Paint surface maintenance operation robot and operation method

Abstract

The invention provides a paint surface maintenance operation robot and an operation method. The paint surface maintenance operation robot is used for repairing the damaged paint surface of the vehicle; the paint surface maintenance operation robot comprises a cabinet body, a control mechanism, and a moving mechanism, a detecting mechanism and a repairing mechanism which are connected to the cabinet body; the moving mechanism can move relative to the cabinet body, and the moving mechanism is used for driving the repairing mechanism to move relative to the damaged paint surface of the vehicle, so that the repairing mechanism can move to the damaged paint surface of the vehicle and repair the damaged paint surface of the vehicle; the detection mechanism is connected to the moving mechanism and used for detecting damage parameters of the damaged paint surface of the vehicle, and the damage parameters comprise the damage depth and the damage area; the control mechanism is configured to control work of the moving mechanism and the repairing mechanism based on detection data of the detection mechanism. According to the paint surface maintenance operation robot, vehicle body covering parts do not need to be disassembled in the whole process, and all operations are completed in the whole vehicle in-situ state.

Description

Technical Field

[0001] This application relates to the field of paint surface repair technology, specifically to paint surface repair robots and operating methods. Background Technology

[0002] In the current automotive aftermarket repair industry, the conventional repair process for defects such as scratches, dents, bubbles, or localized color differences on the car body paint usually relies on a "disassembly-transfer-manual repair-assembly" work mode. Specifically, repair personnel first need to remove the body panels (such as door trim panels, fenders, light assemblies, or bumpers) around the damaged area from the car body. Then, the removed parts are transferred to a dedicated paint booth or sanding station, where technicians use tools such as sandpaper, scrapers, and spray guns to manually sand, fill putty, spray primer and topcoat, and finally polish and reassemble.

[0003] This traditional repair method has a significant technical drawback: for paint damage located in the high part of the vehicle body or with complex structures, disassembling the relevant covering parts is not only time-consuming and labor-intensive, but also easily causes clips to break, seals to deform, or the electrophoretic anti-corrosion layer to break, leading to secondary damage such as assembly noise, rainwater leakage, or corrosion of the metal substrate, seriously affecting the structural integrity, durability, and resale value of the vehicle. Furthermore, because the disassembly and reassembly process is irreversible, even after repair and reinstallation, it is difficult to completely restore the original factory assembly condition.

[0004] Therefore, there is an urgent need for a technical solution that can automatically repair damaged paint surfaces in situ without disassembling body parts, in order to avoid the structural risks and additional costs caused by disassembly. Summary of the Invention

[0005] The embodiments of this application provide a paint surface repair robot and a method for performing the repair.

[0006] In a first aspect, embodiments of this application provide a paint repair robot for repairing damaged paint on vehicles; the paint repair robot includes a cabinet, a control mechanism, and a moving mechanism, a detection mechanism, and a repair mechanism connected to the cabinet;

[0007] The moving mechanism is movable relative to the cabinet. The moving mechanism is used to drive the repair mechanism to move relative to the damaged paint surface of the vehicle, so that the repair mechanism can move to the damaged paint surface of the vehicle and perform repair operations on the damaged paint surface of the vehicle. The detection mechanism is connected to the moving mechanism and is used to detect damage parameters of the vehicle's damaged paint surface. The damage parameters include damage depth and damage area. The control mechanism is configured to control the operation of the moving mechanism and the repair mechanism based on the detection data from the detection mechanism.

[0008] Secondly, embodiments of this application provide a paint surface repair operation method, applied to the paint surface repair operation robot described above; the paint surface repair operation method includes: S1. When the detection mechanism is above the middle area of ​​the damaged paint surface of the vehicle, control the detection mechanism to obtain the damage parameters of the damaged paint surface of the vehicle. S2. Determine that the damage parameters are within a preset range; S3. Control the moving mechanism to move the repair mechanism to the damaged paint surface of the vehicle; S4. Control the repair mechanism to perform repair operations on the damaged paint surface of the vehicle.

[0009] The beneficial effects of the embodiments of this application are as follows: In the embodiments of this application, the cabinet serves as the support and integration platform for the entire machine, housing the control mechanism, power module, and material supply system. The moving mechanism is installed on the top or side of the cabinet and can slide along a horizontal track or adjust its spatial position via a multi-degree-of-freedom robotic arm. The repair mechanism and cleaning mechanism are installed side-by-side on the end effector of the moving mechanism, and their working states can be switched under the command of the control mechanism. The detection mechanism is fixed to the front end of the moving mechanism, with its optical axis pointing towards the area to be repaired, ensuring non-contact scanning is completed before repair. During operation, the moving mechanism first positions the detection mechanism near the damaged paint surface, where the detection mechanism acquires parameters such as damage depth and area and transmits them to the control mechanism. Based on this, the control mechanism plans the repair path and drives the moving mechanism to precisely move the repair mechanism to the damaged location, sequentially performing operations such as grinding, filling, spraying, or polishing. The entire process does not require disassembling the vehicle body panels, and all operations are completed in the original position of the vehicle. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic diagram of the structure of the paint repair robot provided in an embodiment of this application; Figure 2 This is provided by the embodiments of this application. Figure 1 Enlarged structural diagram at point A; Figure 3 This is provided by the embodiments of this application. Figure 1Enlarged structural diagram at point B; Figure 4 This is a flowchart of a paint surface repair operation method provided in an embodiment of this application. Detailed Implementation

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. In addition, it should be understood that the specific embodiments described herein are only for illustration and explanation of this application and are not intended to limit this application. In this application, unless otherwise stated, directional terms such as "upper" and "lower" generally refer to the upper and lower positions of the device in actual use or operation, specifically the drawing directions in the accompanying drawings; while "inner" and "outer" refer to the outline of the device.

[0013] The following is combined Figures 1 to 4 This application describes the paint surface repair robot and its operation method.

[0014] According to an embodiment of the first aspect of this application, this application provides a paint repair robot for repairing damaged paint on vehicles. See also... Figure 1 , Figure 2 and Figure 3 The paint surface repair robot includes a cabinet 1, a control mechanism, and a moving mechanism 2, a detection mechanism 3, and a repair mechanism 4 connected to the cabinet 1. The moving mechanism 2 can move relative to the cabinet 1. The moving mechanism 2 is used to drive the repair mechanism 4 to move relative to the damaged paint surface of the vehicle, so that the repair mechanism 4 can move to the damaged paint surface of the vehicle and perform repair operation on the damaged paint surface of the vehicle. The detection mechanism 3 is connected to the moving mechanism 2. The detection mechanism 3 is used to detect the damage parameters of the vehicle's damaged paint surface. The damage parameters include the damage depth and the damage area. The control mechanism is configured to control the operation of the moving mechanism 2 and the repair mechanism 4 based on the detection data from the detection mechanism 3.

[0015] Understandably, cabinet 1 serves as the support and integration platform for the entire machine, housing the control mechanism, power module, and material supply system. The moving mechanism 2 is installed on the top or side of cabinet 1, allowing it to slide along horizontal tracks or adjust its spatial position via a multi-degree-of-freedom robotic arm. Repair mechanism 4 and cleaning mechanism 5 are mounted side-by-side on the end effector of moving mechanism 2, and their working states can switch under the command of the control mechanism. Detection mechanism 3 is fixed to the front end of moving mechanism 2, with its optical axis pointing towards the area to be repaired, ensuring non-contact scanning is completed before repair. During operation, moving mechanism 2 first positions detection mechanism 3 near the damaged paint surface, where it acquires parameters such as damage depth and area and transmits them to the control mechanism. The control mechanism then plans the repair path and drives moving mechanism 2 to precisely move repair mechanism 4 to the damaged location, sequentially performing operations such as grinding, filling, spraying, or polishing. The entire process requires no disassembly of the vehicle body panels, and all operations are completed in the original position of the vehicle.

[0016] Compared to the traditional process of "disassembly-transfer-manual repair-reassembly" in existing technologies, this application integrates inspection, repair, and cleaning functions into an integrated robot structure, with a control mechanism coordinating the actions of each mechanism. This fundamentally avoids secondary damage caused by disassembling door panels, bumpers, or light assemblies, such as buckle breakage, seal failure, or damage to the anti-corrosion layer. At the same time, since the repair work is carried out directly in the original assembled state of the vehicle body, the geometric continuity and mechanical integrity of the paint repair area and the surrounding structure are ensured, effectively preventing problems such as assembly noise, rainwater leakage, and substrate corrosion, significantly improving repair quality and vehicle resale value. In addition, automated operation greatly shortens the time for manual intervention, improving repair consistency and efficiency.

[0017] In some examples, cabinet 1 is equipped with casters and an electromagnetic braking device at the bottom for easy and flexible positioning in the workshop; the moving mechanism 2 is a six-axis collaborative robotic arm with a repeatability of ±0.05 mm; the detection mechanism 3 is a line laser 3D scanner with a sampling frequency of not less than 2 kHz; the repair mechanism 4 integrates an electric sanding head, a micro-metering putty nozzle, a low-pressure atomizing spray gun, and a soft polishing wheel; and the cleaning mechanism 5 includes a negative pressure dust suction port and a rotating brush.

[0018] In some embodiments, see Figure 1 , Figure 2 and Figure 3 The paint repair robot also includes a cleaning mechanism 5, which is connected to the cabinet 1. The cleaning mechanism 5 is used to clean the impurities generated by the repair mechanism 4 when it repairs the damaged paint surface of the vehicle.

[0019] Understandably, the cleaning mechanism 5 is fixedly installed on the front or top of the cabinet 1, and is spatially adjacent to but not interfering with the repair mechanism 4. When the repair mechanism 4 performs sanding, scraping, or spraying operations on the damaged paint surface of the vehicle, dust, putty debris, paint mist, or other particulate impurities will be generated. The cleaning mechanism 5 is activated in real time under the synchronous scheduling of the control mechanism, and removes the above-mentioned impurities from the repair area and surrounding vehicle body surface in a timely manner through negative pressure suction, airflow guidance, or mechanical cleaning, preventing them from adhering to the unrepaired paint surface or seeping into the gaps of the vehicle body. Specifically, the air inlet or dust suction port of the cleaning mechanism 5 faces the repair work point, and its position is optimized by fluid simulation to ensure that a local closed airflow channel is formed while the repair action is in progress, effectively capturing suspended particles. Impurities are transported through pipes to the dust collection bin or filter unit inside the cabinet 1 to achieve solid-gas separation and centralized collection.

[0020] This design solves the problem of secondary pollution caused by dust diffusion in traditional manual repair by directly integrating the cleaning mechanism 5 into the cabinet 1 and linking it with the repair process. Especially in the whole vehicle in-situ repair mode, it avoids impurities falling into sensitive areas such as door rails, light fixture interfaces or sealing strips, thereby preventing abnormal noise, corrosion or functional failure caused by these. At the same time, timely removal of sanding debris ensures that subsequent filling and spraying processes are carried out on a clean substrate, significantly improving paint film adhesion and surface smoothness. More importantly, since the entire cleaning process does not require manual intervention and is completed simultaneously with the repair action, it greatly shortens the process interval time and improves the continuity and efficiency of automated operation.

[0021] In some examples, the cleaning mechanism 5 includes a negative pressure fan, a flexible suction pipe, a wide-angle nozzle, and a high-efficiency filter; the nozzle edge is provided with a soft silicone lip to conform to the curved surface of the vehicle body to enhance sealing; the negative pressure fan is speed-adjusted by the control mechanism according to the repair intensity, using a high airflow mode during the coarse grinding stage and switching to a low-noise silent mode during the fine polishing stage; the dust collection chamber is detachable and equipped with a full-load sensor for easy maintenance.

[0022] In some embodiments, see Figure 1 , Figure 2 and Figure 3 The repair mechanism 4 includes a polishing component 41, which is used to polish the damaged paint surface of the vehicle. The cleaning mechanism 5 includes a water spray component and an air spray component, which are electrically connected to the control mechanism. The control mechanism is configured to: after the polishing component 41 polishes the damaged paint surface of the vehicle, control the water spray component to spray water onto the damaged paint surface of the vehicle to clean the powder generated by the polishing component 41; and after the water spray component sprays water onto the damaged paint surface of the vehicle, control the air spray component to spray air onto the damaged paint surface of the vehicle.

[0023] Understandably, the grinding component 41 is installed at the end of the repair mechanism 4 and can be driven by a motor to rotate or reciprocate, and is used to perform local grinding on scratches, dents or old paint layers; the water spray component and the air spray component are fixedly connected to the cabinet 1 and arranged in the vicinity of the working area of ​​the grinding component 41, with their nozzles facing the grinding point to ensure coverage of the entire grinding area. During operation, the sanding component 41 first completes the sanding operation on the damaged paint surface under the command of the control mechanism, generating fine dust and paint shavings. Subsequently, the control mechanism immediately activates the water spray component, which uses a micro water pump to atomize and spray deionized water or cleaning liquid containing corrosion inhibitors through fine nozzles, wetting and washing away the sanding residue, causing it to form a mud-like substance that adheres to the surface and no longer flies away. After the water spraying action continues for a preset time (e.g., 1–3 seconds), the control mechanism shuts off the water spray component and immediately activates the air spray component. The compressed air source drives clean and dry gas to blow at high speed through the annular nozzles to sweep the wetted area, quickly evaporating the moisture and carrying away the mud residue, restoring the paint surface to a dry and clean state, providing an ideal base for subsequent filling of putty or spraying of primer.

[0024] This structure effectively solves the problems of dust diffusion, contamination of body seams, and impact on subsequent coating adhesion caused by dry sanding through a three-step in-situ cleaning mechanism of "sanding-water spraying-air spraying". The water spraying process not only suppresses dust but also cools the sanded area, preventing local overheating and damage to the original paint film. The air spraying process avoids water stains and eliminates the risk of rust or paint film blistering caused by moisture. The entire cleaning process is automatically executed in sequence by the control mechanism, without the need for manual wiping or additional cleaning stations, significantly improving the automation and environmental friendliness of the repair operation, while ensuring the cleanliness and reliability of in-situ repair in the condition of the whole vehicle.

[0025] In some examples, the abrasive component 41 is a replaceable sandpaper disc or diamond grinding head with a grit size range of P400 to P2000; the water spray component includes a miniature diaphragm pump, a water tank, and a fan-shaped atomizing nozzle with a spray angle of 30°–60°; the air jet component is supplied with air by an oil-free silent air compressor with an outlet pressure of 0.2–0.4 MPa, and the nozzle has a porous annular structure to achieve uniform purging; the water tank and air pipeline are integrated inside the cabinet 1 and are equipped with liquid level and air pressure sensors.

[0026] In some embodiments, see Figure 1 , Figure 2 and Figure 3The paint repair robot also includes a drying component 6 connected to the cabinet 1. The repair mechanism 4 also includes a putty scraping component 42, a painting component 43, and a polishing component. The drying component 6, the putty scraping component 42, the painting component 43, and the polishing component are all electrically connected to the control mechanism. The control mechanism is configured to control the sanding component 41, the cleaning mechanism 5, the putty scraping component 42, the painting component 43, the drying component 6, and the polishing component to work sequentially on the damaged paint surface of the vehicle.

[0027] Understandably, cabinet 1 serves as the integrated platform for the entire machine, housing the control mechanism, material supply system, and power unit. Grinding component 41, puttying component 42, painting component 43, and polishing component are modularly mounted on the end effector of the moving mechanism 2, and can be switched or worked collaboratively by the control mechanism. Water spraying component, air spraying component, and drying component 6 are fixedly connected to the front of cabinet 1, with their outlets facing the work area, forming a front-to-back collaborative layout with the repair mechanism 4 in space. During operation, the control mechanism first drives the grinding component 41 to perform coarse and fine grinding on the damaged paint surface, removing the oxide layer and uneven areas. Then, the cleaning mechanism 5 is activated, using water spray to clean the dust generated during grinding, and air spray to dry the surface. After the paint surface is clean and dry, the control mechanism switches to the putty component 42, which evenly applies the prepared putty to the depressions and initially smooths them. Next, the painting component 43 sprays the primer and color paint according to the preset number of layers. After spraying, the drying component 6 is activated, using infrared radiation or hot air to accelerate the curing of the paint film. Finally, the polishing component performs fine polishing on the cured paint surface, restoring gloss and smoothness. The entire process is automatically scheduled by the control mechanism according to a strict time sequence, with seamless coordination between the actions of each component, requiring no manual intervention.

[0028] This structure integrates six major functions—sanding, cleaning, scraping, spraying, drying, and polishing—into a single robotic platform, with a control mechanism enabling closed-loop control of the entire process. This completely eliminates the traditional "disassembly-transfer-manual repair" work mode. All processes are completed in the original position of the vehicle body, avoiding secondary damage such as clip breakage, seal failure, or damage to the anti-corrosion layer caused by disassembling body panels. At the same time, automated operation ensures the consistency of parameters for each process (such as putty thickness, number of paint layers, and drying temperature), significantly improving repair quality and appearance restoration. Especially in the repair of high-end vehicles, it can achieve a visual and tactile effect that is almost indistinguishable from the original paint, effectively maintaining the vehicle's residual value and customer satisfaction.

[0029] In some examples, the sanding component 41 uses a replaceable sanding disc driven by a brushless motor, with a grit size covering P400–P3000; the puttying component 42 uses a micro screw metering pump with a flat scraper head to achieve thickness control at the 0.1 mm level; the painting component 43 is a low-pressure atomizing spray gun that supports automatic mixing of two-component varnishes; the drying component 6 is an infrared ceramic heating plate with a temperature control range of 50–80℃ and an adjustable heating rate; the polishing component is equipped with a wool wheel and a microfiber disc with stepless speed adjustment from 0–3000 rpm; and the cabinet 1 has a built-in water circulation filtration system, a paint recovery device, and an activated carbon adsorption module for waste gas.

[0030] In some embodiments, see Figure 1 The paint surface repair robot also includes a placement mechanism 7, which is fixedly installed on the top of the cabinet 1 and is used to support the repair mechanism 4.

[0031] It is understandable that the cabinet 1, as the main support frame of the entire machine, has a placement mechanism 7 on its top. The placement mechanism 7 is a rigid platform or a bracket with a limiting slot, and its surface is equipped with anti-slip rubber pads or magnetic positioning points. When the repair mechanism 4 is not performing a work task, the moving mechanism 2 moves it back to the top of the cabinet 1 and accurately places it on the placement mechanism 7. The placement mechanism 7 provides stable support, dust protection, and standby positioning for the repair mechanism 4, ensuring that the repair mechanism 4 remains clean, safe, and in a preset initial position when not in operation. At the same time, since the repair mechanism 4 integrates a variety of precision tools (such as the nozzle of the paint spraying part 43 and the measuring head of the putty scraping part 42), placing it in the high placement mechanism 7 can avoid contamination or damage caused by ground dust, oil stains, or accidental contact by personnel. In addition, the position of the placement mechanism 7 is optically calibrated so that the repair mechanism 4 can maintain a consistent spatial reference after each return to its position, providing repeatability assurance for subsequent high-precision operations.

[0032] This design effectively solves the storage problem of the multi-functional repair mechanism 4 during idle periods by setting a dedicated placement mechanism 7 on the top of the cabinet 1. It not only improves the neatness and safety of the overall layout, but also significantly shortens the work preparation time. The moving mechanism 2 can directly grab the repair mechanism 4 from the placement mechanism 7 and quickly enter the working position without additional leveling or calibration. More importantly, in the scenario of whole vehicle in-situ repair, when the repair mechanism 4 frequently switches tools or returns to the maintenance position, the placement mechanism 7 provides a reliable "relay docking point", ensuring the continuity and reliability of the automated process and avoiding mechanical displacement or functional failure caused by the mechanism being suspended or placed randomly.

[0033] In some examples, the placement mechanism 7 is made of aluminum alloy plate with a concave positioning groove on the surface that matches the bottom contour of the repair mechanism 4; a permanent magnet or photoelectric alignment mark is embedded in the groove, which works with the Hall sensor or visual recognition module at the end of the moving mechanism 2 to achieve automatic alignment; a micro switch is integrated on the edge of the placement mechanism 7 to detect whether the repair mechanism 4 is fully in place and to feed back a status signal to the control mechanism.

[0034] In some embodiments, see Figure 1 and Figure 3 The paint surface repair robot also includes an observation device 8, which is installed on the right side of the cabinet 1. The observation device 8 can move freely up and down on the side rod of the cabinet 1, and is equipped with a fastening device. This observation device 8 is used to freely monitor the repair progress.

[0035] For example, the observation device 8 uses an independent linear slide rail module.

[0036] For example, the observation device 8 is manually fixed.

[0037] For example, the observation device 8 is directly connected to the control circuit board, and its cables are nested in the cabinet 1.

[0038] According to an embodiment of the second aspect of this application, this application also provides a paint surface repair method, applied to the aforementioned paint surface repair robot. See also... Figure 4 Paint surface repair procedures include: S1. When the detection mechanism 3 is positioned above the middle area of ​​the damaged paint surface of the vehicle, the detection mechanism 3 is controlled to acquire the damage parameters of the damaged paint surface of the vehicle.

[0039] For example, the control mechanism drives the moving mechanism 2 to make fine adjustments, so that the detection mechanism 3 is precisely positioned above the central area of ​​the damaged paint surface on the vehicle, ensuring that its optical axis is perpendicular to the curved surface of the vehicle body. Subsequently, the control mechanism activates the line laser 3D scanner in the detection mechanism 3 to perform a non-contact scan of the damaged area at a sampling frequency of not less than 2 kHz, acquiring point cloud data; the system calculates the damage parameters based on this point cloud data, including a damage depth of 0.35 mm and a damage area of ​​18 square centimeters. Here, "damage parameters" refer to quantitative indicators characterizing the geometric features of paint surface defects, including two core data points: damage depth and damage area.

[0040] S2. Determine that the damage parameters are within a preset range; For example, the control mechanism reads a preset range: the damage depth is between 0.1 and 1.0 mm, and the damage area is less than 50 square centimeters. Since the current damage depth of 0.35 mm and damage area of ​​18 square centimeters are both within this range, the system determines that the damage parameters meet the conditions for automatic repair and allows the subsequent repair process to proceed. If the damage exceeds the range (e.g., depth > 1.0 mm or area > 50 square centimeters), the automatic repair is terminated and a manual intervention prompt is generated.

[0041] S3. Control the moving mechanism 2 to move the repair mechanism 4 to the damaged paint surface of the vehicle; For example, the control mechanism plans the repair path based on the damage parameters and the three-dimensional model of the vehicle body, and generates motion commands to send to the moving mechanism 2. The moving mechanism 2 responds to the commands and moves smoothly along a six-degree-of-freedom spatial trajectory, accurately transporting the repair mechanism 4 from the placement mechanism 7 at the top of the cabinet 1 to the damaged paint surface of the vehicle, so that the working end face of the repair mechanism 4 maintains a preset working distance of 5 mm from the damaged area, and its posture is consistent with the normal of the vehicle body surface.

[0042] S4, the control and repair mechanism 4 performs repair operations on the damaged paint surface of the vehicle.

[0043] For example, the control mechanism activates the repair mechanism 4 to execute a complete repair operation sequence: First, the grinding component 41 is driven to reciprocate and grind the damaged paint surface to remove the oxide layer and burrs; then, the water spray component in the cleaning mechanism 5 is controlled to spray atomized deionized water onto the grinding area for 2 seconds to clean the generated dust; next, the air jet component is activated to blow the surface for 3 seconds to achieve rapid drying; then, the putty component 42 is switched to evenly apply putty and level it; then, the paint spray component 43 is activated to spray the primer and color paint; after spraying, the drying component 6 is activated to perform infrared heating and curing; finally, the polishing component completes the fine polishing. The entire repair operation is automatically scheduled by the control mechanism according to a preset time sequence, without the need for manual intervention.

[0044] Understandably, when the detection mechanism 3 is positioned above the central area of ​​the damaged paint surface on the vehicle, in step S1, it is controlled to acquire the damage depth and area. In step S2, the damage parameters are determined to be within a preset range that can be automatically repaired. In step S3, the moving mechanism 2 is driven to precisely move the repair mechanism 4 to the damaged location. In step S4, the repair mechanism 4 is controlled to perform a complete repair operation encompassing grinding, cleaning, filling, spraying, drying, and polishing. Through the coordinated execution of the above four steps, the paint inspection robot achieves a fully automated closed loop from defect identification to high-quality repair. This method completely abandons the traditional model that relies on manual judgment and disassembly and transportation, not only significantly shortening the repair cycle and improving operational consistency, but also fundamentally eliminating the risk of structural damage and sealing failure caused by disassembly and assembly. All processes are completed in the original assembled state of the vehicle, ensuring that the repair area and the original paint surface are highly matched in terms of geometry, optics, and mechanical properties, effectively maintaining the vehicle's residual value and user satisfaction, and providing an efficient, reliable, and high-precision intelligent solution for the automotive aftermarket repair field.

[0045] In some embodiments, the detection mechanism 3 includes a vision sensor, and the step of controlling the detection mechanism 3 to acquire damage parameters of the vehicle's damaged paint surface includes: S11. Control the vision sensor to acquire images of the damaged paint surface of the vehicle to obtain the original image; For example, the control mechanism activates the high-resolution color industrial camera (2448×2048 resolution, 30 frames / second) in the detection mechanism 3 to acquire a single-frame image of the damaged paint surface of the vehicle, obtaining the original image. This image includes the scratches, dents, and the surrounding intact paint area.

[0046] S12. Based on the original image, perform mild bilateral filtering, non-local mean denoising, and guided filtering sequentially according to preset filtering rules to obtain the denoised image. For example, the control mechanism invokes the image preprocessing module to perform three-stage filtering on the original image: First, a light bilateral filtering is performed (spatial domain standard deviation σ_s=2.0, gray-level domain standard deviation σ_r=30) to preserve edges while smoothing textures; then, nonlocal mean denoising is used (search window 21×21, similarity window 7×7) to suppress random salt-and-pepper noise; finally, guided filtering is applied (guided image is the original image, radius r=5, regularization parameter ε=0.01) to further eliminate illumination gradient artifacts. After this processing, the denoised image is obtained, with a signal-to-noise ratio improved by approximately 12 dB, while maintaining good edge sharpness.

[0047] S13. Based on the denoised image, the closed boundary of the damaged area is identified by the contour registration algorithm, and the connected region within the closed boundary is filled by the region growing algorithm to obtain the damaged area mask. For example, the control mechanism converts the denoised image into a grayscale image and uses the Canny operator combined with morphological closing operation to extract the initial edges; then it calls the contour registration algorithm (based on minimum bounding rectangle constraint and curvature continuity verification) to identify the closed boundary of the damaged area; on this basis, using any pixel within the boundary as a seed point, it executes the region growing algorithm (the growing condition is grayscale difference ≤ 15) to fill all connected pixels and generate a binarized damaged area mask.

[0048] S14. Based on the damage area mask, obtain the damage area of ​​the vehicle's damaged paint surface; For example, the control mechanism counts the total number of pixels with a value of 1 in the mask of the damaged area, multiplies it by the actual physical area corresponding to a single pixel (obtained by camera calibration, which is 0.04 square millimeters / pixel), and calculates that the damaged area is 18 square centimeters.

[0049] S15. Based on each pixel within the mask coverage area of ​​the damaged region in the denoised image, determine the brightness distribution data, which includes the brightness value of each pixel. For example, the control mechanism extracts all pixels covered by the mask of the damaged area in the denoised image, reads their Y channel (luminance component) values, and forms a luminance distribution dataset containing the luminance value of each pixel (range 0–255).

[0050] S16. Based on the color distribution within the mask coverage area of ​​the damaged region in the denoised image, identify the set of the darkest color regions and the set of the lightest color regions. For example, the control mechanism performs cluster analysis on the brightness distribution data to identify the top 5% of pixels with the highest brightness values ​​as the set of darkest colored regions (actually the recessed bottom shadow area), and the top 5% of pixels with the lowest brightness values ​​as the set of lightest colored regions (actually the raised reflective area). Here, "darkest colored" and "lightest colored" correspond to low brightness and high brightness, respectively, reflecting the three-dimensional morphological characteristics of the damaged area.

[0051] S17. Based on the center coordinates of the high-brightness region and the center coordinates of the low-brightness region, control the distance measurement of the center point of the high-brightness region and the center point of the low-brightness region to obtain a first depth value and a second depth value, wherein the brightness value of the pixel in the high-brightness region is greater than the first threshold, and the brightness value of the pixel in the low-brightness region is less than the second threshold. For example, the control mechanism sets a first threshold of 220 and a second threshold of 80. Pixels with a brightness value > 220 are defined as high-brightness areas, and their center coordinates are calculated to be (1250, 980); pixels with a brightness value < 80 are defined as low-brightness areas, and their center coordinates are calculated to be (1265, 995). Subsequently, the control mechanism calls the infrared ranging module in the detection mechanism 3 to perform non-contact distance measurements on these two center points: the distance from the center point of the high-brightness area to the sensor is measured to be 302.5 mm, and the distance from the center point of the low-brightness area to the sensor is measured to be 303.2 mm. Since the ideal plane of the vehicle body surface should be at the same depth, the difference between the two values ​​reflects the depth of the depression. Based on this, the system calculates the first depth value to be 302.5 mm and the second depth value to be 303.2 mm.

[0052] S18. Based on the first depth value and the second depth value, determine the damage depth of the vehicle's damaged paint surface.

[0053] For example, the control mechanism calculates the absolute difference between the first depth value and the second depth value: |303.2 302.5| = 0.7 mm, and combined with the vehicle body curvature compensation factor (provided by the whole vehicle CAD model, which is 0.95 in this area), the damage depth of the vehicle's damaged paint surface is finally determined to be 0.7 × 0.95 ≈ 0.67 mm.

[0054] Understandably, by controlling the visual sensor to acquire the original image, step S12 sequentially performs mild bilateral filtering, non-local mean denoising, and guided filtering to obtain a denoised image. In step S13, a damage area mask is generated by combining contour registration and region growing algorithms. In step S14, the damage area is calculated based on the mask. In step S15, the brightness distribution data of pixels within the mask is extracted. In step S16, the sets of the darkest and lightest color regions are identified. In step S17, the first and second depth values ​​are obtained based on infrared ranging of the center points of the high and low brightness regions. Finally, in step S18, the damage depth is determined accordingly. These eight sub-steps collaboratively construct an efficient and low-cost visual-infrared fusion damage quantification mechanism. This scheme overcomes the limitations of a single sensor in terms of accuracy or cost, utilizing image semantics to guide sparse depth measurement, significantly improving the accuracy of geometric parameter extraction for shallow paint surface defects. The resulting damage depth and area data provide a reliable basis for subsequent repair path planning and material supply, ensuring that sanding is not excessive, putty does not overflow, and spraying does not accumulate, fundamentally guaranteeing the quality consistency and appearance restoration of in-situ repairs.

[0055] In some embodiments, step S4 includes: Based on the damage region mask and region growth algorithm, the filling order and growth direction inside the damage region are determined to obtain the region growth direction information. For example, the control mechanism performs morphological analysis on the mask of the damaged area to identify its principal axis direction (determined by the long side of the smallest bounding rectangle, which forms a 15° angle with the longitudinal direction of the vehicle body). Then, using the point with the lowest brightness within the mask (i.e., the deepest depression) as the seed point, an improved region growing algorithm is executed: during the growing process, expansion is prioritized along the principal axis direction, and the offset vector of each newly added pixel relative to the seed point is recorded. Based on this, the system determines the filling order within the damaged area to be "from the center to the edge, reciprocating along the principal axis," and the growth direction information is represented as a unit vector d = (cos15°, sin15°).

[0056] Based on the depth values ​​of multiple sampling points within the damaged area, depth distribution feedback data is generated; For example, the control mechanism extracts the depth values ​​(range 302.3–303.5 mm) of 12 uniformly distributed sampling points within the damaged area from the infrared ranging data obtained in step S17, and calculates the relative indentation amount of each point in conjunction with the ideal curved surface model of the vehicle body to generate depth distribution feedback data. This data is stored in the form of a two-dimensional array, with each element containing coordinates (x, y) and the corresponding indentation depth δ (unit: mm). For example, δ = 0.67 mm at the center point (1258, 988) and δ = 0.12 mm at the edge point (1280, 1010).

[0057] Based on the growth direction information of the region and the depth distribution feedback data, the area covered by the mask in the damaged region is calibrated point by point to obtain the calibration coordinate sequence. For example, the control mechanism divides the damaged area mask into 50×50 micrometer grid cells. For each grid center point covered by the mask, its access sequence number in the reciprocating path is determined based on the region's growth direction information, and a Z-axis compensation value (i.e., the displacement corresponding to the grinding pressure) is assigned based on depth distribution feedback data. Finally, all grid center points are arranged in sequence to form a calibration coordinate sequence containing three-dimensional coordinates: , where z i = Reference height k·δ i (k is the process coefficient, taken as 0.8).

[0058] Based on the calibration coordinate sequence, the reciprocating operation trajectory is obtained; For example, the control mechanism performs trajectory smoothing on the calibration coordinate sequence: fifth-order polynomial interpolation is used to connect adjacent points, and the maximum radius of curvature is constrained to ≥50 mm to ensure smooth movement of the moving mechanism 2; simultaneously, the unidirectional path is converted into a zigzag reciprocating work trajectory—after completing each line of forward scanning, it is laterally offset by one tool head width (8 mm) and scans back in the opposite direction until the entire damaged area is covered. The final output reciprocating work trajectory includes position, velocity, acceleration, and Z-axis compensation commands.

[0059] Based on the reciprocating work trajectory, a motion command is sent to the moving mechanism 2 to drive the repair mechanism 4 to move along the reciprocating work trajectory and perform the repair operation.

[0060] For example, the control mechanism decomposes the reciprocating work trajectory into real-time motion commands, which are sent to the six-axis servo drive of the moving mechanism 2 via the EtherCAT bus. The moving mechanism 2 responds to the commands, driving the repair mechanism 4 (currently the grinding part 41) to move precisely along the trajectory: the XY plane travels along the reciprocating path, and the Z-axis dynamically adjusts the downward pressure (maximum 0.5 mm) according to the depth distribution, achieving adaptive grinding of "deep areas require more grinding, shallow areas require less grinding." The entire process lasts 45 seconds, after which it automatically switches to the cleaning mechanism 5 to perform water-air spraying operations.

[0061] Understandably, the process involves determining the filling order and growth direction based on a damaged area mask and a region growth algorithm to obtain region growth direction information. Depth distribution feedback data is generated based on the depth values ​​of multiple sampling points within the damaged area. The damaged area is then calibrated point-by-point based on the region growth direction information and the depth distribution feedback data to obtain a calibration coordinate sequence. A reciprocating work trajectory is generated based on this calibration coordinate sequence, and motion commands are sent to the moving mechanism 2 to drive the repair mechanism 4 to perform repair operations along this trajectory. Through the coordinated execution of these five sub-steps, the paint repair robot achieves a leap from static defect identification to dynamic adaptive operation. This solution deeply integrates the geometric shape of the damage with three-dimensional depth information, generating a spatial trajectory that combines full coverage and process adaptability, enabling the repair mechanism 4 to "intelligently follow" the damage morphology for differentiated processing. This not only completely avoids the problems of missed or over-grinding in manual polishing but also significantly improves repair efficiency and surface consistency, providing core technical support for achieving repair quality comparable to original factory paint in the vehicle's original condition.

[0062] In some embodiments, the repair mechanism 4 includes a grinding element 41; step S4 includes: Based on the reciprocating work trajectory, the moving mechanism 2 is controlled to drive the polishing part 41 to perform polishing operation on the damaged paint surface of the vehicle along the reciprocating work trajectory; For example, the control mechanism sends the reciprocating work trajectory to the mobile mechanism 2, driving it to move the grinding component 41 (equipped with a P800 grit sandpaper disc, rotating at 2500 rpm) along a preset path to perform a grinding operation on the damaged paint surface of the vehicle. During the grinding process, the mobile mechanism 2 provides real-time feedback on the end position and posture, ensuring that the Z-axis downward pressure is dynamically adjusted according to the depth distribution, thereby prioritizing the removal of dented areas.

[0063] Based on the image of the polishing area collected in real time by the vision sensor during the polishing operation, the brightness value of each pixel in the image is extracted, and it is determined whether there is a continuous area with a brightness value greater than a preset gloss threshold, thus obtaining a first flatness determination result; at the same time, based on the distance values ​​measured by the infrared ranging sensor at multiple sampling points in the polishing area, it is determined whether the distance of each sampling point is within a preset thickness tolerance range, thus obtaining a second flatness determination result. For example, while polishing is underway, the control mechanism simultaneously activates the vision sensor in the detection mechanism 3 to acquire real-time images of the polished area at a frequency of 10 frames per second. For each frame, the RGB channels are extracted and converted into luminance values ​​(Y = 0.299R + 0.587G + 0.114B) to form a luminance matrix. The system sets a preset gloss threshold of 200 (8-bit image) and traverses the matrix to determine whether there exists a continuous region (8-neighborhood connectivity) with an area ≥ 5 square millimeters and all pixel luminance values ​​> 200. In this example, a continuous bright region of 6.2 square millimeters is detected at the 32nd second, and the first flatness determination result is "a continuous bright region exists".

[0064] Simultaneously, the control mechanism dispatches infrared ranging sensors to perform non-contact ranging on 9 grid sampling points (uniformly distributed in 3×3) within the grinding area, obtaining a distance value sequence: [302.8, 302.9, 302.7, 302.8, 302.8, 302.9, 302.7, 302.8, 302.8] mm. The preset thickness tolerance range is ±0.1 mm (based on an ideal body surface curvature of 302.8 mm). All ranging values ​​fall within the [302.7, 302.9] interval; therefore, the second flatness judgment result is "all sampling point distances are within the thickness tolerance range".

[0065] Based on the first flatness determination result that there is a continuous bright area and the second flatness determination result that the distance between all sampling points is within the thickness tolerance range, a polishing completion signal is generated; For example, the control mechanism performs a logical AND operation on the two judgment results: since the first flatness judgment result is "there is a continuous bright area" (indicating that a uniform reflective layer has been formed on the surface), and the second flatness judgment result is "the distance between all sampling points is within the thickness tolerance range" (indicating that the geometric shape has been flat), the system generates a polishing completion signal.

[0066] Based on the polishing completion signal, the cleaning mechanism 5 is controlled to clean the damaged paint surface of the vehicle.

[0067] For example, in response to the sanding completion signal, the control mechanism immediately stops the rotation of the sanding part 41 and instructs the moving mechanism 2 to raise the repair mechanism 4 to a safe height; then, the cleaning mechanism 5 is activated: first, the water spraying part is controlled to spray atomized deionized water onto the sanding area for 2 seconds to moisten and wash away residual dust; then, the air spraying part is controlled to blow at a pressure of 0.3 MPa for 3 seconds to thoroughly remove moisture and mud residue, so that the paint surface is restored to a clean and dry state, and a qualified base is prepared for the subsequent puttying process.

[0068] Understandably, the reciprocating trajectory control mechanism 2 drives the grinding component 41 to perform grinding operations. During the grinding process, a visual sensor extracts brightness values ​​in real time and determines whether there are continuous high-brightness areas exceeding a preset gloss threshold to obtain the first flatness judgment result. Simultaneously, an infrared ranging sensor detects whether the distance between multiple sampling points is within a preset thickness tolerance range to obtain the second flatness judgment result. When both judgment results meet the conditions, a grinding completion signal is generated, which triggers the cleaning mechanism 5 to perform cleaning and drying operations. Through the coordinated execution of the above four sub-steps, the paint repair robot constructs a multimodal, closed-loop grinding quality verification mechanism. This solution abandons the traditional extensive control mode that relies on fixed time or manual visual inspection. Through the dual constraints of optical reflection characteristics and three-dimensional geometric accuracy, it accurately identifies the grinding endpoint, ensuring that the substrate has neither microscopic undulations nor uniform gloss. Thus, it not only avoids material waste and rework risks caused by over-grinding or under-grinding, but also provides highly reliable input for subsequent filling and painting processes, fundamentally ensuring the process integrity and appearance restoration of the whole vehicle in situ repair.

[0069] In some embodiments, after the step of controlling the cleaning mechanism 5 to clean the damaged paint surface of the vehicle, the method further includes: The vision sensor and the infrared ranging sensor are controlled to scan the damaged paint surface of the vehicle to obtain the detection image and depth data after polishing. For example, the control mechanism drives the moving mechanism 2 to reposition the detection mechanism 3 directly above the polishing area, activates the vision sensor to acquire a high-resolution image as the post-polishing detection image, and simultaneously schedules the infrared ranging sensor to perform a 9-point scan of the same area to obtain post-polishing depth data: the distance values ​​of each point are [302.85, 302.90, 302.78, 302.82, 302.88, 302.91, 302.79, 302.84, 302.86] mm.

[0070] Based on the post-polishing detection image and the post-polishing depth data, S13 to S18 are repeated to obtain the post-polishing damage area and post-polishing damage depth. For example, the control mechanism invokes the same image and depth processing flow as steps S13 to S18: First, perform mild bilateral filtering, non-local mean denoising, and guided filtering on the polished detection image; then, identify the residual defect boundary through the contour registration algorithm, and generate a new damage area mask by combining it with the region growing algorithm; based on the mask, calculate the polished damage area as 2.3 square centimeters; at the same time, based on the pixel brightness distribution within the mask and the infrared ranging extreme points, calculate the polished damage depth as 0.18 millimeters.

[0071] Based on the fact that the damaged area after grinding is greater than a preset area value or the damaged depth after grinding is greater than a preset flatness threshold, a local re-grinding command is generated, and a local re-grinding trajectory is regenerated based on the mask of the current damaged area after grinding, wherein the local re-grinding trajectory is different from the reciprocating operation trajectory. For example, the control mechanism reads a preset area value of 2.0 square centimeters and a preset flatness threshold (i.e., the maximum allowable residual depth) of 0.15 millimeters. Since both the damaged area (2.3 cm² > 2.0 cm²) and the damaged depth (0.18 mm > 0.15 mm) exceed the tolerances after grinding, the system determines that the repair is substandard and generates a local re-grinding command. Subsequently, based on the current mask of the damaged area after grinding (covering only a small unground area at the original damage center), a spiral filling strategy is used to replan the trajectory: starting from the mask's centroid, an Archimedean spiral path is generated from the inside out in a counter-clockwise direction, with a step size of 60% (4.8 mm) of the tool head radius, forming a local re-grinding trajectory. This trajectory is significantly different from the original reciprocating operation trajectory and is specifically designed for small-area fine finishing.

[0072] Based on the local re-grinding trajectory, the moving mechanism 2 is controlled to drive the grinding part 41 to move along the local re-grinding trajectory and perform the repair operation.

[0073] For example, the control mechanism sends the local regrinding trajectory to the moving mechanism 2, driving it to slowly move the grinding part 41 (switching to a finer-grit P1200 sandpaper disc, with the rotation speed reduced to 1800 rpm) along the spiral path. The Z-axis downward pressure is reduced to 50% of its original value, gently grinding the remaining high points. The entire regrinding process lasts 18 seconds. After completion, cleaning and re-inspection are triggered again until both indicators meet the requirements.

[0074] Understandably, after the cleaning operation is completed, the vision sensor and infrared ranging sensor are used to rescan the polished area to obtain the post-polishing detection image and depth data. The processing steps S13 to S18 are repeated to obtain the post-polishing damage area and depth. When any indicator exceeds the preset tolerance, a local re-polishing command is generated, and a local re-polishing trajectory (such as a spiral path) different from the original reciprocating trajectory is generated based on a new mask. Then, the polished part 41 is controlled to perform targeted repair. Through the coordinated execution of the above four sub-steps, the paint repair robot achieves a complete quality closed loop of "detection-execution-verification-correction". This solution breaks through the inherent limitations of open-loop polishing. By triggering adaptive re-polishing through high-precision re-inspection, it ensures that the final substrate geometry and optical performance meet the standards. The differentiated design of the local re-polishing trajectory takes into account both efficiency and precision. It avoids material waste caused by global rework and eliminates the retention of micro-defects, providing a near-ideal flat substrate for subsequent filling and spraying, significantly improving the first-pass yield and appearance consistency of the whole vehicle in-situ repair.

[0075] In some embodiments, during the step of the control and repair mechanism 4 performing the repair operation on the damaged paint surface of the vehicle, when the repair mechanism 4 performs the putty application operation, the following step is further included: S501. Control the moving mechanism 2 to move to the placement position of the putty scraper 42, pick up the putty scraper 42 and then reset it; S502. Based on the preset scraping trajectory, control the moving mechanism 2 to drive the putty scraping component 42 to perform putty scraping operation on the damaged paint surface of the vehicle; S503. During the putty application process, the distance between the scraper and the damaged paint surface of the vehicle is detected in real time by an infrared distance sensor, and the pressure applied by the scraper is determined based on the distance. S504. Simultaneously acquire real-time images and infrared depth data of the scraping area, determine the gloss uniformity of the putty layer based on the color depth and brightness values ​​of the real-time images, and determine whether the thickness of the putty layer is within the preset tolerance range based on the infrared depth data. S505. When it is determined that the putty layer is completely and smoothly applied, the moving mechanism 2 is controlled to return the putty scraper 42 to its original position, thus completing the reset.

[0076] It is understood that this embodiment uses the control mechanism of the paint repair robot as the main body to perform the fully automated operation of the putty application stage. Before proceeding to this embodiment, the sanding and cleaning processes have been completed, the damaged paint surface of the vehicle is clean and dry, and the system has confirmed that it has entered the filling stage. The following sub-steps are executed sequentially, and subsequent steps directly use the results obtained from the preceding steps, without repeating or interpreting the already defined data.

[0077] By controlling the moving mechanism 2 to automatically pick up and return the putty scraper 42, the scraper pressure is adjusted in real time based on infrared ranging during the scraping process. Simultaneously, the color depth and brightness of the visual image are used to judge the gloss uniformity, and infrared depth data is used to judge the putty layer thickness. The system only resets after confirming that the application is complete and smooth. These five sub-steps work together to construct a high-precision, closed-loop automatic putty scraping control mechanism. This solution completely abandons the operation mode that relies on manual experience. Through multi-sensor fusion and closed-loop adjustment of process parameters, it ensures uniform putty layer thickness, a smooth surface, and clear edges. This not only avoids shrinkage cracking caused by excessively thick putty or insufficient coverage caused by excessively thin putty, but also prevents material overflow from contaminating the surrounding original paint surface, providing a high-quality base for subsequent painting processes and significantly improving the reliability and appearance restoration of the vehicle's in-situ repair.

[0078] For example, Specific implementation of step S501: The control mechanism generates a pick-up command, driving the moving mechanism 2 to move along a preset path to the placement mechanism 7 at the top of the cabinet 1, where a dedicated slot for the putty scraper 42 is provided. The end effector of the moving mechanism 2 precisely picks up the putty scraper 42 (its front end is a flat stainless steel scraper, and its rear end integrates a micro screw metering pump) through an electromagnetic adsorption device, then lifts it up and returns it to the starting position of the operation, completing the reset.

[0079] Specific implementation of step S502: The control mechanism invokes a scraping trajectory generated based on the mask of the damaged area after sanding—this trajectory is a reciprocating filling path with a row spacing of 80% (6.4 mm) of the scraper width. The moving mechanism 2 drives the putty scraper 42 to move along this trajectory, while simultaneously activating the metering pump to extrude the putty at a rate of 0.12 ml / s. The scraper adheres to the paint surface at a constant angle (15°) to achieve uniform scraping.

[0080] Specific implementation of step S503: During the scraping process, the control mechanism continuously measures the distance between the scraper tip and the vehicle body surface using an infrared ranging sensor. The initial target distance is set at 2.0 mm (corresponding to a standard putty layer thickness of 0.8 mm). If the measured distance decreases to 1.8 mm, it indicates that the scraper pressure is too high, and the system automatically reduces the Z-axis driving force; if the distance increases to 2.3 mm, it indicates insufficient pressure, and the downward pressure is increased. This achieves closed-loop control of the scraper pressure. Optionally, a miniature pressure sensor can also be integrated on the putty scraper component 42 to directly feedback the contact force and fuse it with infrared data for verification.

[0081] Specific implementation of step S504: The control mechanism simultaneously activates the visual sensor and infrared ranging module: the visual sensor acquires two frames of images of the scraped area per second, extracts RGB values, and calculates color depth (CIE Labs). The gloss is determined by the mean ΔE (exposure angle) and brightness in space; if ΔE < 3 and the standard deviation of brightness < 10, the gloss is considered uniform; the infrared module measures the distance at 9 sampling points, and the putty layer thickness is calculated by combining it with the ideal curved surface model. If the thickness at all points is within the range of 0.7–0.9 mm, the thickness is considered acceptable. Only when both conditions are met is the putty considered "completely applied and smooth".

[0082] Specific implementation of step S505: Upon receiving feedback that the putty application is complete and smooth, the control mechanism stops the metering pump and movement, and drives the moving mechanism 2 to smoothly transport the putty piece 42 back to the placement mechanism 7 at the top of the cabinet 1. The electromagnetic adsorption device is de-energized and released, completing its reset. The system records the status as "putty application complete," ready to enter the spraying stage.

[0083] In some embodiments, during the step of the control and repair mechanism 4 repairing the damaged paint surface of the vehicle, when the repair mechanism 4 performs a painting operation, the following step is further included: S601, Control the moving mechanism 2 to move to the placement position of the paint spraying part 43, pick up the paint spraying part 43 and then reset it; S602. Based on the preset spraying trajectory, control the moving mechanism 2 to drive the paint spraying part 43 to perform a paint spraying operation on the damaged paint surface of the vehicle. S603. During the painting operation, real-time images and infrared depth data of the spraying area are collected simultaneously; S604. Determine the gloss uniformity of the paint film based on the color depth and brightness values ​​of the real-time image, and determine whether the thickness of the paint film is within the preset tolerance range based on the infrared depth data. S605. When it is determined that the paint film is completely and evenly applied, the moving mechanism 2 is controlled to return the painted part 43 to its original position, thus completing the reset.

[0084] It is understood that this embodiment uses the control mechanism of the paint repair robot as the main body to perform the fully automated operation of the painting stage. Before proceeding to this embodiment, the puttying and drying processes have been completed, the putty layer has been cured and the surface is smooth, and the system has confirmed that it has entered the painting stage. The following sub-steps are executed sequentially, and subsequent steps directly use the results obtained from the preceding steps, without repeating the acquisition or interpretation of already defined data.

[0085] By controlling the moving mechanism 2 to automatically pick up and return the paint-spraying parts 43, real-time images and infrared depth data are simultaneously collected during the spraying process. The uniformity of the paint film gloss is judged based on color depth and brightness values, and the paint film thickness is judged based on infrared ranging to determine whether it meets the process tolerance. Resetting is only completed after confirming that the spraying is complete and smooth. These five sub-steps collaboratively construct a closed-loop, high-precision automatic paint spraying control mechanism. This solution breaks through the limitations of traditional open-loop spraying that relies on experience-based adjustments. It utilizes multimodal sensing to achieve online quantitative evaluation of the wet film state, ensuring that each layer of paint film has uniform thickness, consistent color, and no defects. This not only significantly improves the first-pass yield rate but also avoids runs caused by excessively thick paint films or exposed substrate caused by excessively thin films, providing core technical support for achieving paint repair effects comparable to the original factory finish in the vehicle's original condition.

[0086] For example, Specific implementation of step S601: The control mechanism generates a pick-up command, driving the moving mechanism 2 to move along a preset path to the slot position of the paint spraying part 43 in the placement mechanism 7 at the top of the cabinet 1. The paint spraying part 43 is a low-pressure atomizing spray gun, integrating a two-component automatic mixing valve and a micro flow meter. The end of the moving mechanism 2 precisely picks up the paint spraying part 43 through an electromagnetic adsorption interface, then lifts it up and returns it to the spraying start position, completing the reset.

[0087] Specific implementation of step S602: The control mechanism invokes a spraying trajectory generated based on the mask of the damaged area—this trajectory is a multi-layer reciprocating scanning path, with the row spacing of each layer being 70% of the spray width (approximately 14 mm), and a total of three layers are sprayed (one layer of primer and two layers of color paint). The moving mechanism 2 drives the painting part 43 to move at a uniform speed along the trajectory, while simultaneously opening the air path and paint path according to preset parameters: the nozzle is 150 mm away from the paint surface, the paint output is 0.8 ml / s, and the atomizing air pressure is 0.25 MPa, ensuring uniform paint mist coverage.

[0088] Specific implementation of step S603: During the spraying process, the control mechanism simultaneously activates the vision sensor and infrared ranging sensor in the detection mechanism 3: the vision sensor captures images of the wet film state at a frequency of 5 frames per second; the infrared ranging sensor continuously measures the distance of 9 fixed sampling points in the spraying area to obtain data on the cumulative thickness change of the paint film.

[0089] Specific implementation of step S604: The control unit performs CIE Lab color space analysis on the real-time images, calculating the color depth deviation ΔE and brightness standard deviation within the sprayed area. If ΔE < 2 (indicating no visible color difference) and brightness standard deviation < 8, the gloss is considered uniform. Simultaneously, based on infrared ranging data and subtracting the putty layer reference height, the paint film thickness at each point is calculated. If the thickness at all points is within the range of 25–35 micrometers (corresponding to the single-layer standard), the thickness is considered acceptable. Only when both indicators are met is the system deemed "painting complete and smooth".

[0090] Specific implementation of step S605: After receiving feedback that the paint application is complete and smooth, the control mechanism shuts off the paint and air paths of the painted part 43, drives the moving mechanism 2 to smoothly transport the painted part 43 back to the designated placement position at the top of the cabinet 1, and de-energizes the electromagnetic adsorption device, completing the reset. The system records the status as "painting complete," ready to proceed to the next drying or polishing process.

[0091] In some embodiments, after the step of the control repair mechanism 4 performing the repair operation on the damaged paint surface of the vehicle, a curing stage is further included, the curing stage including the following steps: S701, the infrared curing lamps on both sides of the control cabinet 1 are turned on to perform light curing treatment on the sprayed area. S702. During the curing process, real-time images and infrared depth data of the curing area are collected simultaneously; S703. Determine the gloss change trend of the paint film based on the color depth and brightness values ​​of the real-time image, and determine whether the paint film thickness is stable based on the infrared depth data. S704. When it is determined that the gloss of the paint film reaches the preset stable threshold and the thickness change rate is lower than the preset convergence threshold, a curing completion signal is generated and the infrared curing lamp is turned off.

[0092] It is understood that this embodiment uses the control mechanism of the paint repair robot as the main body to execute a post-treatment process for the paint film based on photocuring. Before proceeding to this embodiment, the painting operation has been completed, and the damaged area of ​​the vehicle has been covered with a putty layer made of photocurable putty and matching hardener, as well as a two-component color paint. The system confirms that it has entered the curing stage. The following sub-steps are executed sequentially, and subsequent steps directly use the results obtained from the preceding steps, without repeating or interpreting already defined data.

[0093] The infrared curing lamps on both sides of the control cabinet 1 perform photocuring on the sprayed area. During the curing process, real-time images and infrared depth data are simultaneously acquired, and a dual judgment is made based on the gloss change trend and the stability of the paint film thickness. The light source is only turned off when both converge to a preset threshold. These four sub-steps work together to construct an intelligent, closed-loop photocuring control mechanism. This solution fully utilizes the rapid reaction characteristics of photocurable materials, combined with non-contact online monitoring, to achieve accurate identification of the curing endpoint, avoiding insufficient adhesion due to under-curing or embrittlement and cracking caused by over-curing. Therefore, it not only significantly shortens the curing cycle (by more than 60% compared to traditional hot air), but also ensures that the repaired paint film is consistent with the original factory in terms of optical and mechanical properties, providing an efficient and reliable post-processing guarantee for high-quality in-situ repair of the entire vehicle.

[0094] In some embodiments, during the step of the control repair mechanism 4 repairing the damaged paint surface of the vehicle, when the repair mechanism 4 performs a polishing operation, the following step is further included: S801. Control the moving mechanism 2 to move to the placement position of the polished part, pick up the polished part and then reset; S802. Based on a preset polishing trajectory, control the moving mechanism 2 to drive the polishing part to perform polishing operation on the damaged paint surface of the vehicle; S803. During the polishing operation, the distance between the polishing wheel and the damaged paint surface of the vehicle is detected in real time by an infrared ranging sensor, and the polishing pressure is determined based on the distance. S804. Simultaneously acquire real-time images and infrared depth data of the polishing area, determine the gloss of the paint surface based on the color depth and brightness values ​​of the real-time images, and determine whether the paint film thickness is within the safe margin range based on the infrared depth data. S805. When it is determined that the gloss of the paint surface reaches the preset target value and the paint film thickness is not lower than the minimum protection threshold, the moving mechanism 2 is controlled to put the polished part back to its placement position to complete the reset.

[0095] It is understood that this embodiment uses the control mechanism of the paint surface repair robot as the main body to perform the fully automated operation of the polishing stage. Before entering this embodiment, the curing stage has been completed, the paint film has been fully cross-linked and hardened, and the system confirms that it has entered the final surface finishing stage. The following sub-steps are executed sequentially, and subsequent steps directly use the results obtained from the previous steps, without repeating the acquisition or interpretation of already defined data.

[0096] By controlling the moving mechanism 2 to automatically pick up and return polishing parts, the polishing pressure is adjusted in real time based on infrared ranging during the polishing process. Simultaneously, the color depth and brightness of the visual image are used to determine the paint gloss, and infrared depth data is used to determine the remaining paint film thickness. Resetting is only completed when the gloss meets the standard and the thickness is not below the safety threshold. These five sub-steps work together to construct a safe, precise, and closed-loop automatic polishing control mechanism. This solution breaks through the limitations of traditional reliance on experience and visual judgment, achieving refined operation of "brightening without damaging the paint" through multi-sensor fusion. Therefore, it not only avoids the risk of penetrating the original clear coat layer due to over-polishing, but also ensures that the repaired area is optically highly consistent with the surrounding paint surface, providing a final high-quality guarantee for in-situ vehicle repair, significantly improving user satisfaction and repair professionalism.

[0097] For example, Specific implementation of step S801: The control mechanism generates a pick-up command, driving the moving mechanism 2 to move along a preset path to the polishing part slot in the placement mechanism 7 at the top of the cabinet 1. The polishing part includes a replaceable wool wheel (for coarse polishing) or a microfiber disc (for fine polishing), with a brushless motor integrated at the rear (speed infinitely adjustable from 0–3000 rpm). The moving mechanism 2 precisely picks up the polishing part via an electromagnetic adsorption interface, then lifts it and returns it to the polishing starting position, completing the reset.

[0098] Specific implementation of step S802: The control mechanism invokes a polishing trajectory generated by expanding the original damaged area mask by 10%—this trajectory is a concentric spiral path that covers the entire repair and transition area from the inside out, ensuring no "halo" at the edges. The moving mechanism 2 drives the polished part to move at a constant speed along the trajectory, while the motor starts running at 2200 rpm and applies initial downward pressure.

[0099] Specific implementation of step S803: During polishing, the control mechanism uses an infrared ranging sensor to continuously measure the distance between the polishing wheel surface and the vehicle body paint. The target working distance is set at 3.0 mm (corresponding to standard polishing pressure). If the measured distance decreases to 2.7 mm, it indicates excessive downward pressure, which may excessively wear down the paint film, and the system automatically increases the Z-axis height. If the distance increases to 3.3 mm, it indicates insufficient contact, and the height is reduced to increase the cutting force. Optionally, a miniature pressure sensor can also be integrated on the polished part to directly feedback the contact force, fusing it with infrared data to achieve more precise pressure closed-loop control.

[0100] Specific implementation of step S804: The control mechanism simultaneously activates the visual sensor and infrared ranging module: the visual sensor acquires one frame of high dynamic range image per second, calculates the CIE L value (brightness) and specular reflectance; if L ≥ 92 and the standard deviation of reflectance < 5%, the gloss level is deemed satisfactory; simultaneously, the infrared module measures the distance at 9 sampling points, and combined with the initial paint film thickness (approximately 30 micrometers), determines whether the current remaining thickness is ≥ 15 micrometers (minimum protection threshold). Polishing is considered "completed" only when the gloss level is satisfactory and the thickness is safe.

[0101] Specific implementation of step S805: Upon receiving the "polishing complete" feedback, the control mechanism stops the polishing motor and drives the moving mechanism 2 to smoothly transport the polished part back to the designated placement position at the top of cabinet 1. The electromagnetic adsorption device is de-energized and released, completing the reset. The system records the status as "polishing complete" and sends a successful repair report to the user terminal.

[0102] In some embodiments, after the control and repair mechanism 4 completes the repair operation, the following steps are further included: S901. Control the detection mechanism 3 to perform a final inspection scan on the repaired vehicle paint damage surface to obtain the final inspection image and final inspection depth data; S902. Based on the final inspection image, calculate the color difference value between the repair area and the surrounding original paint surface; based on the final inspection depth data, calculate the depth difference between the deepest point and the shallowest point within the repair area. S903. If the color difference value is greater than the preset color difference threshold or the depth difference value is greater than the preset flatness threshold, the repair is deemed unqualified, a manual intervention instruction is generated and reported to the host computer. S904. If the color difference value is not greater than the preset color difference threshold and the depth difference value is not greater than the preset flatness threshold, then the repair is deemed qualified, a test report containing all process parameters is generated, and the repair data is stored. S905. Control the moving mechanism 2 to perform an automatic reset and output a "repair completed" prompt signal.

[0103] It is understood that this embodiment primarily utilizes the control mechanism of the paint repair robot to perform the final quality assessment, system decision-making, and overall machine reset after the repair is completed. Before proceeding to this embodiment, the polishing stage has been completed, all repair procedures are finished, and the damaged paint surface of the vehicle is under natural light. The following sub-steps are executed sequentially, and subsequent steps directly use the results obtained from the preceding steps, without repeating or interpreting already defined data.

[0104] After repair, the system controls the inspection mechanism 3 to acquire final inspection images and depth data. Based on color difference and depth difference values, a two-dimensional quality assessment is performed. If the work is substandard, manual intervention is initiated; if it is satisfactory, a full-process inspection report is generated. Finally, the moving mechanism 2 is reset and a completion notification is output. These five sub-steps collaboratively construct an intelligent and traceable final inspection and system closure mechanism for repair quality. This solution overcomes the subjectivity and lag of traditional manual visual inspection, utilizing quantitative indicators to achieve objective and accurate quality acceptance. Therefore, it not only ensures that every repair meets original factory-grade appearance and morphology standards but also supports process iteration and accountability through structured data recording, providing complete closed-loop technical support for the digital and intelligent transformation of automotive aftermarket repair.

[0105] An exemplary implementation of step S901: The control mechanism drives the moving mechanism 2 to precisely position the detection mechanism 3 directly above the repair area, activates the vision sensor to acquire a high-resolution final inspection image (resolution 2448×2048), and simultaneously dispatches the infrared ranging sensor to perform a dense scan (sampling points ≥100) of the original paint surface within a 5 cm radius of the repair area to obtain the final inspection depth data.

[0106] Specific implementation of step S902: The control unit converts the final inspection image to the CIE Lab color space, selects a 5×5 pixel block at the center of the repaired area and an area of ​​the same size as the surrounding intact paint surface, and calculates the color difference value ΔE = √[(L1 L2)² + (a1 a2)² + (b1 b2)²], we get ΔE = 2.8; at the same time, we extract the maximum depth value of 302.92 mm and the minimum depth value of 302.85 mm in the repair area from the infrared data, and calculate the depth difference as 0.07 mm.

[0107] Specific implementation of step S903: The control mechanism reads the preset color difference threshold as 3.0 (the boundary is indistinguishable to the human eye) and the preset flatness threshold as 0.10 mm. Since ΔE = 2.8 ≤ 3.0 and the depth difference 0.07 mm ≤ 0.10 mm, this example is judged as qualified; however, if ΔE = 4.2 or the depth difference = 0.15 mm, the repair is judged as unqualified, the system generates a manual intervention command, and uploads the repair area coordinates, failed process, original image and other data to the host computer maintenance management platform through the Ethernet interface, and pops up a "Manual intervention required" alarm on the local HMI interface.

[0108] Specific implementation of step S904: Since the judgment in this example is qualified, the control mechanism automatically generates a structured inspection report, including: initial damage parameters, trajectories and parameters of each process, number of grinding / scraping / spraying / polishing times, final inspection color difference and depth difference, time consumption and material consumption, etc., and encrypts and stores the report and the original sensing data in the solid-state drive in cabinet 1 to support later traceability and process optimization.

[0109] Specific implementation of step S905: The control mechanism sends a reset instruction, the moving mechanism 2 zeros the end effector to the safe docking position, all tool heads have been placed back in the placement mechanism 7, and the door lock of cabinet 1 is closed. Subsequently, the system triggers the sound and light prompt to emit a short beep, displays "repair completed" on the operation screen, and sends a job completion signal to the workshop MES system to end this maintenance task.

[0110] In some embodiments, after the control mechanism controls the detection mechanism 3 to obtain the damage parameters of the damaged paint surface of the vehicle, it further includes: S2a. If multiple separated damage areas are detected, generate a regional priority sequence based on the damage area, damage depth and spatial distance of each damage area; S2b. Plan a multi-region serial-parallel hybrid repair path based on the regional priority sequence and the tool switching time of the repair mechanism 4; S2c. When performing the repair operation, complete the repair of each damage area in sequence according to the hybrid repair path, and reuse the same tool head between adjacent areas to reduce the switching times.

[0111] It can be understood that by identifying multiple damage areas and generating a priority sequence after detection, planning a serial-parallel hybrid repair path based on the tool switching cost, and reusing the tool head during execution to reduce switching; this mechanism significantly improves the operation efficiency in multi-defect scenarios. The traditional single-region repair mode starts and stops frequently and has redundant tool switching, while this solution realizes "one clamping, multi-region cooperation" through intelligent scheduling, greatly reducing the idle stroke and part replacement time while ensuring the repair quality of each region, and is especially suitable for the rapid processing of multi-point damage of accident vehicles, providing an expandable automation strategy for high-throughput repair workshops.

[0112] Exemplarily, in this embodiment, taking the control mechanism as the main body, two independent scratches are found on the car door during the detection stage: area A (area 15 cm², depth 0.4 mm), area B (area 8 cm², depth 0.2 mm), and the distance between the two is 35 cm.

[0113] Specific implementation of step S2a: The control mechanism calculates the priority score = area × depth / distance reference value, and gets the score of area A as 6.0 and the score of area B as 1.6, and generates a priority sequence: [A, B].

[0114] Specific implementation of step S2b: System assessment: Switching to sanding part 41 takes 8 seconds, while moving from A to B only takes 5 seconds. To reduce downtime, the planned path is: complete sanding of area A → move to area B → sanding of area B → return to area A for puttying → return to area B for puttying... that is, "grouping by process and continuous execution across areas", forming a hybrid path.

[0115] Specific implementation of step S2c: The mobile mechanism 2 first completes the sanding of areas A and B (using a single sanding piece 41), and then uniformly performs subsequent processes such as puttying and spraying. The number of tool switching times is reduced from 8 to 3, and the total operation time is shortened by 22%.

[0116] In some embodiments, prior to performing puttying, painting, or curing operations, the method further includes: Sx1, Obtain real-time temperature and humidity data of the interior of cabinet 1 or the workshop environment; Sx2. Based on the temperature and humidity data, query the pre-stored environment-process parameter mapping table to obtain the putty curing rate correction coefficient, paint viscosity compensation value and infrared curing power adjustment amount under the current environment. Sx3. Based on the correction coefficient, compensation value, and adjustment amount, dynamically adjust the putty extrusion rate, paint atomization air pressure, and infrared curing lamp power.

[0117] Understandably, by acquiring ambient temperature and humidity before key process steps and dynamically adjusting putty extrusion, spraying air pressure, and curing power using a mapping table, this mechanism enables the robot to adapt to different environments. Traditional fixed-parameter modes are prone to problems such as delayed curing and paint film defects under extreme climates, while this solution ensures process stability through feedforward compensation. Therefore, it can output consistently high-quality repair results regardless of whether it's a dry winter or a humid summer, significantly improving the equipment's applicability and robustness in different regions and seasons.

[0118] For example, this embodiment performs the repair in a high humidity environment during summer (temperature 32°C, humidity 75% RH).

[0119] Specific implementation of step Sx1: The control mechanism reads the data from the built-in temperature and humidity sensor in cabinet 1: T=32℃, RH=75%.

[0120] Specific implementation of step Sx2: According to the environment-process mapping table: Under high humidity, the curing rate of putty decreases by 15%, so the correction factor is 1.15; the viscosity of water-based paint increases, so the atomization pressure needs to be increased by 0.05 MPa; infrared curing requires a 10% increase in power to compensate for the heat absorption of moisture evaporation.

[0121] Specific implementation of step Sx3: The putty extrusion rate was increased from 0.12 mL / s to 0.14 mL / s; the spray painting air pressure was adjusted from 0.25 MPa to 0.30 MPa; and the initial power of the infrared curing lamp was increased from 60% to 70%. The final paint film was free of sagging and orange peel, and the curing time was consistent with the standard environment.

[0122] In some embodiments, after generating the test report, the method further includes: Sy1. Store the initial damage parameters, execution parameters of each process, and final inspection quality results of this repair as a training sample in the local process database. Sy2. When the cumulative number of samples exceeds the preset threshold, the lightweight regression model is invoked to update the optimal grinding pressure, putty thickness or polishing speed for the same type of damage online. Sy3. Write the updated process parameters into the preset process library to guide subsequent repair operations for similar damage.

[0123] Understandably, by storing process-quality data after each repair, and accumulating it to a certain scale, the optimal parameters for similar damage are updated online, feeding back into subsequent operations. This mechanism endows the robot with a self-evolving ability to "get smarter with use." Traditional fixed process libraries cannot adapt to batch differences in materials or changes in tool wear, while this solution approximates the true optimal solution through continuous learning. As a result, repair quality steadily improves over long-term operation, significantly reducing the need for manual parameter tuning, laying the foundation for building a self-evolving intelligent maintenance system.

[0124] For example, in this embodiment, learning is triggered after the 100th scratch repair with a depth of 0.3–0.5 mm is completed.

[0125] Specific implementation of step Sy1: System records: initial depth 0.42 mm, grinding pressure 0.8 N, final inspection ΔE=2.5, depth difference 0.06 mm.

[0126] Specific implementation of step Sy2: When the sample size reached 100, the control mechanism ran a linear regression model and found that for this type of damage, ΔE decreased by an average of 0.3 when the grinding pressure was 0.75N.

[0127] Specific implementation of step Sy3: The default polishing pressure for "scratches with a depth of 0.3–0.5 mm" was updated from 0.8 N to 0.75 N and added to the process library. The new parameter was used for the 101st similar repair, and the final inspection ΔE was reduced to 2.2.

[0128] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A paint repair robot for repairing damaged paint on vehicles; characterized in that, It includes a cabinet, a control mechanism, and a moving mechanism, a detection mechanism, and a repair mechanism connected to the cabinet; The moving mechanism is movable relative to the cabinet. The moving mechanism is used to drive the repair mechanism to move relative to the damaged paint surface of the vehicle, so that the repair mechanism can move to the damaged paint surface of the vehicle and perform repair operations on the damaged paint surface of the vehicle. The detection mechanism is connected to the moving mechanism and is used to detect damage parameters of the vehicle's damaged paint surface. The damage parameters include damage depth and damage area. The control mechanism is configured to control the operation of the moving mechanism and the repair mechanism based on the detection data from the detection mechanism.

2. The paint surface repair robot according to claim 1, characterized in that, The paint repair robot also includes a cleaning mechanism connected to the cabinet. The cleaning mechanism is used to clean up impurities generated by the repair mechanism when repairing damaged paint surfaces on vehicles.

3. The paint surface repair robot according to claim 2, characterized in that, The repair mechanism includes a polishing component used to polish the damaged paint surface of the vehicle. The cleaning mechanism includes a water spray component and an air spray component, which are electrically connected to the control mechanism. The control mechanism is configured to: after the polishing component polishes the damaged paint surface of the vehicle, control the water spray component to spray water onto the damaged paint surface of the vehicle to clean the powder generated by the polishing component; and after the water spray component sprays water onto the damaged paint surface of the vehicle, control the air spray component to spray air onto the damaged paint surface of the vehicle.

4. The paint surface repair robot according to claim 2, characterized in that, The paint repair robot also includes a drying component connected to the cabinet. The repair mechanism also includes a putty scraping component, a painting component, and a polishing component. The drying component, the putty scraping component, the painting component, and the polishing component are all electrically connected to the control mechanism. The control mechanism is configured to control the sanding component, the cleaning mechanism, the putty scraping component, the painting component, the drying component, and the polishing component to work sequentially on the damaged paint surface of the vehicle.

5. The paint surface repair robot according to any one of claims 1 to 4, characterized in that, The paint surface repair robot also includes a placement mechanism, which is fixedly installed on the top of the cabinet and is used to support the repair mechanism.

6. A method for paint surface inspection, applied to the paint surface inspection robot as described in any one of claims 1 to 5, characterized in that, include: S1. When the detection mechanism is above the middle area of ​​the damaged paint surface of the vehicle, control the detection mechanism to obtain the damage parameters of the damaged paint surface of the vehicle. S2. Determine that the damage parameters are within a preset range; S3. Control the moving mechanism to move the repair mechanism to the damaged paint surface of the vehicle; S4. Control the repair mechanism to perform repair operations on the damaged paint surface of the vehicle.

7. The paint surface repair method according to claim 6, characterized in that, The detection mechanism includes a vision sensor, and the step of controlling the detection mechanism to acquire damage parameters of the vehicle's damaged paint surface includes: S11. Control the vision sensor to acquire images of the damaged paint surface of the vehicle to obtain the original image; S12. Based on the original image, perform mild bilateral filtering, non-local mean denoising, and guided filtering sequentially according to preset filtering rules to obtain the denoised image. S13. Based on the denoised image, the closed boundary of the damaged area is identified by the contour registration algorithm, and the connected region within the closed boundary is filled by the region growing algorithm to obtain the damaged area mask. S14. Based on the damage area mask, obtain the damage area of ​​the vehicle's damaged paint surface; S15. Based on each pixel within the mask coverage area of ​​the damaged region in the denoised image, determine the brightness distribution data, which includes the brightness value of each pixel. S16. Based on the color distribution within the mask coverage area of ​​the damaged region in the denoised image, identify the set of the darkest color regions and the set of the lightest color regions. S17. Based on the center coordinates of the high-brightness region and the center coordinates of the low-brightness region, control the distance measurement of the center point of the high-brightness region and the center point of the low-brightness region to obtain a first depth value and a second depth value, wherein the brightness value of the pixel in the high-brightness region is greater than the first threshold, and the brightness value of the pixel in the low-brightness region is less than the second threshold. S18. Based on the first depth value and the second depth value, determine the damage depth of the vehicle's damaged paint surface.

8. The paint surface repair method according to claim 7, characterized in that, Step S4 includes: Based on the damage region mask and region growth algorithm, the filling order and growth direction inside the damage region are determined to obtain the region growth direction information. Based on the depth values ​​of multiple sampling points within the damaged area, depth distribution feedback data is generated; Based on the growth direction information of the region and the depth distribution feedback data, the area covered by the mask in the damaged region is calibrated point by point to obtain the calibration coordinate sequence. Based on the calibration coordinate sequence, the reciprocating operation trajectory is obtained; Based on the reciprocating work trajectory, a motion command is sent to the moving mechanism to drive the repair mechanism to move along the reciprocating work trajectory and perform the repair operation.

9. The paint surface repair method according to claim 8, characterized in that, The repair mechanism includes a grinding component; step S4 includes: Based on the reciprocating work trajectory, the moving mechanism is controlled to drive the grinding part to perform grinding operations on the damaged paint surface of the vehicle along the reciprocating work trajectory; Based on the image of the polishing area collected in real time by the vision sensor during the polishing operation, the brightness value of each pixel in the image is extracted, and it is determined whether there is a continuous area with a brightness value greater than a preset gloss threshold, thus obtaining a first flatness determination result; at the same time, based on the distance values ​​measured by the infrared ranging sensor at multiple sampling points in the polishing area, it is determined whether the distance of each sampling point is within a preset thickness tolerance range, thus obtaining a second flatness determination result. Based on the first flatness determination result that there is a continuous bright area and the second flatness determination result that the distance between all sampling points is within the thickness tolerance range, a polishing completion signal is generated; Based on the polishing completion signal, the cleaning mechanism is controlled to clean the damaged paint surface of the vehicle.

10. The paint surface repair method according to claim 9, characterized in that, After the step of controlling the cleaning mechanism to clean the damaged paint surface of the vehicle, the method further includes: The vision sensor and the infrared ranging sensor are controlled to scan the damaged paint surface of the vehicle to obtain the detection image and depth data after polishing. Based on the post-polishing detection image and the post-polishing depth data, S13 to S18 are repeated to obtain the post-polishing damage area and post-polishing damage depth. Based on the fact that the damaged area after grinding is greater than a preset area value or the damaged depth after grinding is greater than a preset flatness threshold, a local re-grinding command is generated, and a local re-grinding trajectory is regenerated based on the mask of the current damaged area after grinding, wherein the local re-grinding trajectory is different from the reciprocating operation trajectory. Based on the local re-grinding trajectory, the moving mechanism is controlled to drive the grinding part to move along the local re-grinding trajectory and perform the repair operation.

Images