light camera configured at the end of the sorting robot (7), the coordinate information of the missing part is extracted; Converting the coordinate information of the missing part into the projection coordinate system to determine the three-dimensional point cloud coordinate information of the missing part; The host computer (6) constructs a point cloud model of the part to be inspected based on the received three-dimensional point cloud data, including: Supplementing the three-dimensional point cloud coordinate information of the missing part to the three-dimensional point cloud data to obtain combined three-dimensional point cloud data; Using Hausdorff distance to simplify the combined three-dimensional point cloud data; The principal component analysis method is adopted to align the simplified combined three-dimensional point cloud data to obtain the point cloud model.

6. The method according to claim 5, characterized in that The acquiring of the point cloud edge data includes: Projecting the three-dimensional point cloud data onto a two-dimensional plane; Acquiring the point cloud edge data based on two-dimensional image edge processing; Determine the edge data of the missing part based on the image of the part to be detected and the edge data of the point cloud using two-dimensional image edge extraction; Supplementing the missing edge data of the point cloud edge data to determine combined point cloud edge data, wherein the combined three-dimensional point cloud data is obtained by mapping the combined point cloud edge data into the projection coordinate system; in, The coordinates of point P on the inspection surface of the part to be inspected, whose coordinates are (x1, y1, z1), are projected onto plane M: Ax+By+Cz+D=0: , (x p ,y p , z p ) is the projection point of point P on plane M; When C=0 and D=0, plane M is Ax+By=0, and the projection point of point P on plane M is: 。 7. The method according to claim 5, characterized in that The simplifying of the combined three-dimensional point cloud data by using the Hausdorff distance includes: estimating the principal curvatures of all points of the three-dimensional point cloud data using a quadratic parabola fitting method; Calculate the Hausdorff distance between a point in the three-dimensional point cloud data and its neighboring points, and take the maximum value as the Hausdorff value of the point; Traversing all points of the three-dimensional point cloud data and calculating the Hausdorff values ​​of all points; According to the Hausdorff value, the three-dimensional point cloud data is divided into multiple intervals, and a threshold is set for each interval; All intervals are traversed, and points of the three-dimensional point cloud data whose Hausdorff values ​​are less than the corresponding threshold are deleted.

8. The method according to claim 5, characterized in that Before registering the simplified combined three-dimensional point cloud data, the method further includes: Denoise point cloud data; Fitting and filtering processing are performed on the basis of registering the combined three-dimensional point cloud data using the principal component analysis method.

Images