Camera polishing system for automotive vehicle structure

Abstract

The invention relates to a system (10) for polishing a protruding asperity on a mechanical part (12) such as a motor vehicle structure. The polishing system comprises a mechanical part receiving space (15), a first articulated arm (22) equipped with a polishing tool (24); a control unit (16). The polishing system further comprises a second articulated arm (28) configured to be controlled by the control unit. The second articulated arm (28) comprises vision means (30) capable of detecting a protruding asperity on the mechanical part. The control unit (16) is configured to control the first articulated arm (22) in order to polish the mechanical part (12) with the polishing tool (24) so as to remove said asperity detected via said vision means. The invention also relates to a polishing method and a computer program. Figure to be published with the abstract: Figure 1

Description

Title of the invention: Camera polishing system for motor vehicle structure

[0001] The invention relates to the surface control of parts, for example in the automotive field. The invention relates more specifically to a system and a method for polishing a part. The invention also relates to a computer program.

[0002] A motor vehicle has a structure, or body, of complex shape. The latter comprises a plurality of sheets stamped and then welded together. Following the ironwork, the raw structure undergoes various finishing operations contributing to the perceived quality of the motor vehicle. In order to guarantee this quality according to pre-established standards, the structure undergoes polishing operations in order to eliminate certain imperfections.

[0003] Polishing and sanding operations in automotive manufacturing have evolved from manual operation by an operator using polishing tools to automation using robots and rotary tools. This technique allows for the automatic removal of weld defects by polishing the surface along a predetermined path that covers all areas of the bodywork. This improves the appearance and final finish of the motor vehicle by avoiding the arduous conditions of manual labor.

[0004] During the automated welding process in a production line, material spatter is a common and random form of defect. This material spatter occurs when the molten metal does not completely fuse with the base metal and is thrown outward from the weld, leaving a buildup of material near the weld point. The excess molten material and marks are then removed by polishing or sanding the metal surface with an automatic polishing tool attached to a robot. The polishing tool follows a path connecting the weld points.

[0005] Robot-assisted automated weld defect polishing systems are precise and uniform, improving the quality of the final finish. In addition, robots can work in hazardous or difficult conditions; particularly in environments containing metal dust. These robots thus reduce safety costs. However, excessive polishing and sanding of weld defects can present several risks, such as excessive material removal. Excessive material is removed, weakening the weld and / or the sheets. This increases the risk of failure since the mechanical properties and characteristics of the weld can be altered.

[0006] Furthermore, excessive sanding and polishing is likely to generate heat which can affect the metal structure and cause residual stresses. It is also likely to create new defects such as scratches or imperfections on the surface of the weld.

[0007] However, the precision of retouching is limited. Indeed, the structure commonly has a coating whose thickness is of the order of a tenth of a millimeter. It is likely to be removed locally when the polishing is too pronounced. Therefore, the structure is no longer effectively protected at the location of the retouching of the roughness.

[0008] There is a need to provide a solution minimizing at least one of these risks.

[0009] Document CN114310962A discloses an intelligent robot communication control system for grinding and a method therefor. The system comprises an upper computer, an IRB6700-200 type grinding robot, a control demonstrator, a floor type belt sander, a clamping jaw, an air compressor, a three-dimensional laser scanner and other visual image acquisition modules. The upper computer serves as a master station, the demonstrator serves as a slave station, the upper computer sends an instruction to the control demonstrator, and the demonstrator receives and sends the instruction to the grinding robot and the belt sander so as to grind a workpiece.The visual image acquisition module is used to perform feature recognition on the outer surface of the polished workpiece; determine a posture track and send the posture track to an upper computer. The actual position of the workpiece to be polished is obtained through the position of a camera arranged at the rear end of the polishing robot.

[0010] Document CN109955122A and document CN209811886U disclose fully automatic grinding systems based on computer vision. A system comprises a workpiece clamping platform, a grinding device, a computer vision image system, an upper computer control system, and a robot control cabinet. The grinding device comprises an industrial robot mounted on one side of the workpiece clamping platform, and a grinding execution device used for grinding the workpieces to be ground is mounted on the industrial robot. The computer vision image system is used for collecting image data of the grinding faces of the workpieces to be ground. The upper computer control system establishes a three-dimensional model of the grinding faces according to the image data collected by the computer vision image system.

[0011] However, these systems are limited in terms of speed. Furthermore, each image system is exposed to projections during the polishing phases. In addition, the presence of the grinding device limits the field of vision by forming a physical mask which conceals certain defects.

[0012] The invention aims to address at least one of the problems or incon encountered in the prior art. In particular, the invention aims to increase the rate of polishing and inspection of mechanical parts. The invention also aims to optimize the rate, wear of the polishing tools, safety and precision of retouching a mechanical part with a coating, in particular a structure of a motor vehicle.

[0013] According to a first aspect, the invention proposes a system for polishing a protruding asperity on a mechanical part; the polishing system comprising a mechanical part receiving space, a first articulated arm equipped with a polishing tool; a control unit; remarkable in that the polishing system further comprises a second articulated arm configured to be controlled by the control unit, the second articulated arm comprising vision means capable of detecting a protruding asperity on the mechanical part; the control unit being configured to control the first articulated arm in order to polish the mechanical part with the polishing tool so as to remove said asperity detected via said vision means.

[0014] It will be clearly understood that the invention proposes a system for polishing a protruding roughness on a mechanical part; the polishing system comprising a space for receiving a mechanical part, a control unit for a first articulated arm equipped with a polishing tool and a second articulated arm with vision means capable of detecting a protruding roughness on the mechanical part; the control unit being configured to control the first articulated arm in order to polish the mechanical part with the polishing tool so as to abrade the detected roughness via said vision means.

[0015] Document CN111496580A discloses a multi-machine cooperative processing system and method for a large-caliber aspherical surface optical element. The multi-machine cooperative processing system comprises a robot body, a robot body controller, a polishing tool, a workbench, a polishing liquid supply system, and a programming simulation test system. By means of a control instruction, multiple robots can drive the polishing tool to polish a workpiece along a planned route. However, polishing along the planned route cannot adapt to production hazards.

[0016] Preferably, the vision means are configured to estimate a roughness thickness; the control unit being configured to control the first articulated arm and the polishing tool as a function of said roughness thickness.

[0017] Preferably, the vision means are configured to be sensitive to light in the visible range and in the infrared range.

[0018] Preferably, the second articulated arm comprises a light source associated with the vision means.

[0019] Preferably, the polishing tool comprises an abrasive disc and / or an abrasive brush.

[0020] Preferably, the vision means comprise a camera.

[0021] Preferably, the polishing system further comprises a third articulated arm identical to the second articulated arm.

[0022] Preferably, the receiving zone comprises a support capable of receiving the mechanical part.

[0023] Preferably, the receiving zone comprises means for holding the mechanical part.

[0024] Preferably, the mechanical part is metallic.

[0025] Preferably, the vision means are capable of inspecting the mechanical part.

[0026] Preferably, the first articulated arm and the second articulated arm are arms articulated with at least five axes of mobility.

[0027] Preferably, the polishing system comprises a user interface.

[0028] Preferably, the polishing tool comprises a diameter of at most 10 cm, more preferably of at most 5 cm.

[0029] Preferably, the vision means comprise a module for recognizing roughness based on reference images.

[0030] Preferably, the recognition module is configured to recognize a roughness according to an automatic learning process.

[0031] According to another aspect, the invention proposes a method for polishing a protruding asperity on a mechanical part using a polishing system; remarkable in that the polishing system is in accordance with the invention, and in that the polishing method comprises the following steps: a) inspection of a surface of the mechanical part with the vision means of the second articulated arm; b) detection of a protruding asperity on said surface; c) polishing the asperity with the polishing tool of the first articulated arm; then preferably, d) inspection of the surface at the location of the polishing of the polished asperity during step c) polishing.

[0032] Preferably, the polishing method comprises a step e) estimation of wear of the polishing tool of the first articulated arm.

[0033] Preferably, the mechanical part comprises two metal elements fixed by a weld at the level of the asperity detected in step b) detection; preferably, the polishing method comprises a step f) emission of a weld anomaly signal when the detected asperity meets an anomaly criterion.

[0034] Preferably, the mechanical part is a motor vehicle structure.

[0035] Preferably, the structure comprises a coating forming said surface.

[0036] Preferably, the coating comprises zinc.

[0037] Preferably, the control method comprises a step of determining rotation speed of the polishing tool and / or pressure applied by the polishing tool on the asperity detected in step b) detection.

[0038] According to another aspect, the invention provides a computer program comprising instructions which, when the program is executed by a computer, cause the latter to implement the polishing method according to the invention.

[0039] Preferably, the polishing method comprises a step g) calculation of a trajectory of the polishing tool of the second articulated arm, said trajectory passing through the asperity detected in step b) detection.

[0040] Preferably, the computer is integrated into the control unit.

[0041] According to another aspect, the invention proposes a method for monitoring wear of a polishing tool of a roughness polishing system; remarkable in that the polishing system is in accordance with the invention, and in that the method for monitoring wear of a polishing tool comprises the steps: a) inspection of a surface of the mechanical part with the vision means of the second articulated arm; b) detection of a roughness projecting from said surface; c) polishing the roughness with the polishing tool of the first articulated arm; e) estimation of wear of the polishing tool of the first articulated arm.

[0042] According to another aspect, the invention provides a computer program comprising instructions which, when the program is executed by a computer, cause the latter to implement the wear monitoring method according to the invention.

[0043] Each characteristic introduced by the expression “preferably” given in relation to one of the aspects of the invention applies to all the other aspects of the invention.

[0044] The invention will be well understood and other aspects and advantages will appear clearly on reading the description which follows, given with reference to the figures appended and listed below.

[0045] [Fig.l] represents a polishing system according to the invention.

[0046] [Fig.2] illustrates vision means and a polishing tool analyzing and processing a rough surface of a mechanical part in a polishing system according to the invention.

[0047] [Fig. 3] is a diagram of a polishing process according to the invention.

[0048] In the following description, the term "comprise" is synonymous with "include" and is not limiting in that it allows the presence of other elements in the polishing system or other steps in the polishing process to which it relates. It is understood that the term "comprise" includes the terms "consist of". The terms "external" and "internal" will designate respectively that which is oriented towards the exterior of the vehicle and towards the interior of the vehicle.

[0049] In the present description, the ranges of values include the limits which de- limit.

[0050] In the present description, the equalities between the values are not to be understood in the strict sense insofar as each equality authorizes a variation of at most 10%, preferably at most 5%, more preferably at most 2%, between these values.

[0051] In the present description, the technical characteristics are defined in the mounting configuration of the polishing system, unless explicitly mentioned otherwise.

[0052] Throughout the description, the different figures use the same reference signs to designate identical or similar entities.

[0053] [Fig.l] represents a system 10 for polishing the roughness of a mechanical part 12. In the present illustration, the mechanical part 12 is a structure 14 of a motor vehicle. The structure 14 delimits the passenger compartment of the motor vehicle. It extends over the entire width of the motor vehicle. It can extend substantially over the entire length of the motor vehicle. The structure 14 comprises a length of between 2 m and 5 m; preferably between 3 m and 4.5 m.

[0054] The polishing system 10 is intended to be used in the automotive field. In particular, it meets speed and precision constraints.

[0055] The polishing system 10 is capable of polishing asperities (not visible) protruding from the mechanical part 12. The asperities form excess thicknesses. They form bumps. They form irregularities, defects on the surface of the mechanical part 12. They increase the surface roughness, and harm the surface homogeneity, and therefore the perceived quality. The asperities comprise a length of at most 6 mm, preferably at most 2.0 mm, more preferably at most 1.0 mm. The length is measured along the surface from which the asperity protrudes.

[0056] The polishing system 10 comprises a receiving space 15. The receiving space 15 preferably comprises a fixed holding support (not shown) for the mechanical part 12, which makes it possible to hold it firmly in a precise position. According to an alternative of the invention, the polishing system comprises a conveyor moving the mechanical part to the receiving space.

[0057] The polishing system 10 comprises a control unit 16. The control unit 16 controls the moving entities in the mechanical part 12, and in particular the arrival and then the departure of the mechanical part 12. It is coupled to the actuators and the sensors. The control unit 16 comprises, for example, a human-machine interface 18, for example with a display screen and means for entering instructions by an operator.

[0058] The polishing system 10 comprises a first robot 20 with a first articulated arm 22 equipped with a polishing tool 24. The first robot 20 is controlled by the control unit 16. The first articulated arm 22, thanks to its polishing tool 24, is able to reach all surfaces of the mechanical part 12. The first articulated arm 22 is in the receiving space 15, or adjacent to the receiving space 15.

[0059] The polishing system 10 also comprises a second robot 26 with a second articulated arm 28. The second articulated arm 28 is equipped with vision means 30. The vision means 30 are computer vision means, also called digital vision means. They comprise photosensitive sensors, and are capable of acquiring at least one digital image of the mechanical part 12. The digital images are transmitted to the control unit 16 which processes them.

[0060] The vision means 30 contribute to the computer detection of protruding asperity on the mechanical part 12. The control unit 16 is configured to control the first robot 20 in order to polish the mechanical part with the polishing tool of the first articulated arm 22 so as to remove said asperity detected by said vision means 30.

[0061] Each robot is fixed in the reception space 15, or mobile thanks to a mobile base. Each mobile base is able to move around the mechanical part 12. Each articulated arm comprises at least one mobile segment, preferably several mobile segments; for example three segments mobile relative to each other and relative to the optional base.

[0062] The arms are movable in rotation, and are moved by motors themselves controlled by the control unit 16. Thus, the polishing tool 24 and the vision means 30 can scan all the surfaces of the mechanical part 12. In the context of the motor vehicle structure, the polishing tool 24 and the vision means 30 can reach all or part of the internal surface of the motor vehicle structure 14. Each articulated arm comprises at least two axes of rotation, preferably at least three axes of rotation, more preferably at least four axes of rotation, even more preferably at least five axes of rotation.

[0063] According to one option of the invention, the first articulated arm and the second articulated arm are arranged on the same robot. They are linked to the same base. They are movable relative to each other.

[0064] [Fig.2] shows a polishing tool 24 and vision means of a system of polishing 10 facing a mechanical part 12. The polishing system 10 may correspond to that presented in relation to [Fig.l]; just like the mechanical part 12.

[0065] The polishing tool 24 is at a free end of the first articulated arm 22. The vision means 30 are at a free end of the second articulated arm 28, for example at the last segment. They are facing a surface 32 of the mechanical part 12.

[0066] The vision means 30 are capable of identifying and detecting a roughness (not re presented) projecting from the mechanical part 12. The vision means 30 make it possible to position the detected roughness. The positioning can be according to a reference point specific to the mechanical part 12 or to the receiving space 15. The control unit 16 is adapted to control the first robot, in particular the first articulated arm 22, in order to polish the mechanical part 12 with the polishing tool 24 so as to remove said roughness detected by said vision means.

[0067] An asperity generally forms a bump having a thickness between 0.01 mm and 3 mm, preferably between 0.03 mm and 1 mm; more preferably between 0.2 mm and 0.5 mm. The thickness is measured perpendicular to the surface 32.

[0068] The vision means 30 are configured to estimate a roughness thickness. The roughness thickness is estimated statistically as a function of its width, or using two images taken from different viewing angles. The control unit 16 is configured to control the first articulated arm 22 as a function of said roughness thickness. It modulates at least: the polishing time, the rotation speed, the pressure of the polishing tool 24, as a function of the roughness thickness.

[0069] The vision means 30 comprise a camera. The camera is for example a hyperspectral camera. They may comprise an infrared camera. The vision means 30 are configured to be sensitive to light in the visible range and in the infrared range. Generally, the vision means 30 comprise a camera. The vision means 30 are sensitive to light with a wavelength between 700 nm and 2000 nm.

[0070] According to one option, the second articulated arm 28 comprises a light source 34 associated with the vision means 30. The light source 34 may be circular. It may surround an optical axis of the vision means 30. The light source 34 is integral with the vision means 30. In the present embodiment, the light source 34 forms a ring; however, it may describe any other shape.

[0071] The polishing tool 24 comprises a rotating tool. The rotating tool is circular. The polishing tool 24 comprises an abrasive disc and / or an abrasive brush. It may form a cylinder. The polishing tool 24 has a diameter of less than 10 cm, more preferably less than 5 cm. A limited diameter makes it possible to reduce the polished area and to access hidden and confined surfaces. Combined with the vision means 30, the radius makes it possible to rationalize the polished area, and to have access to it despite the presence of physical obstacles.

[0072] According to an alternative or option of the invention, the polishing tool comprises sandpaper.

[0073] The mechanical part 12 is metallic. It can be made of steel or aluminum. The mechanical part 12 comprises at least two metallic elements fixed by welding. at the level of the asperity. Preferably, the mechanical part 12 comprises at least tens of metal elements welded to each other. They can be welded using electric welding spots. The metal elements can be sheets. The sheets have thicknesses between 0.5 mm and 1.5 mm, preferably between 0.65 mm and 0.85 mm.

[0074] The mechanical part 12 comprises a coating forming said surface 32. The surface 32 is preferably a curved surface. It optionally forms a rebate. The coating may be zinc. It may correspond to a galvanic layer. For example, after welding, the metal elements are immersed in a bath of molten zinc.

[0075] According to an option of the invention, the polishing system further comprises a third articulated arm (not shown) identical to the second articulated arm. The third articulated arm allows a visual inspection of the mechanical part after polishing. The third articulated arm is for example specific to a third robot. This multitude of articulated arms optimizes production rates, rectifications and inspections.

[0076] The invention makes it possible to physically decouple visual inspection from polishing, and to carry out these operations at the same time. Furthermore, depending on the mechanical parts and their typical roughness; visual inspection may be slower than polishing; or vice versa. Thus, the two-arm configuration limits the impact of differences in operating times of the vision means and the polishing tool.

[0077] The control unit 16 comprises a processor 36 and a memory 38. The human-machine interface 18 is connected to the first robot and to the second robot via the control unit 16.

[0078] The processor 36 may include one or more programmable electronic microprocessors or microcontrollers. In addition, the processor 36 may include a central processing unit (CPU), a memory (in addition to or like the separate memory illustrated by the reference numeral 38), and an input / output (I / O) interface through which the processor 36 may receive a plurality of input signals. Such an I / O interface is also configured to generate a plurality of output signals including, but not limited to, those used to control and / or provide data to the display of the human-machine interface 18.

[0079] The memory 38 is provided for storing data and instructions or code (i.e., software) for and readable and / or writable by the processor 36. The memory 38 may include various forms of non-volatile (i.e., non-transient) memory. The non-volatile memory includes flash memory or read-only memory (ROM), any type of programmable read-only memory (e.g., PROM, EPROM, EEPROM). The memory 38 optionally includes volatile memory, including random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), and synchronous dynamic random access memory (SDRAM). According to the invention, the memory may be internal to the processor 36; or alternatively, form a separate component.

[0080] [Fig. 3] shows a diagram of a roughness polishing process with a polishing system. The polishing system and / or the mechanical part may correspond to the one or those presented in relation to one of Figures 1 to 2.

[0081] The polishing process comprises the following steps:

[0082] a) inspection 100 of a surface of the mechanical part with the vision means of the second articulated arm;

[0083] b) detection 102 of an asperity protruding from the surface;

[0084] c) polishing 104 of the roughness with the polishing tool of the first articulated arm; and preferably,

[0085] d) control 106 of the polished surface at the location of the polishing of the asperity detected during step b) detection 102;

[0086] e) estimation 108 of wear of the polishing tool of the first articulated arm;

[0087] f) emission 110 of a welding anomaly signal when the detected asperity meets an anomaly criterion;

[0088] g) calculation 112 of a trajectory of the polishing tool of the first articulated arm, said trajectory passing through the asperity detected in step b) detection 102;

[0089] h) determination 114 of rotation speed and / or pressure of the polishing tool.

[0090] The mechanical part comprises at least two metal elements fixed by a weld. During the welding operation, their fixing zone is likely to have a projection of droplets of molten metal. The latter then form protruding asperities. When the weld is an electric spot weld, the weld zone may become dented and form a protruding asperity. Such asperities are detected during step b) detection 102.

[0091] The mechanical part preferably comprises a coating forming said surface. During step c) polishing 104, the roughness is removed by stopping before reaching the coating on the surface. Thus, the polishing operation is limited in thickness; in addition to being limited over the extent of the mechanical part. It does not thin the thickness of the coating around the roughness. This preserves the coating and the physical integrity of the mechanical part so as not to weaken it. This is relevant in the automotive field since the thickness of the sheets is optimized according to the strength and mass. Pronounced polishing is likely to weaken the structure. If necessary, limiting the polishing preserves the mechanical strength of the welds.

[0092] According to an alternative of the invention, the roughness comprises a burr or a run, or a bubble. Each of these rough spots forms a defect that remains visible, especially after the application of one or more layers of paint and varnish; as generally applied to a motor vehicle body. The burr can result from a sheet metal cutting step. The drip can be a zinc drip, for example during treatment in a bath. They penalize the perceived quality.

[0093] Step b) detection 102 is carried out by means of an artificial intelligence algorithm, in particular a deep learning algorithm. Such an algorithm is generally referred to by the English expression “Deep Learning”. The deep learning algorithm makes it possible to detect welding defects, bumps, peaks and other artifacts specific to the surface of the mechanical part. The deep learning algorithm detects these elements in the images from the vision means, and predicts the defects, their size, their type and their severity. It implements a convolutional neural network, generally referred to by the acronym “CNN” corresponding to the English expression “Convolutional Neural Network”.

[0094] The input data of the convolutional neural network are images captured by the vision means. The images can be resized, filtered, reoriented, cut. The output data are a series of regions of interest found in these images, each with metadata corresponding to: the size, type and thickness of the asperity. The algorithm is trained according to a supervised or unsupervised method. It is trained with defect-free images and images with defects labeled according to the type of the defect. The deep learning algorithm is trained until it reaches a deep learning model having a reliability of at least 95%, preferably at least 99%.

[0095] Step e) estimation 108 is carried out by comparing the images obtained during step a) inspection 100 and step d) control 106. When the asperity persists despite a predefined polishing time, tool wear is identified. According to an alternative of the invention, step e) estimation 108 is carried out by calculating the usage time of the polishing tool.

[0096] In step f) emission 110, if the asperity has a thickness and / or a length greater than a predefined threshold, the anomaly signal is emitted. The signal can be emitted if the asperity has a certain shape, or a given color.

[0097] For step f) emission 110, a machine learning algorithm is implemented. Such a solution is generally designated by the acronym “ML” which corresponds to the English expression “Machine Learning”. The purpose of this machine learning algorithm is to correlate the results of the robotic polishing tool with upstream manufacturing processes; such as a welding assembly process. The objective is to find the causes of the appearance of the defects. For example, if the defects are caused by a misalignment of the robots welding, this machine learning algorithm highlights a correlation of data that may indicate such causality.

[0098] The machine learning algorithm comprises a neural network with several hidden layers. The neural network can be fully connected, or partially connected. The mathematical formulas for connection between neurons include sigmoid-type activation functions. Other activation functions are considered. The network architecture is, for example, a forward propagation neural network.

[0099] The input data includes all process data from previous manufacturing steps, such as cycle times, welding parameters; references, temperatures. Other data are conceivable. The welding parameters include current, electrical quantities, pressure.

[0100] The output data comprises a series of correlation rules linking the input data and the defective parts. The machine learning algorithm is trained by creating a dataset with previous occurrences in which the data is labeled. The training data includes measured data and possibly simulation data.

[0101] Step d) control 106 can be carried out by the second articulated arm or by the optional third articulated arm. In the event that the result does not meet quality criteria, the method again carries out step c) polishing 104.

[0102] The invention comprises the combination of all the embodiments illustrated by all the figures.

Claims

Claims

1. Polishing system (10) for a protruding asperity on a mechanical part (12); the polishing system (10) comprising a receiving space (15) for the mechanical part (12), a first articulated arm (22) equipped with a polishing tool (24); a control unit (16); characterized in that the polishing system (10) further comprises a second articulated arm (28) configured to be controlled by the control unit (16), the second articulated arm (28) comprising vision means (30) capable of detecting a protruding asperity on the mechanical part (12); the control unit (16) being configured to control the first articulated arm (22) in order to polish the mechanical part (12) with the polishing tool (24) so as to remove said asperity detected via said vision means (30).

2. Polishing system (10) according to claim 1, characterized in that the vision means (30) are configured to estimate a roughness thickness; the control unit (16) being configured to control the first articulated arm (22) and the polishing tool (24) as a function of said roughness thickness.

3. Polishing system (10) according to one of claims 1 to 2, characterized in that the vision means (30) are configured to be sensitive to light in the visible range and in the infrared range; preferably, the second articulated arm (28) comprises a light source (34) associated with the vision means (30).

4. Polishing system (10) according to one of claims 1 to 3, characterized in that the polishing tool (24) comprises an abrasive disc and / or an abrasive brush; preferably, the vision means (30) comprise a camera.

5. Polishing system (10) according to one of claims 1 to 4, characterized in that the polishing system (10) further comprises a third articulated arm identical to the second articulated arm (28).

6. Method for polishing protruding roughness on a mechanical part (12) using a polishing system (10); characterized in that the polishing system (10) is in accordance with one of claims 1 to 5, and in that the polishing method comprises the following steps: • a) inspection (100) of a surface (32) of the mechanical part (12) with the vision means (30) of the second articulated arm (28); • b) detection (102) of an asperity protruding from said surface (32); • c) polishing (104) of the asperity with the polishing tool (24) of the first articulated arm (22); then preferably, • d) control (106) of the surface (32) at the location of the polishing of the asperity polished during step c) polishing (104).

7. Polishing method according to claim 6, characterized in that the polishing method comprises a step e) estimation (108) of wear of the polishing tool (24) of the first articulated arm (22).

8. Polishing method according to one of claims 6 to 7, characterized in that the mechanical part (12) comprises two metal elements fixed by a weld at the level of the asperity detected in step b) detection (102); preferably, the polishing method comprises a step f) emission (110) of a weld anomaly signal when the detected asperity meets an anomaly criterion.

9. Polishing method according to one of claims 6 to 8, characterized in that the mechanical part (12) is a structure (14) of a motor vehicle; preferably, the structure (14) comprises a coating forming said surface (32).

10. Computer program comprising instructions which, when the program is executed by a computer, cause the latter to implement the polishing method according to one of claims 6 to 9; preferably, the polishing method comprises a step g) calculation (112) of a trajectory of the polishing tool (24) of the second articulated arm (28), said trajectory passing through the asperity detected in step b) detection (102).

Images