Automatic grinding system and grinding method

Abstract

The invention relates to an automatic grinding system and a grinding method. The method comprises the following steps that an image of a panel is obtained through an imaging device; acquiring the position of the defect part of the panel according to the acquired image; bringing a polishing robot into surface contact with the panel at the position; and operating the polishing robot while maintaining the surface contact between the polishing robot and the panel.

Description

Technical Field

[0001] This invention relates to an automated polishing system and method for surfaces. Background Technology

[0002] Recently, with the rapid popularization of electric vehicles, there is active research into solutions for lightweight vehicle bodies. As part of this, the application of aluminum sheets in vehicle bodies is constantly expanding.

[0003] Due to its material properties, aluminum is prone to defects such as chips, dents, scratches, and pits during manufacturing. These defects are visible to the naked eye after the painting process and can negatively impact the vehicle's appearance. Therefore, to remove these defects, manual full inspection and polishing are currently employed.

[0004] However, this manual polishing process, performed by operators, consumes excessive time, contributing to the increased unit price of components. Furthermore, aluminum dust is a well-known hazardous substance.

[0005] Existing technical documents

[0006] Republic of Korea Patent Publication No. 10-2023-0136805 (Publication Date: 2023.09.27) Summary of the Invention

[0007] The present invention is proposed to solve the above-mentioned problems and aims to provide an automatic grinding system and method that can automate grinding operations.

[0008] The present invention aims to provide an automated polishing system and method that can reduce labor time and costs.

[0009] The present invention aims to provide an automated polishing system and method capable of absorbing manufacturing tolerances and deviations that may occur during the production process.

[0010] The purpose of this invention is not limited to the purposes mentioned above, and other purposes not mentioned can be clearly understood from the following description by those skilled in the art to which this invention pertains (hereinafter referred to as "those skilled in the art").

[0011] In order to achieve the objectives of the invention as described above and to perform the specific functions of the invention as described below, the invention includes the following features.

[0012] According to some embodiments of the present invention, an automatic polishing method includes the following steps: acquiring an image of a panel using an imaging device; obtaining the location of a defective portion existing in the panel using a computer based on the acquired image; guiding a polishing robot including a polishing tool to the aforementioned location; bringing the polishing tool into contact with the panel surface at the aforementioned location; and operating the polishing robot while maintaining surface contact between the polishing tool and the panel.

[0013] According to a partial embodiment of the present invention, an automated polishing system includes: an imaging device configured to acquire three-dimensional data of an object; a multi-jointed robot including a polishing tool configured to polish the object; and a computer configured to determine, based on the three-dimensional data, the location of a defective portion existing in the object, and to control the robot to polish the defective portion at the specified location. The computer is configured to control each axis of the robot to perform polishing while maintaining a uniformly distributed pressure on the defective portion using the polishing tool.

[0014] Based on the present invention, an automatic grinding system and method capable of automating grinding operations are provided.

[0015] Based on the present invention, an automated polishing system and method that can reduce working hours and costs are provided.

[0016] According to the present invention, an automatic polishing system and method are provided that can absorb manufacturing tolerances and deviations that may occur during the production process.

[0017] The effects of the present invention are not limited to those described above, and those skilled in the art will clearly understand from the following description other effects not mentioned. Attached Figure Description

[0018] Figure 1 This is a structural diagram of an automatic polishing system based on an embodiment of the present invention.

[0019] Figure 2 A deep learning model of an automated polishing system based on an embodiment of the present invention is shown.

[0020] Figure 3 A robot based on an embodiment of the present invention is shown.

[0021] Figure 4 A force-torque sensor for an automated polishing system based on an embodiment of the present invention is shown.

[0022] Figure 5 The invention illustrates the grinding process of a grinding tool in an automated grinding system based on an embodiment of the invention.

[0023] Figure 6This is a flowchart of an automatic polishing method based on an embodiment of the present invention.

[0024] Figure 7 This is a flowchart of the operation of the robot in the automatic polishing method based on an embodiment of the present invention.

[0025] Explanation of reference numerals in the attached figures

[0026] 1: Polishing system 2: Object

[0027] 4: Defective Part 20: Imaging Device

[0028] 30: Multi-joint robot 40: Robot

[0029] 50: Grinding tools 60: Force-torque sensor

[0030] 62: Strain gauge; 70: Robot controller

[0031] 100: Computer; 110: Deep Learning Model

[0032] P: Path. Detailed Implementation

[0033] The specific structural and functional descriptions disclosed in the embodiments of this invention are merely examples provided to illustrate embodiments based on the concepts of this invention; embodiments based on the concepts of this invention can be implemented in various forms. Therefore, this invention should not be limited to the embodiments described in this specification, but should be understood to include all modified, equivalent, and alternative embodiments of this invention. All embodiments falling within the scope of the concept and technology of this invention are considered to be included within the protection scope of this invention.

[0034] Furthermore, in this invention, the terms "first," "second," etc., are used only to distinguish different components, and not to limit the nature of the components. For example, without departing from the concept of this invention, "first component" can be called "second component," and vice versa.

[0035] It is important to understand that when describing a component as "connected" or "connected" to another component, this connection can be a direct connection or an indirect connection through an intermediate component. Conversely, when it is said that a component is "directly connected" or "directly in contact" with another component, it should be understood that there are no other components between them. Furthermore, different expressions used to describe the relationship between components, such as "between" and "directly between," or "adjacent to..." and "directly adjacent to...", should be interpreted in the same way.

[0036] In this specification, the same reference numerals denote the same parts. The terminology used in this specification is for describing embodiments and is not intended to limit the invention. Unless otherwise specified, singular words in the specification also include their plural forms. Furthermore, the terms "comprising" and / or "including" as used in the specification should be understood to mean that the stated parts, steps, operations, and / or elements do not exclude the presence or addition of one or more other parts, steps, operations, and / or elements.

[0037] The present invention will now be described in detail with reference to the accompanying drawings.

[0038] As mentioned above, manual polishing of materials, including aluminum, can increase labor time and costs, and negatively impact the health of workers. Therefore, while the development of automated polishing technology is urgent, there are situations where automation of polishing is difficult to achieve. For example, due to the properties of aluminum, automated polishing is challenging. Specifically, aluminum generates excessive dust during forming and manufacturing, and defects such as dents or unevenness may occasionally appear in unpredictable areas.

[0039] To automatically remove defects such as bumps and dents in materials located in unpredictable areas, a robot, including a grinding tool, needs to be moved to the defective location and the grinding tool needs to adapt and grind the surface in real time, adapting to the curvature. Only in this way can defect-free products be obtained in mass production.

[0040] While automation can be achieved through teaching robots to operate, if the robot simply moves along a pre-taught path or position, it struggles to handle deviations from the product, the robot itself, or the manufacturing process. Therefore, it's difficult to ensure uniform polishing quality, and the likelihood of defective products is high. Furthermore, calibration of the pre-set teaching operation for the entire area of ​​the product requires regular adjustments and excessive time.

[0041] In response, the present invention provides an automatic polishing system and method that requires no teaching and is capable of adaptive control, enabling the polishing tool to make real-time surface contact with the curved surface of the object being polished.

[0042] like Figure 1 As shown, the polishing system 1 based on the present invention is configured to polish the surface of an object 2. The surface of the object 2 may have curvature or include curved portions. In one example, the object 2 may be a panel of a vehicle body. In one example, the object 2 may be a welded curved portion including a moving part of the vehicle body. In one example, the object 2 may be an aluminum sheet. However, the object 2 polished by the polishing system 1 is not limited to these. The object 2 polished by the polishing system 1 is applicable not only to vehicle bodies in bodywork processes but also to objects in all fields requiring polishing or grinding, such as painting processes.

[0043] In particular, the polishing system 1 is capable of polishing defective portions 4 formed on the surface of the object 2. Defective portions 4 may include bumps, scratches, or dents formed on the surface of the object 2. Such defective portions 4 may form sporadically on the object 2. The polishing system 1 can adapt to the surface of the object 2 in real time during the polishing process, and can polish defective portions 4 without separate teaching, even if the surface of the object 2 contains defective portions 4 with curvature or including bent portions.

[0044] In other words, the polishing system 1 is configured to determine the accurate three-dimensional position of the defect portion 4 through image processing and analysis of the object 2, and to perform polishing while adapting to the surface at the corresponding position in real time. For this purpose, the polishing system 1 includes an imaging device 20, a polishing device, and a computer 100. The computer 100 is configured to perform calculations for the operation of the polishing system 1 and to control the imaging device 20 and the polishing tools. In one embodiment, the computer 100 may perform calculations based on the image acquired by the imaging device 20 to obtain three-dimensional position information of the defect portion 4. Additionally, in another embodiment, the computer 100 is capable of performing control to cause the polishing tools to perform polishing based on the acquired three-dimensional position information.

[0045] Computer 100 includes memory and a processor. The processor, as hardware, is capable of executing computer-readable code or a series of instructions stored in memory and of processing data. As a non-limiting example, the processor may include a central processing unit, a graphics processing unit, a multi-core processor, a multiprocessor, an ASIC (Application-Specific Integrated Circuit), or a FPGA (Field-Programmable Gate Array).

[0046] The memory can store data, code, or a series of instructions that can be executed by a processor. The memory can be volatile or non-volatile. As a non-limiting example, volatile memory can include DRAM (dynamic random access memory) or SRAM (static random access memory). As another non-limiting example, non-volatile memory can include EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory, MRAM (Magnetic RAM), CD-ROM, or DVD-ROM.

[0047] Imaging device 20 is configured to acquire an image of object 2. In some embodiments, imaging device 20 can acquire three-dimensional data of object 2 using structured light. Imaging device 20 includes a camera and a projector. The camera is configured to capture a light pattern projected onto object 2, thereby acquiring two-dimensional data or two-dimensional position information of object 2. In one example, the camera may be a two-dimensional camera. The projector is configured to project the light pattern onto object 2. Imaging device 20 can combine depth information using the camera and the projector to construct three-dimensional data. Computer 100 is configured to match the position information of the two-dimensional data to the three-dimensional data to obtain position coordinates in a three-dimensional coordinate system. In one embodiment, imaging device 20 may be a three-dimensional machine vision system.

[0048] In order to convert specific coordinates in two-dimensional image data into three-dimensional spatial coordinates, computer 100 can use the intrinsic parameters, extrinsic parameters, and depth information of the camera device.

[0049] The internal parameters of the camera device can include the focal length (fx) in the x-direction, the focal length (fy) in the y-direction, and the intrinsic parameter matrix (K) based on the x-coordinate (Cx) and y-coordinate (Cy) of the principal point. The intrinsic parameter matrix (K) can be determined by Equation 1.

[0050] Formula 1

[0051]

[0052] The external parameters of the camera device can include the rotation matrix (R), the translation vector (T), and the extrinsic parameter matrix (M1). The extrinsic parameter matrix (M1) can be obtained using Equation 2.

[0053] Formula 2

[0054] M1 = [R|T]

[0055] Computer 100 can perform calculations using internal and external parameters to convert two-dimensional position coordinates into three-dimensional spatial coordinates. Regarding the above calculations, known techniques for converting coordinates of a two-dimensional image into coordinates of a three-dimensional image can be used; therefore, additional explanations related to this are omitted.

[0056] Reference Figure 2 According to some embodiments of the present invention, the accuracy of identifying the defective portion 4 in an image acquired by the imaging device 20 can be further improved through deep learning of the computer 100. For this purpose, the computer 100 can include a deep learning model 110. The deep learning model 110 is configured to learn an image of an object 2 including the defective portion 4, the two-dimensional position of the defective portion 4 in the image of the object 2, and data of the defective portion 4 in the image of the object 2. The data of the defective portion 4 has a different range of values ​​than the data of the non-defective portion in the object 2, thus it can be determined that the portion having values ​​within that range includes the defective portion 4. Based on the learned data, when the image P1 of the object 2 is input, the deep learning model 110 can output the position coordinates of the defective portion 4 in the image of the object 2 and the data of the defective portion 4 in the image of the object 2 as output P2. Therefore, when the computer 100 identifies the defective portion 4 in an image acquired from the imaging device 20, it can identify the defective portion 4 with a high probability. That is, by calculating the pixel values ​​on the two-dimensional image and matching them with the three-dimensional data, the desired feature region, i.e., the three-dimensional data of the defective portion 4, can be obtained. As described below, the acquired 3D data can be converted into robot coordinates, and robot 40 can move according to these coordinates. This invention enables robot 40 to find the defective part 4 autonomously without the need for teaching, and without relying on the previously required robot programming. In one example, the deep learning model 110 can be a model based on the YOLO (You Only Look Once) deep learning algorithm.

[0057] According to some embodiments of the present invention, at least a portion of the defective portion 4 can be pre-emphasized before acquiring an image via the imaging device 20. For example, at least a portion of the defective portion 4 of the object 2 can be pre-marked by means of coloring or other methods, as can be observed with the naked eye. In another example, scratches are artificially formed on the defective portion 4 that can be seen with the naked eye using an oilstone or sandpaper, thereby increasing the shadow of the defective portion 4 with unevenness, dents, etc., making it easier to identify the defective portion 4 based on the imaging device 20. This is because when the degree of defect of the defective portion 4 on the object 2 is small, there may be cases where it cannot be identified by the imaging device 20 or the computer 100, so the identifiability of the defective portion 4 can be further improved by the above-described process.

[0058] In some embodiments, the two-dimensional image acquired by the camera device can be preprocessed by the computer 100. The computer 100 can preprocess the defect portion 4 in the two-dimensional image through hue, saturation, and value (HSV) color analysis. The preprocessing can include increasing the saturation in the HSV. By adjusting the saturation of the defect portion 4 during the preprocessing, the detection rate of the defect portion 4 can be improved.

[0059] In one example, the imaging device 20 can be mounted on a jointed robot 30. The computer 100 can control the movement of the jointed robot 30 and control the imaging device 20 to position the image on the object 2 by controlling the jointed robot 30.

[0060] The polishing system 1 includes a polishing device. In one embodiment, the polishing device may be a robot 40, and may be a multi-joint robot. As a non-limiting example, the polishing device may be a multi-joint collaborative robot capable of multi-axis control. As a non-limiting example, the polishing device may be a multi-joint collaborative robot capable of six-axis control.

[0061] Robot 40 includes a polishing tool 50. The polishing tool 50 is configured to polish the surface of object 2 like sandpaper. The polishing tool 50 can be detachably mounted to the working end of robot 40. The polishing tool 50 can rotate via a spindle mounted on robot 40. The polishing tool 50 can rotate while applying pressure to the surface of object 2 via robot 40.

[0062] like Figure 3As shown, robot 40 includes force-torque sensor 60. In one example, force-torque sensor 60 may be a multi-axis force-torque sensor. As a non-limiting example, force-torque sensor 60 may be a six-axis force-torque sensor. Force-torque sensor 60 senses the applied force, and robot 40 can adjust the force applied to the surface of object 2 based on the sensed information.

[0063] Reference Figure 4 According to an embodiment of the present invention, the force-torque sensor 60 may be a resistive sensor that converts deformation caused by external force into an electrical signal. In one embodiment, the force-torque sensor 60 includes a strain gauge 62. The force-torque sensor 60 is able to detect deformation as an electrical signal, i.e., a change in impedance, through the strain gauge 62, and provide the detected information to the robot 40.

[0064] According to embodiments of the present invention, the force-torque sensor 60 may include a plurality of strain gauges 62. The strain gauges 62 may be arranged at predetermined intervals within the force-torque sensor 60. As shown in the embodiment, the strain gauges 62 may be arranged at predetermined intervals along the circumference of the force-torque sensor 60. In some embodiments, the force-torque sensor 60 may include four or more strain gauges 62, each strain gauge 62 may be arranged at approximately 90° intervals in the circumferential direction (62a, 62b, 62c are shown in the figures, 62d is omitted). However, the number of strain gauges 62 may vary.

[0065] Specifically, the force-torque sensor 60 measures the pressure applied to the polishing tool 50 as a change in impedance when the polishing tool 50 contacts the surface of the object 2. According to the present invention, when polishing the surface of the object 2, pressure equalization control is achieved by controlling each axis of the robot 40 to make the impedance values ​​measured in each strain gauge 62 uniform. This allows the robot 40 to adapt to the surface of the object 2 in real time and perform polishing without considering whether the surface of the object 2 has undulations or curvature, thus enabling automated polishing operations.

[0066] In one embodiment, such as Figure 5As shown, the robot 40 can be controlled to move such that the polishing tool 50 moves along a path P surrounding the defect portion 4, centered on the defect portion 4. In one embodiment, the robot 40 can be controlled to move the polishing tool 50 along a square path, centering the defect portion 4. For example, in the case of rough polishing, the polishing tool 50 can move along a square path with a side length of 30 mm; in the case of fine polishing, the polishing tool 50 can move along a square path with a side length of 50 mm. Polishing the unevenness of a vehicle body panel requires extensive work. That is, focusing polishing only on one area may leave too many dents or marks. To address this, the present invention is configured to move the polishing tool 50 along a path surrounding the defect portion 4 during the polishing process, thereby preventing the aforementioned problems.

[0067] Robot 40 includes robot controller 70. Robot controller 70 is configured to receive sensing values ​​from force-torque sensor 60. Furthermore, robot controller 70 is configured to control each axis of robot 40 based on the sensing values ​​from force-torque sensor 60.

[0068] The robot controller 70 is configured to communicate with the computer 100. The computer 100 is configured to transmit the three-dimensional position coordinates of the defect portion 4 to the robot controller 70. Furthermore, the robot controller 70 is configured to cause the robot 40 to perform a grinding operation at the corresponding position based on the received three-dimensional position coordinates. In one embodiment, the communication between the computer 100 and the robot controller 70 can use Ethernet TCP (Transmission Control Protocol).

[0069] According to an embodiment of the present invention, the computer 100 is configured to perform calculations that associate the coordinate system of the imaging device 20, the coordinate system of the robot 40, and the coordinate system of the polishing tool 50 with each other.

[0070] Computer 100 is configured to view the coordinates of the defect portion 4 observed by imaging device 20, i.e., the coordinate system (x, y) of imaging device 20. i y i , z i Convert to the robot's 40 coordinate system (x) r y r , z r In some embodiments, in order to adjust the coordinate system (x, y) of the imaging device 20... i y i , z i Mapped to the Cartesian coordinate system of robot 40 (x r y r , z r The transformation matrix (R) can be used. In one example, the coordinate system of robot 40 (x...) ry r , z r The coordinate system can be a Cartesian coordinate system of robot 40. The Cartesian coordinate system uses the origin of robot 40 as a reference to show the movement along the x, y, and z axes. Through the calibration of imaging device 20 and robot 40, the distance between the three-dimensional position coordinates of the defect portion 4 measured by imaging device 20 and the origin of robot 40 is calculated, thereby enabling the setting of the movement distance of robot 40 to the defect portion 4. That is, computer 100 can obtain the target coordinates (x, y, z) that robot 40 needs to move from its origin along the x-axis, y-axis, and z-axis to the defect portion 4. Computer 100 can obtain the target coordinates (x, y, z) and send them to robot controller 70, thereby enabling robot 40 to move to the target coordinates (x, y, z) representing the defect portion 4 to perform grinding.

[0071] In another embodiment, the computer 100 is configured to send target coordinates (x, y, z) taking into account the rotation values ​​of the polishing tool 50. The coordinate system of the polishing tool 50 (x, y, z) t y t , z t The coordinate system is based on the endpoint of the grinding tool 50 for movement. Even through the Cartesian coordinate system (x...) of the robot 40... r y r , z r The movement of robot 40 was determined, but the rotation value of robot 40 was not determined because it was around the coordinate system (x, y) of the grinding tool 50. t y t , z t The rotation of each axis of the object 2 requires calculation of the rotation amount based on the curvature of the object 2. For example, in determining the coordinate system (x, y) of the grinding tool 50... t y t , z t When rotating the robot 40, the axis direction can be aligned with the right thumb, the direction of the other fingers can be set as positive (+), and the opposite direction can be set as negative (-). Considering the above points, the computer 100 can send the coordinates (x, y, z, Rx, Ry, Rz) as the final target coordinates to the robot controller 70. x, y, z can represent the linear movement of the robot 40 along its three axes, and Rx, Ry, Rz can represent the rotation of the robot 40 about each axis.

[0072] Furthermore, the computer 100 is configured to calculate the curvature of the defective portion 4. The rotation value of the robot 40 should be determined based on the curvature of the object 2 so that the grinding tool 50 fits tightly against the defective portion 4 for adaptive control. If the rotation correction value is not reflected when grinding operations are performed on areas with high curvature, surface contact adaptive control may not be possible. To address this, the computer 100 can calculate the curvature change of the object 2 by considering the normal vector in the three-dimensional image. The normal vector can be determined by the computer 100 when acquiring three-dimensional position information from the acquired two-dimensional image.

[0073] Computer 100 can obtain the rotation amount of the normal vector in defect section 4 using Rodrigues' rotation formula. Rodrigues' rotation formula determines the rotation amount along each axis between two vectors. Thus, the curvature of the measurement point relative to an arbitrarily set reference point can be calculated. However, this rotation amount is the rotation amount of imaging device 20, so computer 100 can multiply the rotation amount of the normal vector obtained from the 3D image by a rotation matrix to convert it into the rotation amount of robot 40. Furthermore, the rotation amount of robot 40 follows the rotation amount of the coordinate system of grinding tool 50, so an additional process can be performed to convert the rotation amount of robot 40 in the Cartesian coordinate system into the coordinate system value of grinding tool 50. Through this calculation, the rotation amount of robot 40 can be determined, thereby enabling the transmission of the position coordinates (x, y, z, Rx, Ry, Rz) of the final target location to robot controller 70.

[0074] As described above, the position of the defect portion 4 obtained through image analysis is in the coordinate system (x, y) of the imaging device 20. i y i , z i The position information is generated based on the coordinate system (x, y) of the robot 40. The position information of this defect part 4 is then input into the coordinate system (x, y) of the robot 40. r y r , z r When the robot 40 is in motion, it can move along the robot coordinate system. This is achieved by adjusting the coordinate system (x, y) of the imaging device 20. i y i , z i Mapped to the robot's 40 coordinate system (x r y r , z r This allows robot 40 to move to defect section 4. As a way to unify two different coordinate systems, a transformation matrix (R) can be used. The transformation matrix (R) is a 4x4 matrix, where the initial 3x3 matrix portion (rotation transformation portion) represents the different coordinate systems, and the remaining 4x1 portion (translation transformation portion) represents the amount of origin movement between the two coordinate systems.

[0075]

Formula 3

[0076]

[0077] The rotation transformation can be calculated from the transformation matrix (R) based on the difference in rotation between the two different coordinate systems and the movement of the origin using the Euler angle formula (Equation 4).

[0078]

Formula 4

[0079]

[0080] Here, Ax, Ay, and Az refer to the rotation values ​​along the x-axis, y-axis, and z-axis, respectively. If the rotation amount is known, the rotation transformation part can be calculated using Equation 4. However, even if the imaging device 20 and robot 40 are installed in a pre-set manner, the accurate rotation amount cannot be known due to ground inclination at the installation location, errors in the installation position, etc. Therefore, to reduce this error, this invention is configured to actually calculate the rotation amount. That is, within the imaging area of ​​the imaging device 20, the movement of each axis of the robot 40 is known, thereby enabling the calculation of the rotation amount based on the coordinate system (x-y-z) of the imaging device 20. i y i , z i Obtain the robot's 40 coordinate system (x r y r , z r The displacement of ) can be calculated by inputting the error value obtained based on this into Equation 4.

[0081] The translation transformation can be obtained by additionally establishing three equations. That is, these three equations can be derived using the coordinate system (x, y) of the imaging device 20 and the robot 40. r y r , z r The coordinates of robot 40 can be obtained by multiplying the transformation matrix (R) with the coordinates of imaging device 20. The coordinates of defect part 4 (in the coordinate system of imaging device 20) can be obtained through imaging device 20, and the coordinates of robot 40 of defect part 4 can be obtained through teaching robot 40. Since the three-dimensional coordinates of robot 40 and imaging device 20 are known, the three transformation parts as unknowns can be calculated through three equations, and robot 40 can be moved to defect part 4 based on the mapped coordinate system.

[0082] like Figure 6 As shown, in step S600, the polishing system 1 is configured to acquire a three-dimensional image of the surface of the object 2. As described above, a two-dimensional image of the object 2 can be acquired by the imaging device 20.

[0083] In step S610, the position coordinates of the defective part 4 can be obtained. The computer 100 performs a calculation that associates the coordinate system of the robot 40, the coordinate system of the imaging device 20, and the coordinate system of the polishing tool 50, thereby determining the position coordinates of the robot 40 that it can move to in the coordinate system of the robot 40 to the defective part 4.

[0084] In step S620, based on the acquired position coordinates, the robot 40 can automatically polish the surface of the object 2. The polishing tool 50 can adapt to the surface of the object 2 and perform the polishing operation through the force-torque sensor 60, regardless of whether the surface of the object 2 has curvature.

[0085] Regarding the automatic polishing process, please refer to... Figure 7 In step S700, robot 40 moves to the defect part 4 at the corresponding position coordinates according to the obtained position coordinates.

[0086] When multiple defective parts 4 exist, the computer 100 can generate the optimal movement path for the robot 40. In one embodiment, the optimal movement path may refer to the shortest distance that passes through all defective parts 4 when multiple defective parts 4 exist on the object 2.

[0087] In step S710, it is confirmed whether the polishing tool 50 of the robot 40 is in surface contact with the surface of the object 2. Here, surface contact refers to the state in which the polishing tool 50 is uniformly and closely attached to the surface of the object 2. The computer 100 can determine whether surface contact exists based on the impedance values ​​in each direction of the strain gauge 62 of the force-torque sensor 60. When the impedance values ​​in each direction are uniform, the computer 100 can determine that the polishing tool 50 is in surface contact with the surface of the object 2.

[0088] When it is determined that surface contact has been achieved, the computer 100 begins polishing based on the robot 40 (S720). The robot controller 70 rotates the spindle to rotate the polishing tool 50, and can apply a preset pressure to the surface of the object 2 through axis control of the robot 40.

[0089] The robot controller 70 is configured to receive the sensing values ​​of the force-torque sensor 60 in real time after the grinding process begins (S730). The robot controller 70 is also able to receive the impedance values ​​of each strain gauge 62 and determine whether these values ​​are uniform.

[0090] The robot controller 70 is configured to control each axis of the robot 40 to ensure uniform pressure in each direction of the force-torque sensor 60 (S740). Uniform pressure in each direction indicates that the polishing tool 50 is performing polishing while in close contact with the surface of the object 2. Therefore, according to the present invention, polishing operations can be performed automatically without teaching the robot 40.

[0091] Previously, teaching was required for each sample. This existing technology struggled to adapt flexibly to deviations on curved surfaces and could lead to poor polishing. However, according to the present invention, an automated polishing system is provided that can achieve robot position changes along the curved surface of an object without requiring teaching.

[0092] Previously, after an operator identified defects on a sanding object divided into multiple areas, the sanding robot, based on the operator's pre-set instructions, would move along a predetermined pattern within the area where the defects were located while performing the sanding operation. That is, pre-teaching of the robot's path was required. Furthermore, the sanding robot was configured to move only according to a predetermined pattern, without considering the curvature of the sanding object. In contrast, according to the present invention, accurate coordinates (i.e., accurate positions) of the defects can be obtained based on an image of the object, and the sanding robot moves to the obtained position and automatically performs sanding along the curvature of the object at that position without prior teaching.

[0093] Due to the characteristics of stamping, aluminum materials exhibit curvature deviations. This invention provides an automated polishing system that can flexibly achieve surface contact and perform polishing regardless of whether deviations exist.

[0094] The polishing system and method based on the present invention are applicable to all processes that require surface polishing.

[0095] The present invention described above is not limited to the foregoing embodiments and figures. Various substitutions, modifications and alterations can be made without departing from the technical concept of the present invention, which will be obvious to those skilled in the art.

Claims

1. An automatic polishing method, comprising the following steps: Image of the panel is acquired through an imaging device; Based on the acquired images, the location of the defective portion existing in the panel is determined by computer; Guide the polishing robot, including the polishing tools, to the said position; The polishing tool is brought into contact with the panel surface at the specified position; and The polishing robot is operated while maintaining surface contact between the polishing tool and the panel.

2. The automated polishing method of claim 1, wherein, The steps to locate the defect include: The computer identifies defective portions in the image; and The computer calculates the location coordinates of the identified defect.

3. The automatic polishing method according to claim 2, wherein, The step of identifying defective parts is performed by a pre-learned deep learning model of the computer. The deep learning model is configured to provide location data of the defective parts when inputting an image of the panel.

4. The automatic polishing method according to claim 1, wherein, The steps to locate the defect include: The imaging device captures the light pattern projected onto the panel, thereby obtaining two-dimensional data of the object; The light pattern is projected onto the object using the imaging device; Depth information is extracted by the imaging device based on the deformation of the captured light pattern; and Based on the internal and external parameters of the imaging device and the depth information, three-dimensional data is obtained from the two-dimensional data.

5. The automatic polishing method according to claim 1, wherein, The steps to locate the defect include: The two-dimensional data of the defect portion acquired by the imaging device is mapped into three-dimensional data.

6. The automatic polishing method according to claim 5 further includes the following steps: Obtain the normal vector of the defect portion from the three-dimensional data; Calculate the rotation amount of the normal vector on the defective part; Based on the rotation of the normal vector, determine the change in curvature of the defective portion; and The polishing tool is operated according to the determined amount of curvature change.

7. The automatic polishing method according to claim 1, wherein, The steps to locate the defect include: The position of the defect portion acquired relative to the imaging device is converted to its position relative to the robot.

8. The automatic polishing method according to claim 1, wherein, The steps for operating a polishing robot include: The axes of the grinding robot are controlled based on the sensing values ​​of the force-torque sensors.

9. The automatic polishing method according to claim 8, comprising the following steps: The axes of the grinding robot are controlled such that the deformations of the multiple strain gauges of the force-torque sensor are substantially the same.

10. The automatic polishing method according to claim 1, wherein, The steps for operating a polishing robot include: The polishing robot controls the polishing tool to apply pressure to the panel while rotating at a preset speed.

11. The automatic polishing method according to claim 1, wherein, The steps for operating a polishing robot include: The polishing tool rotates while moving along a path around the defective part, with the defective part as the center.

12. The automatic polishing method according to claim 1, wherein, The panel is made of aluminum and has a curvature.

13. The automatic polishing method according to claim 1, wherein, The defects include dents, scratches, or bumps formed on the panel surface.

14. The automatic polishing method according to claim 1, wherein, The imaging device is a three-dimensional machine vision camera device.

15. An automatic polishing system, comprising: An imaging device configured to acquire three-dimensional data of an object; A multi-jointed robot including a polishing tool configured to polish the object; as well as A computer configured to determine the location of a defective portion existing in the object based on the three-dimensional data, and to control a robot to polish the defective portion at that location. The computer is configured to control each axis of the robot so that the grinding tool maintains a uniform pressure distribution on the defective part while performing grinding.

16. The automatic polishing system according to claim 15, wherein, The robot includes a force-torque sensor, and the computer is configured to determine in real time whether the pressure is uniform based on the sensing values ​​of the force-torque sensor.

17. The automatic polishing system according to claim 15, wherein, The robot includes force-torque sensors. The force-torque sensor includes at least four strain gauges arranged at predetermined intervals. The computer is configured to determine whether the pressure is uniform based on whether the deformation of each strain gauge is basically the same.

18. The automatic polishing system according to claim 15, wherein, The computer includes a pre-learned deep learning model configured to identify defective parts and provide the location of the defective parts when an image of an input object is received.

19. The automatic polishing system according to claim 15, wherein, The object is a car body panel made of aluminum.

20. The automatic polishing system according to claim 15, wherein, The defects include dents, scratches, or bumps formed on the surface of an object that has curvature.

Images