A robot system for dual inspection of a spot weld, and a method

Abstract

The present invention relates to a robot system (1) for non-destructive inspection of a spot weld (30) on a workpiece (40), the robot system comprising a dual spot weld inspection system (50) having a vision-based system (52) configured to capture an image (60) of a surface (35) of the spot weld, and further an ultrasonic transducer (54) configured to acquire data of the spot weld; and a controller (90, 90d) configured to: receive the captured image and the acquired data; perform a first evaluation of the spot weld from the received captured image to identify a first possible anomaly (32) of the spot weld; perform a second evaluation of the spot weld from the received acquired data to identify a second possible anomaly (31) of the spot weld; and further configured to determine that the spot weld is a possible defective spot weld based on the performed first and second evaluations.

Description

[0001] A ROBOT SYSTEM FOR DUAL INSPECTION OF A SPOT WELD, AND A METHOD

[0002] Technical Field

[0003] The present invention relates to the field of automated inspection of welds, in particular during manufacturing and assembly of one or more workpieces comprising one or more spot welds. By way of example, the present invention relates to a robot system, controllers and methods for non-destructive inspection of a spot weld on a workpiece.

[0004] Background

[0005] There is an increasing demand for automating testing and quality control processes in industrial environments, where the use of robot systems can enhance efficiency and consistency. Automating these inspection tasks can reduce the time, cost, and potential errors associated with manual inspections, while improving overall quality assurance. One area of particular interest is the inspection of spot welds performed by robot systems, as the quality of the weld is crucial to ensuring structural integrity. However, detecting defects in spot welds often poses challenges that may affect the possibilities of fully automating the manufacturing processes.

[0006] It would therefore be desirable to provide an improved method for inspecting a weld on a workpiece.

[0007] Summary

[0008] In view of the above, it would be desirable to provide an improved method for automatically inspecting a weld, such as a spot weld, in which the detection of flaws can be performed in a more efficient, reliable, and automated manner. The object is at least partly achieved by a robot system according to independent claim 1, and the present invention as defined in the other independent claims. The dependent claims relate to advantageous embodiments. According to a first aspect of the present invention, there is provided a robot system for non-destructive inspection of a spot weld on a workpiece. The robot system comprises a dual spot weld inspection system having a vision-based system configured to capture an image of the surface of the spot weld and an ultrasonic transducer configured to acquire data of the spot weld. Moreover, the robot system comprises a controller configured to receive the captured image and the acquired data; perform a first evaluation of the spot weld from the received captured image to identify a first possible anomaly of the spot weld; perform a second evaluation of the spot weld from the received acquired data to identify a second possible anomaly of the spot weld; and further configured to determine that the spot weld is a possible defective spot weld based on the performed first and second evaluations.

[0009] Typically, the ultrasonic transducer may be configured to acquire data indicative of the subsurface of the spot weld.

[0010] The present invention is at least partly based on the insight that hitherto known inspection techniques, such as ultrasonic testing (UT), may have limitations when integrated into automatic weld inspection systems. Some ultrasonic methods may struggle to detect certain types of weld defects, such as stick welds, which often exhibit subtle characteristics that are not easily identifiable through ultrasonic data alone.

[0011] Moreover, some prior art systems rely on complex inspection procedures, which can be resource-intensive and time-consuming, especially when multiple inspection technologies are used separately. One challenge, therefore, is to develop an integrated solution that combines the strengths of both visionbased and ultrasonic inspection techniques while reducing complexity and increasing accuracy in detecting both surface and subsurface defects in real time.

[0012] The proposed system addresses these challenges by providing a robot system that integrates vision-based inspection and ultrasonic testing in a unified framework by the configuration of the dual spot weld inspection system. The system provides a dual spot weld inspection system, wherein a vision-based system captures an image of the surface of the spot weld, while an ultrasonic transducer acquires data, such as subsurface data or data indicative of the size of the spot weld. To this end, the two methods are applied to complement each other, allowing the robot system to detect a broader range of weld anomalies in real time.

[0013] Through the configuration of the controller, it becomes possible to process data from both the vision-based system and the ultrasonic transducer. By identifying possible anomalies from either the captured image or the ultrasonic data, the system allows for more reliable detection of defects, whether on the surface or below the surface. Such dual-mode operation may thus reduce the risk of false negatives, where a flaw might be missed if only one method is used.

[0014] Additionally, the system may be particularly useful for the detection of stick welds, which are difficult to identify using ultrasonic methods alone. The system may also be configured to cross-check data from both inspection methods to increase the confidence in identifying a weld as suspicious, thereby reducing the need for re-inspection or complex secondary verification processes.

[0015] By automating the inspection process and integrating both the vision-based system and the ultrasonic transducer in a dual spot weld inspection system of a single robot system, the system provides for real-time inspection, improving the speed and efficiency of weld quality assessment without sacrificing accuracy. Furthermore, the reduction in complexity, as compared to existing systems, may contribute to a more cost-effective solution for post-weld quality estimation.

[0016] The first possible anomaly of the spot weld may be different from a positional misplacement of the spot weld.

[0017] The first evaluation may comprise identifying a possible anomaly in the form of a surface anomaly in the spot weld based on the captured image. The surface anomaly may be any one of a weld spatter, a crack, an incomplete fusion, a surface deformation, a misalignment of the spot weld, and a miscolor of the spot weld. A technical advantage may include improved detection of surface defects such as weld spatter, cracks, and incomplete fusions using vision-based inspection. Hereby, the proposed system may reduce the risk of defective welds passing unnoticed.

[0018] In one example, a surface anomaly may be identified by the controller in the first evaluation if at least a combination of two anomalies is identified among the group of weld spatter, crack, incomplete fusion, surface deformation, misalignment of the spot weld, and miscolor of the spot weld.

[0019] The controller may be configured to identify a possible surface anomaly by comparing the captured image of the spot weld with one or more reference images of ordinary spot welds, so as to detect deviations indicating a surface anomaly. A technical advantage may include enhanced accuracy in surface defect detection through comparison with reference images, allowing the system to more effectively recognize deviations that indicate potential anomalies. By using image-based comparison with reference welds, the system can more reliably identify characteristics specific to spot welds, thereby enhancing the overall detection of problematic welds and reducing the need for additional manual inspection.

[0020] For example, the controller maybe configured to identify a possible anomaly by processing the captured image of the spot weld using a machine learning model configured to evaluate whether the image fits within a set of reference images representing acceptable spot welds, thereby detecting deviations indicative of a surface anomaly. The controller maybe configured to determine that the spot weld is a potential stick weld by comparing the captured image of the spot weld with one or more reference images indicative of a stick weld. The machine learning model is trained on a plurality of images of other spot welds. In one example, the controller may apply the trained machine learning model to the captured image of the spot weld to perform the first evaluation. The machine learning model evaluates surface-specific characteristics of the spot weld, including e.g. miscolor and the shape of the weld. In one example, the trained machine learning model performs the first evaluation to determine whether the spot weld requires further analysis to assess its quality.

[0021] The second evaluation may comprise identifying a possible anomaly in the form of a subsurface anomaly in the spot weld based on the data acquired by the ultrasonic transducer. The subsurface anomaly is any one of an inclusion defect, a porosity defect, and an internal crack. A technical advantage may include the ability to detect subsurface anomalies, such as inclusion defects, porosity, and internal cracks, using ultrasonic testing. Such configuration thus allows for an even more comprehensive assessment of the quality of the weld by identifying internal defects that could compromise structural integrity.

[0022] The controller maybe configured to identify a possible subsurface anomaly by analyzing the reflection and attenuation patterns of the ultrasonic signals acquired by the ultrasonic transducer. A technical advantage may include increased precision in subsurface anomaly detection through the analysis of ultrasonic signal reflection and attenuation patterns.

[0023] The possible subsurface anomaly may be identified based on a comparison with an expected signal profile for a correctly formed spot weld. A technical advantage may include enhanced reliability in identifying subsurface defects by comparing the acquired ultrasonic signals with expected signal profiles for correctly formed spot welds.

[0024] Alternatively, or in addition, the controller may be configured to calculate a size of a welded area within the spot weld based on the acquired data from the ultrasonic transducer. In this manner, controller may determine the extent of the welded area in the spot weld by distinguishing between welded and unwelded regions within the subsurface of the workpiece. By acquiring and analyzing subsurface data, the system calculates the size of the welded area, ensuring the weld meets predefined quality and structural requirements.

[0025] The controller may be configured to control the dual spot weld inspection system to perform a single non-destructive inspection pass over the spot weld, during which the vision-based system captures the image, and the ultrasonic transducer acquires data of the spot weld. The configuration to perform a single non-destructive inspection pass allows for increased efficiency and reduced inspection time by enabling simultaneous or sequential data collection using both the vision-based system and the ultrasonic transducer. Such configuration may also eliminate the need for multiple passes over the spot weld.

[0026] The ultrasonic transducer may acquire data of the spot weld prior to the visionbased system capturing the image during the single non-destructive inspection pass. Acquiring data with the ultrasonic transducer prior to capturing the image enables the detection of potentially critical subsurface anomalies before surface analysis is conducted. Such approach may reduce the risk of unnecessary processing of surface data if a flaw is already detected in the subsurface layer, thereby improving system resources and prioritizing potentially more critical subsurface defects for evaluation.

[0027] The controller maybe configured to perform the second evaluation of the spot weld from the received acquired data to determine the presence or absence of the second possible anomaly in the spot weld; and for any spot weld determined in the second evaluation to be absent of any second possible anomaly, perform the first evaluation of the spot weld from the received captured image to determine that the spot weld is a potential stick weld by comparing the captured image with one or more reference images indicative of a stick weld. The conditional approach of performing the second evaluation first to detect the presence or absence of subsurface anomalies before conducting a surface analysis provides a safeguard against unnecessary computational overhead. Furthermore, such configuration may enhance the reliability of the inspection process by focusing on detecting potential stick welds only when no critical subsurface defects are identified, improving accuracy in quality assessment and ensuring the most relevant defects are prioritized for further analysis.

[0028] The controller may further be configured to notify a user of the determined possible defective spot weld. A technical advantage may include real-time user notification of potential weld anomalies, based on the vision-based inspection and the ultrasonic data. Such capability allows for prompt intervention or further investigation, thus reducing the time between defect detection and corrective action. The controller may further be configured to notify a user of a determined defect in the spot weld based on the identified possible anomaly. A technical advantage may include automated user alerts upon the identification of a possible weld defect and the subsequent determination of the weld defect. In this manner, the presence of flaws can be more quickly communicated to the user, enabling immediate decisions regarding further inspection.

[0029] According to a second aspect of the invention, there is provided a method for non-destructive inspection of a spot weld on a workpiece using a dual spot weld inspection system of a robot system. The dual spot weld inspection system comprises a vision-based system and an ultrasonic transducer. The method comprises capturing an image of a surface of the spot weld using the visionbased system; acquiring data of the spot weld using the ultrasonic transducer; receiving, by processing circuitry of a controller, the captured image and the acquired data; performing, by the processing circuitry of the controller, a first evaluation of the spot weld using the received captured image to identify a first possible anomaly of the spot weld; performing, by the processing circuitry of the controller, a second evaluation of the spot weld using the received acquired data to identify a second possible anomaly of the spot weld; and determining, by the processing circuitry of the controller, that the spot weld is a possible defective spot weld based on the performed first and second evaluations. The second aspect of the invention may seek to solve the same problem as described for the first aspect of the invention. Thus, effects and features of the second aspect of the invention are largely analogous to those described above in connection with the first aspect of the invention.

[0030] Another technical advantage may include the real-time integration of visionbased and ultrasonic inspection methods to detect both surface and subsurface anomalies. Such dual approach provides for a more comprehensive assessment of the weld’s quality by allowing the detection of a broader range of defects, improving the reliability and accuracy of the inspection process. The method may further comprise identifying a possible anomaly by comparing the captured image of the spot weld with one or more reference images of ordinary spot welds, so as to detect deviations indicating a surface anomaly. A technical advantage may include enhanced detection accuracy through image comparison with reference welds.

[0031] The method may further comprise identifying a possible subsurface anomaly by analyzing the reflection and attenuation patterns of the ultrasonic signals acquired by the ultrasonic transducer. A technical advantage may include improved precision in identifying subsurface anomalies through analysis of ultrasonic signal patterns.

[0032] The possible subsurface anomaly may be identified based on a comparison with an expected signal profile for a correctly formed spot weld. A technical advantage may include more reliable subsurface defect detection by comparing the ultrasonic signal profiles with those of correctly formed welds.

[0033] Typically, the provision of capturing an image of a surface of the spot weld using the vision-based system and the provision of acquiring data of the spot weld using the ultrasonic transducer may be performed in a single nondestructive inspection pass over the spot weld.

[0034] In one example, the provision of performing the second evaluation of the spot weld may be completed prior to performing the first evaluation of the spot weld. Moreover, the second evaluation may comprise determining the presence or absence of the second possible anomaly in the spot weld. In addition, for any spot weld determined in the second evaluation to be absent of any second possible anomaly, the method may comprise performing the first evaluation of the spot weld to determine that the spot weld is a potential stick weld by comparing the captured image with one or more reference images indicative of a stick weld.

[0035] The second evaluation of the second possible anomaly of the spot weld from the received acquired data may be performed in response to that the first evaluation of the first possible anomaly of the spot weld does not identify a possible surface anomaly in the spot weld. A technical advantage may include an even more reliable inspection system of spot welds. For example, such a configuration may provide a safeguard system to ensure that subsurface anomalies are still detected even when surface anomalies are absent.

[0036] The first evaluation of the first possible anomaly of the spot weld from the received captured image may be performed in response to that the second evaluation of the second possible anomaly of the spot weld does not identify a possible subsurface anomaly in the spot weld. A technical advantage may include an even more reliable inspection system of spot welds. For example, such configuration provides that surface anomalies are assessed when no subsurface anomalies are found.

[0037] There is also provided a computer program product comprising program code for performing, when executed by a controller, the method according to the second aspect, and a non-transitory computer-readable storage medium comprising instructions, which when executed by a controller, cause the controller to perform the method according to the second aspect. The computer program may be stored or distributed on a data carrier. As used herein, a “data carrier” may be a transitory data carrier, such as modulated electromagnetic or optical waves, or a non-transitory data carrier. Non- transitory data carriers include volatile and non-volatile memories, such as permanent and non-permanent storage media of magnetic, optical or solid- state type. Still within the scope of “data carrier”, such memories may be fixedly mounted or portable.

[0038] Further features of, and advantages with, the present invention will become apparent when studying the appended claims and the following description. The skilled person realize that different features of the present invention may be combined to create embodiments other than those described in the following, without departing from the scope of the present invention.

[0039] Brief Description of the Drawings

[0040] These and other aspects of the present invention will now be described in more detail, with reference to the appended drawings showing example embodiments of the invention, wherein:

[0041] Fig. i schematically illustrates an example of a robot system according to the present invention;

[0042] Fig. 2 schematically illustrates an example of a dual spot weld inspection system having a vision-based system and an ultrasonic transducer, in which the dual spot weld inspection system is arranged on the robot system in Fig. 1, according to the present invention;

[0043] Fig. 3 is a cross sectional view of a spot weld on a workpiece;

[0044] Fig. 4 shows an image of the spot weld in Fig. 3; and

[0045] Fig. 5 is a flow-chart of an example of a method for inspecting a spot weld using the dual spot weld inspection system of Fig. 2, in which the inspection is controlled and performed by the robot system of Fig. 1, according to the present invention.

[0046] Detailed Description

[0047] In the present detailed description, various embodiments of robot systems and methods are mainly described with reference to a robot system comprising an industrial robot. However, the described robot system is suitable for any type of system and arrangement comprising a robot, such as an industrial robot, a service robot and the like. Moreover, the described method, control system and controller may be suitable for remote operations in relation to the robot system. The same or similar reference numerals will be used to denote the same or similar structural features. For ease of reference, the examples are described in relation to a workpiece in the form of a car body. Referring to Fig. 1, there is illustrated a robot system 1 according to one example. The robot system i here comprises four robots 2. Three of the robots 2 are welding robots configured to weld spot welds 30 on one or more workpieces 40, such as vehicle bodies. The workpiece maybe made of a metal. As shown in Fig. 1, the robot system 1 further comprises a robot 2 configured to operate as a weld inspection robot 2a. The weld inspection robot 2a is arranged downstream the other robots 2 to perform inspection of a weld on the workpiece 40 that has been created by one of the other robots 2. In other examples, the weld inspection robot 2a can be arranged in between other robots 2. The weld inspection robot 2a is configured to inspect the one or more spot welds 30 in a non-destructive manner. The weld inspection robot 2a is here an integral part of the robot system 1. In other examples, the robot system 1 may only constitute the weld inspection robot 2a, which is then arranged as a separate working station in an assembly line, or manufacturing line, depending on type of workpiece and type of manufacturing line. For ease of reference, the weld inspection robot 2a may hereinafter be referred to simply as the robot 2a.

[0048] As shown in Fig. 1, the robot 2a comprises a robot manipulator arrangement 3. The robot manipulator arrangement 3 typically comprises a robot arm arrangement having one or more robot arms. The robot 2a may be configured to be in communication with the other robots 2 through a control system 90. Hence, as illustrated in Fig. 1, the robot system 1 may comprise the control system 90, which here includes one or more controllers 90a to 9od for the respective robots 2, 2a. In this example, the control system 90 comprises processing circuitry 92 and a memory 94. The control system 90 may be configured to be in communication with the controllers 90a to 9od. The controllers 90a to 9od are configured to control the operation(s) of the respective robots 2, 2a of the robot system 1.

[0049] The robot systems and methods as described herein are particularly useful for non-destructive inspection of spot welds already formed on the workpieces. Such spot welds can be denoted as post -welds. A post-weld refers to the state or condition of a weld after the welding process has been completed, in contrast to ongoing welding, which takes place during the active formation of the weld. In post-weld inspection or evaluation, the welded joint is no longer being subjected to heat or other factors used to form the weld. Instead, the quality and integrity of the final weld is performed through a non-destructive testing method. Accordingly, the term non-destructive post-weld inspection refers to an inspection of the final quality of a weld.

[0050] One example of the robot 2a will now be described in relation to Figs, i and 2. The robot 2a is configured to perform a non-destructive spot weld inspection method including visual inspection and ultrasonic testing. Thus, the robot 2a comprises a dual spot weld inspection system 50. As shown in Figs. 1 and 2, the dual inspection system 50 is connected to the robot manipulator arrangement 3. Through this configuration, the dual inspection system 50 is configured to be moved over the workpiece 40 and the weld(s) 30.

[0051] In Fig. 1, the robot 2a comprises the controller god, which is configured to control the operations of the dual inspection system 50. Similar to the configurations of the other controllers 90a to 90c, the controller 90b may also be configured to control the operations of the robot manipulator arrangement 3. It should be noted that the controller 90b may be an integral part of the control system 90, or a stand-alone unit configured to be in communication with the control system 90 and the other controllers 90a to 90c. The controller 9od typically comprises corresponding processing circuitry 92. The controller 9od is configured to execute one or more control algorithms and motion instructions, thereby managing the movements and operations of the robot 2a and the dual spot weld inspection system 50. The controller 90b here also comprises a memory 94.

[0052] As shown in Fig. 1 in conjunction with Fig. 2, the dual spot weld inspection system 50 comprises a vision-based system 52. The vision-based system 52 is configured to capture an image 60 of a surface 35 of the spot weld 30, as seen in Fig. 4. The vision-based system 52 is here a still camera configured to capture at least one image 60 of the spot weld 30. The vision-based system 52 is in communication with the controller 9od. The image 60 is transferred to the controller god. In addition, or alternatively, the image 6o is stored in the memory 94, as shown in Fig. 1. Accordingly, the controller god is configured to receive the captured image 60. Typically, the controller god receives the full image from the vision-based system 52 for processing and analysis.

[0053] The vision-based system 52 is arranged in, or on, the dual spot weld inspection 50 at a suitable distance above the spot weld, as shown in Fig. 2. The visionbased system 52 can be controlled by the controller god to capture at least one image 60 of the spot weld 30. The captured image 60 of the spot weld 30 comprises image data representative of a visual illustration of the spot weld 30. As such, the vision-based system 52 captures image data to generate a representation of the spot weld 30. Typically, the controller god may be configured to determine the position of the spot weld on the workpiece 40, and then control the vision-based system 52 to capture an image on a certain spot weld 30 based on the determined position. As such, the dual spot weld inspection system 50 is here also configured to obtain positioning data indicative of an approximate position and orientation of a certain spot weld 30 in relation to the robot 2a. By way of example, the robot 2a is directed to a preprogrammed position, where the spot weld is nominally expected to be located. Given the possibility of positional deviations (e.g., up to 5 mm) from the nominal position, the system can use the vision-based system 52 to capture an image of the area and identify the precise location of the spot weld. This allows for accurate alignment before further inspection of the spot weld, using the vision-based system 52 to capture one or more images.

[0054] Moreover, as shown in Figs. 1 and 2, the dual spot weld inspection system 50 comprises an ultrasonic transducer 54. The ultrasonic transducer 54 is configured to acquire data of the spot weld 30. In one example, the ultrasonic transducer 54 is configured to acquire data indicative of a subsurface 36 of the spot weld 30, as seen in Fig. 3. In this context, the term “subsurface” typically refers to the inner volume or the internal portion of the material of the spot weld 30 that is beneath the surface of the weld. In weld inspection, subsurface refers to regions below the visible surface where defects such as voids, porosity, inclusions, or internal cracks can form during the welding process. The term “surface” typically refers to the external layer of the material of the spot weld 30. In the context of weld inspection, the surface includes the visible exterior of the spot weld 30 where potential defects, such as spatter, cracks, misalignments, or other anomalies, can occur.

[0055] The ultrasonic transducer 54 is here a single ultrasonic transducer. In other examples, the ultrasonic transducer 54 is a phased array ultrasonic scanner comprising an array of ultrasonic transducers. The ultrasonic transducer 54 is moveably arranged on the robot 2a, such that the ultrasonic transducer 54 can be brought in physical contact with the spot weld 30 to be inspected. In other examples, the ultrasonic transducer 54 may comprise a plurality of ultrasonic transducers. In such configurations, it may also be suitable to apply an ultrasonic gel or another appropriate coupling medium in physical contact with both the respective spot weld 30 and the ultrasonic transducers to obtain an even more accurate and valid test measurement. Non-contact ultrasonic inspection of the spot weld 30 is also feasible. In such a configuration, the ultrasonic transducer 54 may use air as the medium for sound wave transmission. For example, the ultrasonic transducer 54 may include one or more air-coupled transducers designed to generate and receive sound waves without direct contact with the surface of the spot weld 30. Other possible noncontact methods include laser ultrasonic testing, where a laser generates and detects ultrasonic waves, and electromagnetic acoustic transducers 54, which induce ultrasonic waves through electromagnetic fields without physical contact.

[0056] The controller 90b is configured to receive data acquired by the ultrasonic transducer 54. Accordingly, the controller 90b is configured to receive the captured image 60 of the spot weld 30 from the vision-based system 52 and also receive acquired data indicative of the subsurface of the spot weld 30 from the ultrasonic transducer 54.

[0057] From the received captured image 60 and the acquired subsurface data, the controller 90b is allowed to evaluate the spot weld 30 and identify one or more possible anomalies 31, 32 of the spot weld 30. In one example, the controller 9od is configured to identify at least one possible anomaly 31, 32 of the spot weld 30 from both the captured image 60 and the acquired data. In another example, the controller 90b is configured to identify a possible anomaly 32 of the spot weld 30 only from the captured image 60. In yet another example, the controller 90b is configured to identify a possible anomaly 31 of the spot weld 30 only from the acquired data. As such, the examples herein include a controller 9od capable of identifying a possible anomaly 31, 32 of the spot weld 30 from the captured image 60 and the acquired data.

[0058] More specifically, the controller 90b is configured to perform a first evaluation of the spot weld 30 from the received captured image 60 to identify a first possible anomaly of the spot weld 30. The first possible anomaly of the spot weld 30 is here a possible surface anomaly 32. Byway of example, the surface anomaly 32 is any one of a weld spatter, a crack, an incomplete fusion, a surface deformation, a misalignment of the spot weld, and a miscolor of the spot weld. Fig. 4 illustrates an example of an image 60 captured by the vision-based system 52. The image 60 is captured from the above of the surface 35 of the spot weld 30, and thus represents a top view of the spot weld 30. The spot weld 30 thus extends in a longitudinal direction X and in a transverse direction Y. Moreover, as shown in Fig. 4, the surface of the spot weld comprises an anomaly in the form of the surface anomaly 32.

[0059] Detecting the surface anomaly 32 from the captured image 60 typically involves evaluating the surface 35 of the spot weld 30, e.g. by analyzing the visual characteristics of the weld surface 35 to identify any deviations from expected patterns. In this example, the controller 90b is configured to perform the first evaluation to identify a possible surface anomaly (first anomaly) by comparing the captured image 60 of the spot weld 30 with one or more reference images 62 of ordinary spot welds. As such, the controller 9od is configured to detect deviations indicating a possible surface anomaly. The reference images 62 are typically also stored in the memory, as shown in Fig. 1. In one example, the controller 90b is configured to determine that the spot weld 30 is a potential stick weld by comparing the captured image 60 of the spot weld 30 with one or more reference images 62. The reference images 62 constitute examples of spot welds having predefined acceptable weld characteristics. The reference images may include data or represent spot weld with defect-free welds for baseline comparison. In one example, the controller 9od maybe configured to apply a machine learning model trained to recognize acceptable and defective weld patterns. Such machine learning model may be an Al -based model, and may include edge detection algorithms, such as Sobel, Canny, or Prewitt filters. Such algorithms are commonly used to highlight changes in pixel intensity, which here may represent cracks or sharp edges on the weld surface. Alternatively, or in addition, the controller 90b may apply a machine learning models configured for pattern recognition. Such machine learning models can be convolutional neural networks (CNNs), that can be trained to recognize and classify different types of surface defects based on the visual data, image segmentation and / or histogram analysis, in which pixel intensity distributions are analyzed to detect areas with deviations from the norm, which might indicate anomalies like discoloration or surface damage. For instance, a detected miscolor could indicate improper heat application during the welding process.

[0060] By comparing the captured image 60 of the spot weld 30 with at least one reference image 62 indicative of an ordinary spot weld, the controller 90b determines whether the spot weld in the captured image 60 contains a possible surface anomaly 32. As such, the controller 90b is configured to determine that the spot weld 30 is a possible defective spot weld 30 based on the first evaluation, i.e. based on the outcome of the first evaluation.

[0061] If no surface anomaly is detected, the spot weld 30 is classified as acceptable, and no further inspection action is taken. The controller 90b may log the inspection result for quality assurance purposes, ensuring traceability and allowing the weld to be approved for continued use in the manufacturing process. However, if the controller 90b detects a surface anomaly in the spot weld 30 based on the outcome of the first evaluation, e.g. based on the above comparison, the spot weld 30 is classified as defective, typically requiring further inspection. Depending on the nature and severity of the anomaly, the controller god can flag the spot weld 30 for further manual inspection by an operator or initiate a subsequent automated process to mark the spot weld 30 for rework or repair. In some cases, the detection of the possible surface anomaly 32 may prompt the controller 90b to conduct additional subsurface inspection using ultrasonic testing, allowing for further analysis of the subsurface structural integrity.

[0062] The controller 90b is configured to notify the user of the detected possible defective spot weld 30, providing information on the type and location of the defect. In addition to notifying the user, the controller 90b logs all relevant data, such as the anomaly detection, image comparisons, and subsequent actions, for subsequent comprehensive quality control. Depending on the application of the controller 90b, the controller 90b may instruct automated tools of the robot system 1 to proceed with further testing or to isolate the workpiece 40 with the defective spot weld 30 for rework.

[0063] Moreover, the controller 9od is configured to perform a second evaluation of the spot weld 30 from the received acquired data to identify a second possible anomaly in the form of a possible subsurface anomaly 31 in the spot weld 30. In this example, the subsurface anomaly 31 refers to any one of an inclusion defect, a porosity defect, and an internal crack. Fig. 3 shows an example of a spot weld 30 with a subsurface anomaly 31. The spot weld 30 of Fig. 3 has been inspected by the ultrasonic transducer 54. Fig. 3 depicts the spot weld 30 along the longitudinal direction X and in a vertical direction Z. The spot weld 30 is here a weld formed to join two parts of the car body 40 (workpiece). Reference 31 indicates the subsurface anomaly of the spot weld 30. Reference 36 indicates the region of the subsurface.

[0064] More specifically, data, typically including measurable data, is acquired by the ultrasonic transducer 54. The acquired data is subsequently transferred from the ultrasonic transducer 54 to the controller 90b for further evaluation, here corresponding to the second evaluation of the spot weld 30. In this example, the controller god is configured to perform the second evaluation of the spot weld 30 to identify a possible subsurface anomaly 31 by analyzing the reflection and attenuation patterns of the ultrasonic signals acquired by the ultrasonic transducer 54. These signals represent how sound waves propagate through the spot weld 30, and any deviations in their behavior can indicate internal defects, such as inclusions, porosity, or cracks.

[0065] In particular, in the second evaluation, the controller god evaluates how the ultrasonic waves are reflected back or attenuated when passing through the material. A correctly formed spot weld 30 will typically produce a predictable reflection and attenuation pattern, based on its uniform internal structure. In contrast, an anomaly, such as an internal crack or void, may cause the ultrasonic waves to reflect at unexpected angles or attenuate more significantly, due to the disruption in the material’s consistency.

[0066] By comparing the acquired data to an expected signal profile for a correctly formed spot weld 30, the controller god determines whether a subsurface anomaly 31 is present. If the signal patterns match the expected profile, the controller god concludes that no subsurface anomaly exists, and the spot weld 30 is classified as structurally sound. No further subsurface inspection maybe needed. Additionally, the controller god may log the inspection results for traceability and quality assurance purposes.

[0067] However, if the controller god detects deviations from the expected signal profile, the controller god identifies the weld as containing a possible subsurface defect, such as an internal crack, void, or porosity. Subsequently, the controller god determines that the spot weld 30 is a possible defective spot weld 30 based on the second evaluation. Typically, the spot weld 30 is also flagged for further inspection or rework, depending on the severity of the defect. The controller god may instruct automated systems to mark the spot weld 30 for rework or repair, or it may send a notification to the user, detailing the type and location of the detected anomaly. The controller god may also log the detected possible defect of the spot weld 30 and the actions taken. The ability of the controller god to analyze and compare ultrasonic signal patterns allows for a more precise detection of subsurface defects that are otherwise invisible through surface inspection methods. Such complementary analysis provides a more comprehensive evaluation of the weld's overall integrity.

[0068] To this end, the controller god is configured to determine that the spot weld 30 is a possible defective spot weld 30 based on the performed first and second evaluations. The term “defective spot weld” refers to a spot weld that exhibits one or more anomalies, either on the surface or subsurface. These anomalies may arise during the welding process and can affect the performance, durability, or safety of the workpiece. By combining visual inspection and ultrasonic evaluation, the system ensures comprehensive detection and classification of such defects.

[0069] In examples where both surface and subsurface anomalies are detected in the first and second evaluations, the defect classification may be escalated by the controller god, and an appropriate precautionary action maybe taken by the controller god, such as notifying the user.

[0070] As such, the controller god may typically be configured to notify a user of a determined possible defective spot weld 30 based on the determinations made from the captured image 60 in the performed first evaluation and from the analysis of the acquired data in the performed second evaluation.

[0071] In one example, the controller god is further configured to notify a user of a determined type of the possible anomaly in the spot weld 30. In such example, the controller god notifies the user whether the detected possible anomaly amounts to a surface anomaly, a subsurface anomaly, or a combination thereof. As such, by performing the first evaluation, the control god can identify a possible surface anomaly based on the captured image 60, and by performing the second evaluation, the control god can identify a possible subsurface anomaly. Determining that the spot weld 30 is a possible defective spot weld based on the performed first and second evaluations can be performed in several different manners. For example, the controller 90b may be configured to integrate the results of the two evaluations. If both evaluations indicate anomalies, the controller 90b typically determines the weld 30 to be a possible defective spot weld. In addition, or alternatively, if the subsurface evaluation in the second evaluation reveals a critical anomaly (e.g., void or incomplete penetration), the determination of a possible defect can be made without relying on the surface evaluation in the first evaluation. In addition, or alternatively, if only surface anomalies are detected (e.g., miscolor or spatter) in the first evaluation, the determination of a possible defect is typically made after confirming no critical subsurface flaws in the second evaluation.

[0072] The determination of a possible defective spot weld 30 can be based on predefined thresholds relating to the identified possible anomalies in the first and second evaluations. For surface anomalies, the threshold may be indicative of miscolor intensity, crack length, and / or spatter size exceeding acceptable limits. Such anomalies may independently indicate a defect. For subsurface anomalies, the threshold may be indicative of an insufficient penetration depth and / or excessive void size detected through ultrasonic data. Such anomalies may independently indicate a defect. For example, in the first evaluation, the captured image 60 shows a miscolor pattern indicative of improper heat distribution, while, in the second evaluation, the acquired subsurface data confirms insufficient penetration depth. In such situation, the controller 90b determines the spot weld as possibly defective. In another example, in the second evaluation, the ultrasonic data reveals voids in the spot weld 30, but, in the first evaluation, the captured image 60 shows no surface anomalies. However, in such situation, the controller 90b still marks the spot weld 30 as potentially defective due to the critical subsurface anomaly. In some example, the controller 90b maybe configured to assign weights to anomalies from the two evaluations to calculate a defect likelihood score. For example surface anomaly contributes 40% to the score, while subsurface anomaly contributes 6o% to the score. If the combined score exceeds a threshold, the spot weld 30 is marked as potentially defective by the controller god.

[0073] The controller god maybe configured to control the dual spot weld inspection system 50 to perform a single non-destructive inspection pass over the spot weld 30, during which the vision-based system 52 captures the image 60 and the ultrasonic transducer 52 acquires data of the spot weld 30. The configuration to perform a single non-destructive inspection pass allows for increased efficiency and reduced inspection time by enabling simultaneous or sequential data collection using both the vision-based system and the ultrasonic transducer. Such configuration may also eliminate the need for multiple passes over the spot weld.

[0074] The ultrasonic transducer 54 may acquire data of the spot weld 30 prior to the vision-based system 52 capturing the image 60 during the single nondestructive inspection pass. Acquiring data with the ultrasonic transducer 54 prior to capturing the image 60 allows for detecting more critical subsurface anomalies first. Such approach may reduce the risk of unnecessary processing of surface data if a flaw is already detected in the subsurface layer, thereby improving system resources and prioritizing potentially more critical subsurface defects for evaluation.

[0075] The controller god maybe configured to perform the second evaluation of the spot weld 30 from the received acquired data to determine the presence or absence of the second possible anomaly in the spot weld 30. Moreover, for any spot weld 30 determined in the second evaluation to be absent of at least one second possible anomaly, the controller god performs the first evaluation of the spot weld 30 from the received captured image 60 to determine that the spot weld 30 is a potential stick weld by comparing the captured image 60 with one or more reference images 62 indicative of a stick weld. The conditional approach of performing the second evaluation in a first operation of the controller god to detect the presence or absence of subsurface anomalies before conducting a surface analysis (in the first evaluation) provides a safeguard against unnecessary computational overhead. Furthermore, such configuration may enhance the reliability of the inspection process by focusing on detecting potential stick welds when no critical subsurface defects are identified, improving accuracy in quality assessment, and ensuring the most relevant defects are prioritized for further analysis. The configuration of the controller god that performs the first evaluation and the second evaluation is based on the challenge that ultrasonic testing may face difficulty in reliably detecting stick welds. A stick weld occurs when no weld nugget forms, and the zinc coating soldered the plates of the workpiece together. In ultrasonic testing, the presence of air between the plates may indicate the absence of a weld. However, in stick welds, there is no air gap, making it challenging to identify such flaw using standard ultrasonic methods only. Instead, advanced algorithms must often be applied to ultrasonic signals to identify potential stick welds, but such algorithms are complex and unreliable. As a result, stick welds are frequently missed during ultrasonic inspection. Stick welds also tend to have a paler appearance compared to more "healthy" welds. This is because stick welds are typically caused by insufficient current during the spot-welding process, which results in less heat and a lack of the characteristic "burned" appearance of a properly formed weld. To address such challenge, the robot 2a uses the vision-based system 52 to capture an image of the spot weld 30. The captured image 60 is then analyzed by the controller god in the first evaluation to identify a likelihood of a stick weld. In contrast, other types of flaws may be more reliably identified in the second evaluation based on ultrasonic testing, where the vision-based system 52 may be less effective. To this end, by combining visual data from the vision-based system 52 with data from the ultrasonic transducer 54, the inspection performed by the controller 90b benefits from the complementary strengths of the two methods. The first evaluation based on the vision-based system 52 excels at detecting stick welds, a known weakness of ultrasound, while the second evaluation based on ultrasound is more effective at identifying other flaw types. Typically, if the first evaluation identifies a high likelihood of a flaw in the spot weld 30, but the second evaluation using ultrasonic indicates that the spot weld is acceptable, the controller 90b still determines that the spot weld 30 is a possible defective spot weld. The controller 90b thus marks the spot weld 30 as suspicious and flags it for further inspection. Such combined inspection process may thus reduce the risk of flawed welds leaving the production line. By leveraging the complementary strengths of vision and ultrasonic methods, the system can achieve higher accuracy than conventional ultrasonic-only inspection systems. Detecting even a small number of additional stick welds represents an improvement, as stick welds can be catastrophic flaws that prevent the plates from being held together with sufficient force.

[0076] In one example, the controller god is configured to control the ultrasonic transducer 54 to acquire data before the vision-based system 52 is controlled to capture an image 60 of the spot weld 30. In one example, the controller 90b is configured to control the vision-based system 52 to capture an image of the spot weld 30 before the ultrasonic transducer 54 acquires data of the spot weld 30. In other examples, the controller 90b controls the ultrasonic transducer 54 to acquire data in parallel with the vision-based system 52 to capture image.

[0077] In one example, the controller 9od is configured to always perform the first evaluation whenever no subsurface anomaly is detected in the second evaluation. Such configuration serves as a safeguard when the ultrasonic inspection does not detect any subsurface anomalies. Typically, the ultrasonic transducer 54 may primarily check for subsurface anomalies, which may be hidden within the spot weld 30 and not visible to the naked eye. However, ultrasonic testing alone may occasionally miss certain surface defects or subtle issues that do not affect the internal structure but still compromise the overall quality of the weld. By configuring the controller 90b to always perform the first evaluation, using the vision-based surface inspection, when no subsurface anomaly is found, the controller 90b is configured to act as a safeguard system. That is, in examples where the ultrasonic transducer 54 does not detect a problem internally, the surface inspection using the captured image 60 can still identify surface anomalies (e.g., spatter, cracks, or misalignment) that may affect the performance of the weld. For completeness, in examples where no anomaly is identified, the controller 9od may typically approve the spot weld, and allow the workpiece 40 to proceed in the manufacturing process.

[0078] As such, whether confirming weld integrity or detecting a defect, the robot system 1 with the configuration of the controller god provides a thorough evaluation of both surface and subsurface conditions.

[0079] In one example, the controller 90b is configured to calculate a size of a welded area within the spot weld 30 based on the acquired data from the ultrasonic transducer 54. By way of example, the ultrasonic transducer 54 emits high- frequency sound waves into the spot weld area. Reflected signals (echoes) are captured by the transducer 54, with variations in signal intensity and timing indicating transitions between welded and unwelded regions. The received echoes are processed by the controller 90b to generate a subsurface map of the spot weld. In such example, welded regions are characterized by specific acoustic impedance patterns that differ from unwelded regions. Typically, a predefined threshold can be applied to distinguish welded areas from unwelded ones based on echo amplitude and time-of-flight data. Machine learning models or advanced signal analysis algorithms may be employed to improve classification accuracy, particularly for complex weld geometries. The detected welded regions are then quantified by measuring their dimensions in the subsurface map. Optionally, an algorithm may be applied to calculate the total area of the weld by integrating the detected welded regions. The processed data is subsequently analyzed by the controller 90b for irregular patterns, such as voids or cracks, that could indicate defects. The calculated welded area and any detected anomalies are the output to the controller 9od.

[0080] It should be noted that the operation of determining the position of the spot weld 30 on the workpiece 40 can be performed in several different ways. Typically, the robot 2a is directed to a pre-programmed position, where the spot weld 30 is nominally expected to be located. Given the possibility of positional deviations (e.g., up to 5 mm) from the nominal position, the robot 2a uses the vision-based system 52 to capture an image 60 of the area and identify the precise location of the spot weld 30. This allows for accurate alignment before further inspection. Alternatively, the ultrasonic transducer 54 can be employed to measure the weld at the nominal position. This provides feedback to the control system 90, and / or the controller 90b, regarding the actual position of the weld. If the ultrasonic transducer 54 detects part of the spot weld 30, the controller 90b can adjust the position of the robot 2a and take another measurement to refine the alignment.

[0081] In one example, the control system 90 may transfer data indicative of a nominal position of each spot weld 30 in relation to the vehicle body 40 to the controller 90b of the robot 2a. Analogously, the controller 90b may be configured to communicate the nominal position of each spot weld 30 in relation to the vehicle body 40 to the dual spot weld inspection system 50. The nominal position may be a predetermined theoretical position that has been determined by means of a welding program running in respective welding robot controller 90a to 90c of the welding robots 2.

[0082] To this end, the controller 90b maybe configured to determine, on the basis of the nominal positions in relation to the vehicle body 40, an approximate position of each spot weld 30 in relation to the robot 2a. Alternatively, or in addition, the controller 90b may receive data indicative of the approximate positions and orientations of the spot welds 30 in relation to the robot 2a from the memory 94 of the control system 90. As such, the position and orientation of the vehicle body 40 in relation to the robot 2a can be determined for the purpose of performing the inspection of the spot weld by the dual spot weld inspection system 50.

[0083] As the approximate positions and orientations of the spot welds 30 in relation to the robot 2a are known, the robot 2a is configured to be moved over the relevant area of the vehicle body to direct the vision-based system 52 towards certain individual spot welds 30 or certain groups of spot welds 30. In this manner, the vision-based system 52 will be in the position to analyze the resulting image data to determine the real position of each spot weld 30 in relation to the robot 2a. As soon as the real position of each spot weld 30 in relation to the robot 2a is established, the robot 2a is in the position to bring the ultrasonic transducer 54 in contact with each spot weld 30 to obtain a respective test measurement, and also to capture one or more additional images by the vision-based system 52.

[0084] The ultrasonic transducer 54 should typically be moved against each spot weld 30 with the same constant force. The ultrasonic transducer 54 can be arranged in the dual inspection system 50 in various ways. Fig. 2 depicts one example, in which the dual inspection system 50 comprises a damper arrangement 73. The function of the damper arrangement 73 is to regulate the force with which the ultrasonic transducer 54 is moved in contact with the vehicle body 40 and the spot weld 30. The ultrasonic transducer 54 should typically exert the same constant force on each spot weld 30 regardless of the orientation of the ultrasonic transducer 54 in relation to the respective spot weld 30. The damper arrangement 73 may comprise a spring 76 pushing the ultrasonic transducer 54 in a linear direction 72 in relation to the damper arrangement 73, and a distance measurement device 74 measuring a distance 76 between the ultrasonic transducer 54 and a reference position 75 within the damper arrangement 73. As the ultrasonic transducer 54 comes into contact with the spot weld 30 on the vehicle body 40, the spring 76 starts to be pushed together and the distance 71 between the ultrasonic transducer 54 and the reference position 75 starts to decrease, which is detected by the distance measurement device 74. The distance 71 and thereby the force exerted by the ultrasonic transducer 54 on the spot weld 30 can be kept constant from one test measurement to another on the basis of the reading of the distance measurement device 74. The damper arrangement 73 may comprise a pneumatic cylinder pushing the ultrasonic transducer 54 in the linear direction 72 in relation to the damper arrangement 73. The pressure within the cylinder can be controlled to compensate for gravity such that the same constant force is applied irrespective of the orientation of the ultrasonic transducer 54 in relation to the gravity field.

[0085] Fig. 5 is a flowchart of exemplary steps of one example of a method 100 for non-destructive inspection of a spot weld 30 on a workpiece 40. In this example, the method 100 is performed by the controller god. As such, the controller god here comprises processing circuitry g2 configured to perform the method 100 according to the examples. Accordingly, the method 100 is here a computer-implemented method. In other examples, the overall control system go of the robot system i maybe configured to implement the method 100 according to examples. The method 100 is intended for inspecting postwelds on the workpiece 40 using the robot system 1 and the robot 2a in FIGS. 1 to 3. The processing circuitry g2 is configured to control the dual inspection system 50. The processing circuitry g2 is configured to control the vision-based system 52 and the ultrasonic transducer 54 of the dual inspection system 50. The processing circuitry g2 is configured to perform the following steps. The method 100 comprises a step no of capturing an image 60 of the surface of the spot weld 30 using the vision-based system 52. Moreover, the method 100 comprises a step 120 of acquiring data of the spot weld, such as data indicative of the subsurface of the spot weld, using the ultrasonic transducer 54. Subsequently, the method 100 comprises a step 130 of receiving the captured image 60 and the acquired data. Furthermore, the method 100 comprises a step 140 of performing a first evaluation of the spot weld 30 using the received captured image 60 to identify a first possible anomaly of the spot weld 30. In addition, the method 100 comprises a step 150 of performing a second evaluation of the spot weld 30 using the received acquired data to identify a second possible anomaly of the spot weld 30. Subsequently, the method 100 comprises a step 160 of determining that the spot weld 30 is a possible defective spot weld 30 based on the performed first and second evaluations.

[0086] In one example, the step no of capturing an image 60 of a surface of the spot weld 30 using the vision-based system 52 and the step of acquiring 120 data of the spot weld 30 using the ultrasonic transducer 54 are performed in a single non-destructive inspection pass over the spot weld.

[0087] In one example, the step of performing the second evaluation of the spot weld 30 is completed prior to the step of performing the first evaluation of the spot weld 30. Moreover, in such an example, the second evaluation comprises determining the presence or absence of the second possible anomaly in the spot weld 30; and for any spot weld 30 determined in the second evaluation to be absent of at least one second possible anomaly, the step of performing the first evaluation of the spot weld 30 comprises to determine that the spot weld 30 is a potential stick weld by comparing the captured image 60 with one or more reference images indicative of a stick weld.

[0088] In one example, the method 100 comprises performing the second evaluation of the second possible anomaly of the spot weld 30 from the received acquired data in response to that the first evaluation of the first possible anomaly of the spot weld identifies a possible surface anomaly in the spot weld 30.

[0089] In one example, the method 100 comprises performing the second evaluation of the second possible anomaly of the spot weld 30 from the received acquired data in response to that the first evaluation of the first possible anomaly of the spot weld does not identify a possible surface anomaly in the spot weld 30.

[0090] In one example, the method 100 comprises performing the first evaluation of the first possible anomaly of the spot weld 30 in response to that the second evaluation of the second possible anomaly of the spot weld 30 identifies a possible subsurface anomaly in the spot weld.

[0091] In one example, the method 100 comprises performing the first evaluation of the first possible anomaly of the spot weld 30 in response to that the second evaluation of the second possible anomaly of the spot weld 30 does not identify a possible subsurface anomaly in the spot weld.

[0092] In one example, step 140 comprises identifying a possible surface anomaly by comparing the captured image 60 of the spot weld 30 with one or more reference images of ordinary spot welds, so as to detect deviations indicating a surface anomaly.

[0093] In one example, step 150 comprises identifying a possible subsurface anomaly by analyzing the reflection and attenuation patterns of the ultrasonic signals acquired by the ultrasonic transducer 54. In one example, the possible subsurface anomaly is identified based on a comparison with an expected signal profile for a correctly formed spot weld.

[0094] In one example, the step of identifying a possible subsurface anomaly of the spot weld from the acquired data is only performed if the captured image 6o indicates a possible surface anomaly in the spot weld 30.

[0095] In one example, the step of identifying a possible surface anomaly of the spot weld from the captured image 60 is only performed if the acquired data indicates a possible subsurface anomaly in the spot weld 30.

[0096] In one example, step 140 of performing the first evaluation to identify a possible surface anomaly of the spot weld 30 from the captured image 60 may be performed by processing the captured image 60 of the spot weld using a machine learning model configured to evaluate whether the captured image 60 fits within a set of reference images representing acceptable spot welds. The machine learning model is trained based on data from a plurality of images of acceptable spot welds and defective spot welds. Hereby, the method may detect deviations indicative of a surface anomaly. If the method 100 determines that the spot weld 30 in the captured image 60 is not within the set of reference images representing acceptable spot welds, the method in step 140 determines that the spot weld 30 in the captured image 60 contains a possible surface anomaly. As such, the method in step 160 determines that the inspected spot weld 30 is a possible defective spot weld based on the first evaluation in step 140.

[0097] By way of example, step 140 comprises applying the trained machine learning model to the captured image 60 of the spot weld 30 to perform the first evaluation. The machine learning model evaluates surface-specific characteristics of the spot weld, including e.g. miscolor and the shape of the weld. In one example, the trained machine learning model performs the first evaluation to determine whether the spot weld 30 requires further analysis to assess the surface anomaly of the spot weld 30. It should be noted that the vision-based system 52 and the ultrasonic transducer 54 can be arranged in several different ways in the dual spot weld inspection system 50. In one example, as illustrated in Fig. 2, the ultrasonic transducer 54 is arranged in a leading part of the dual spot weld inspection system 50 and the vision-based system 52 is arranged in a trailing part. In another example, the ultrasonic transducer 54 is arranged in the trailing part of the dual spot weld inspection system 50 and the vision-based system 52 is arranged in the leading part. The dual spot weld inspection system 50 may thus comprise a leading part and a trailing part. The leading part refers to the portion at the front of the dual spot weld inspection system 50, positioned forward relative to a direction of movement of the dual spot weld inspection system 50 along the workpiece 40. The trailing part refers to the portion at the rear of the dual spot weld inspection system 50, positioned behind the leading part. The terms "leading" and "trailing" describe the spatial relationship of the parts of the dual spot weld inspection system 50, where "leading" indicates the front portion and "trailing" indicates the rear portion.

[0098] The above presentation of the robot system 1 should also be regarded as disclosing a control system 90 having one or more controllers 90a, 90b, 90c, 9od for controlling the robot system 1, for instance using the controller 90b and the processing circuitry 92. In addition, there is disclosed a controller 9od comprising processing circuitry 92 configured to control the robot system 1, the robot 2a and the dual spot weld inspection system 50.

[0099] As described herein, the disclosure also relates to the control system 90, comprising controllers 90a to 9od configured to execute the method 100 according to the above examples. The disclosure also relates a computer program comprising instructions to cause the controller 90b to execute the method of any of the above examples. It should be noted that the controller, as described herein, may include a microprocessor, microcontroller, programmable digital signal processor or another programmable processor device. The processing circuitry may also include a microprocessor, microcontroller, programmable digital signal processor or another programmable processor device, or instead, include an application specific integrated circuit, a programmable gate array or programmable array logic, a programmable logic device, or a digital signal processor. Where the processing circuitry includes a programmable device such as the microprocessor, microcontroller or programmable digital signal processor mentioned above, the processor may further include computer executable code that controls operation of the programmable device.

[0100] Even though the invention has been described with reference to specific exemplifying embodiments thereof, many different alterations, modifications and the like will become apparent for those skilled in the art. Also, it should be noted that parts of the system and method may be omitted, interchanged or arranged in various ways, the system and method yet being able to perform the functionality of the present invention.

[0101] Additionally, variations to the disclosed embodiments can be understood and effected by the skilled person in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

Claims

32Claims1. A robot system (1) for non-destructive inspection of a spot weld (30) on a workpiece (40), wherein the robot system comprises:- a dual spot weld inspection system (50) having a vision-based system (52) configured to capture an image (60) of a surface (35) of the spot weld, and further an ultrasonic transducer (54) configured to acquire data of the spot weld; and- a controller (90, god) configured to: receive the captured image and the acquired data; perform a first evaluation of the spot weld from the received captured image to identify a first possible anomaly (32) of the spot weld; perform a second evaluation of the spot weld from the received acquired data to identify a second possible anomaly (31) of the spot weld; and further configured to determine that the spot weld is a possible defective spot weld based on the performed first and second evaluations.

2. The robot system according to claim 1, wherein the first possible anomaly of the spot weld is different from a positional misplacement of the spot weld.

3. The robot system according to claim 1 or claim 2, wherein the first evaluation comprises identifying a possible anomaly in the form of a surface anomaly in the spot weld (30) based on the captured image.

4. The robot system according to claim 3, wherein the controller is configured to identify the possible surface anomaly by comparing the captured image of the spot weld with one or more reference images of ordinary spot welds.

5. The robot system according to any one of the preceding claims, wherein the second evaluation comprises identifying a possible anomaly in the form of a subsurface anomaly in the spot weld based on the data acquired by the ultrasonic transducer.

336. The robot system according to claim 5, wherein the controller is configured to identify a possible subsurface anomaly by analyzing the reflection and attenuation patterns of the ultrasonic signals acquired by the ultrasonic transducer.

7. The robot system according to claim 6, wherein the possible subsurface anomaly is identified based on a comparison with an expected signal profile for a correctly formed spot weld.

8. The robot system according to any one of the preceding claims, wherein the controller is configured to calculate a size of a welded area within the spot weld based on the acquired data from the ultrasonic transducer.

9. The robot system according to any one of the preceding claims, wherein the controller is configured to control the dual spot weld inspection system to perform a single non-destructive inspection pass over the spot weld, during which the vision-based system captures the image, and the ultrasonic transducer acquires data of the spot weld.

10. The robot system according to claim 9, wherein the ultrasonic transducer acquires data of the spot weld prior to the vision-based system capturing the image during the single non-destructive inspection pass.

11. The robot system according to any one of the preceding claims, wherein the controller is configured to perform the second evaluation of the spot weld from the received acquired data to determine the presence or absence of the second possible anomaly in the spot weld; and for any spot weld determined in the second evaluation to be absent of any second possible anomaly, perform the first evaluation of the spot weld from the received captured image to determine that the spot weld is a potential stick weld by comparing the captured image with one or more reference images indicative of a stick weld.

12. The robot system according to any one of the preceding claims, wherein the controller is further configured to notify a user of the determined possible defective spot weld.

13. A method (100) for non-destructive inspection of a spot weld (30) on a workpiece (40) using a dual spot weld inspection system (50) of a robot system (1), the dual spot weld inspection system comprising a vision-based system (52) and an ultrasonic transducer (54), the method comprising:- capturing (no) an image (60) of a surface of the spot weld using the vision-based system (52);- acquiring (120) data of the spot weld using the ultrasonic transducer (54);- receiving (130), by processing circuitry of a controller, the captured image and the acquired data;- performing (140), by the processing circuitry of the controller, a first evaluation of the spot weld using the received captured image to identify a first possible anomaly of the spot weld;- performing (150), by the processing circuitry of the controller, a second evaluation of the spot weld using the received acquired data to identify a second possible anomaly of the spot weld; and- determining (160), by the processing circuitry of the controller, that the spot weld is a possible defective spot weld based on the performed first and second evaluations.

14. The method according to claim 13, wherein capturing (no) an image (60) of a surface of the spot weld using the vision-based system (52) and acquiring (120) data of the spot weld using the ultrasonic transducer (54) are performed in a single non-destructive inspection pass over the spot weld.15- The method according to claim 13 or claim 14, wherein performing the second evaluation of the spot weld is completed prior to performing the first evaluation of the spot weld, the second evaluation comprises determining the presence or absence of the second possible anomaly in the spot weld; and for any spot weld determined in the second evaluation to be absent of any second possible anomaly, further performing the first evaluation of the spot weld to determine that the spot weld is a potential stick weld by comparing the captured image with one or more reference images indicative of a stick weld.

16. A computer program product comprising program code for performing, when executed by a controller, the method of any of claims 13 to 15.

17. A non-transitory computer-readable storage medium comprising instructions, which when executed by a controller, cause the controller to perform the method of any of claims 13 to 15.

Images