Welding robot system

Abstract

[Problem] To provide a remote-controlled welding robot system capable of reducing misalignment between a welding part of a workpiece and a welding torch. [Solution] A welding robot system 10 comprises: a welding robot 11; a camera 12 that captures an image of a welding part that has not been welded in a workpiece 5; a display device 13 that displays the captured image; a reception unit 15 that receives an instruction to correct a welding position from an operator in response to displaying of the image; a robot control unit 16 that controls the welding robot 11 on the basis of instruction data to perform welding and that performs control for correcting the welding position in response to the received instruction; an accumulation unit 17 that accumulates positions of a plurality of points on a track of the welding performed by the welding robot 11; and a prediction unit 18 that predicts a welding path by using the accumulated positions of the plurality of points. The robot control unit 16 controls the welding robot 11 so as to move a welding torch on the predicted welding path.

Description

Welding Robot System

[0001] The present invention relates to a welding robot system that operates a welding robot by remote control.

[0002] Welding of large structures such as ships and plant tanks is often performed in locations that are difficult for workers to access, such as high altitudes or narrow spaces, and therefore automated welding using welding robots has been introduced. Furthermore, in such welding, high-quality welding, for example, highly airtight welding, is required because any welding defects that may occur could lead to serious accidents. Since fully automated welding may not be able to ensure such high quality, a welding robot system that automatically welds workpieces is remotely operated by a worker who views video captured by a camera from a remote location and corrects the target position of a welding torch to compensate for deviations in the welded portion due to thermal distortion or the like (see, for example, Patent Document 1).

[0003] Special Publication No. 61-044595

[0004] 6, ideally, welding torch 111 should be able to move along a target welding path L100, but actual welding trajectory L103 may deviate from the target welding path L100. If the degree of deviation is within the range of boundary lines L101 and L102 that indicate the tolerance range, no problem will arise in the quality of the welding. However, if deviation exceeds boundary lines L101 and L102, a problem will arise in the quality of the welding.

[0005] In a remote-controlled welding robot system, as shown in FIG. 7 , the tip of the welding torch moves at a constant speed in the direction of travel along a teaching line L1 indicated by teaching data. In such a situation, for example, as shown in FIG. 7 , the welding portion L2 of the workpiece may change in the direction of distortion due to thermal distortion of the workpiece. To perform welding that tracks this change, the operator corrects the aim position of the welding torch while viewing the image captured by the camera, as indicated by the solid arrow in FIG. 7 . By making this correction, the deviation of the welding trajectory L201 from the welding portion L2 can be kept within an acceptable range. In a situation where such remote correction is being performed, if the image skips for a predetermined period due to, for example, a communication delay, correction cannot be made during the period without the image, as shown in FIG. 7 . Therefore, the welding torch moves in the direction of the teaching line L1. As a result, the deviation of the welding trajectory L201 from the welding portion L2 of the workpiece exceeds the acceptable range, resulting in a problem with the quality of the welding.

[0006] Even if there are no such periods without images, if the change in the welded portion L2 due to thermal distortion of the workpiece is significant, the frequency of correction of the welding trajectory L202 increases, as indicated by the solid arrow in Figure 8. Even if thermal distortion does not progress during welding and the welding trajectory L202 becomes linear halfway through, frequent corrections are required once the trajectory deviates from the initial teaching line L1. Because the worker makes corrections while viewing only the camera image, frequent corrections increase the worker's fatigue. This reduces the time the worker can concentrate on their work, requiring replacement workers to be prepared. This places a heavy burden on the worker when there are few skilled workers.

[0007] The present invention has been made to solve the above-mentioned problems, and aims to provide a remote-controlled welding robot system that can further reduce the misalignment between the welding portion of the workpiece and the welding torch.

[0008] In order to achieve the above-mentioned object, a welding robot system according to one aspect of the present invention comprises a welding robot that moves a welding torch, a camera attached to the welding robot that photographs the unwelded portions of the workpiece, a display device that displays the image photographed by the camera, a reception unit that receives instructions to correct the welding position from an operator in accordance with the image displayed on the display device, a robot control unit that controls the welding robot to perform welding based on teaching data and controls the correction of the welding position in accordance with the instructions received by the reception unit, a storage unit that accumulates the positions of multiple points on the trajectory of welding by the welding robot, and a prediction unit that predicts the welding path using the positions of the multiple points accumulated by the storage unit, and the robot control unit controls the welding robot so that the welding torch moves along the welding path predicted by the prediction unit.

[0009] With this configuration, even if the unwelded portion of the workpiece changes due to thermal distortion, the welding torch can be moved along the predicted welding path to reduce the misalignment between the unwelded portion of the workpiece and the welding torch. As a result, the frequency of remote corrections performed by the operator is reduced, thereby reducing operator fatigue. Therefore, for example, remote corrections can be performed by fewer operators. Furthermore, by moving the welding torch along the predicted welding path, even if the image skips for a predetermined period due to communication delays, the possibility that the welding torch misalignment will exceed the allowable range during that time can be reduced. In other words, a welding robot system that is more robust against communication delays can be provided.

[0010] In a welding robot system according to one aspect of the present invention, the prediction unit may identify the position of the approximation line so as to optimize an objective function corresponding to the difference between the multiple points and a predetermined linear approximation line, and predict a welding path corresponding to the identified position of the approximation line. This configuration makes it possible to predict a welding path that fits multiple points on the welding trajectory. In particular, if the approximation line is predicted using only the most recent points on the welding trajectory, the predicted approximation line may deviate significantly from the original approximation line if the operator mistakenly makes an excessively large correction. On the other hand, by identifying the position of the approximation line using multiple points on the welding trajectory, it is possible to identify the position of the approximation line that minimizes the error from the multiple points, thereby enabling the identification of an approximation line that changes smoothly over time without being overly sensitive to errors at some points.

[0011] In addition, in a welding robot system according to one aspect of the present invention, the prediction unit may optimize an objective function in which weights are assigned to the differences between the multiple points and the approximation line, with the weight decreasing as the difference between the current time and the welding time increases. This configuration allows the influence of more recent portions of the welding trajectory to be greater when predicting the welding path, thereby enabling a prediction of a welding path that better fits the current welding point of the welding torch. Furthermore, even if the shape of the unwelded portion of the workpiece gradually changes due to thermal distortion or the like, it is possible to predict a welding path that corresponds to the latest situation. For example, if the approximation line is predicted using only the most recent points, as described above, there is a possibility that the predicted approximation line will significantly deviate from the original approximation line if an erroneous correction is made. On the other hand, if the weights of the multiple points are equalized, the approximation line will not reflect the latest trends in the welding trajectory. In contrast, by using weights that decrease as the difference between the current time and the welding time increases, it is possible to identify the appropriate approximation line position, which solves these issues.

[0012] In the welding robot system according to an aspect of the present invention, the predetermined line shape may be a straight line. With this configuration, the welding path can be more easily predicted.

[0013] In the welding robot system according to an aspect of the present invention, the display device may display an image in which the direction of the welding path is oriented in a predetermined direction when the welding torch moves along the welding path predicted by the prediction unit. With this configuration, even if the direction of the welding path changes, the worker can correct the welding position while viewing the image in which the direction of the welding path is oriented in the predetermined direction, which reduces the worker's workload.

[0014] In the welding robot system according to an aspect of the present invention, the receiving unit may receive an instruction to correct the welding torch in a direction perpendicular to the welding path when the welding torch moves along the welding path predicted by the prediction unit. With this configuration, when the image is displayed so that the direction of the welding path is aligned in a predetermined direction, the operator can make more intuitive corrections.

[0015] According to a welding robot system according to one aspect of the present invention, it is possible to reduce the displacement of the welding torch from the welding portion of the workpiece.

[0016] Schematic diagram showing the configuration of a welding robot system according to an embodiment of the present invention. Schematic diagram showing an example of a welding robot system in the embodiment. FIG. Illustrates an example of a plurality of points on a welding trajectory and an approximation line in the embodiment. Flowchart showing the operation of a welding robot system in the embodiment. FIG. Illustrates a welding trajectory and its correction ... tolerance range. FIG. Illustrates a conventional welding trajectory and its correction. FIG. Illustrates a conventional welding trajectory and its correction.

[0017] A welding robot system according to the present invention will be described below using an embodiment. In the following embodiments, components and steps denoted by the same reference numerals are the same or equivalent, and repeated description may be omitted. The welding robot system according to this embodiment remotely corrects the welding target position and moves the welding torch along a welding path predicted using past welding trajectories.

[0018] 1 is a schematic diagram showing the configuration of a welding robot system 10 according to this embodiment. The welding robot system 10 according to this embodiment includes a welding robot 11 that moves a welding torch 11a, a camera 12 attached to the welding robot 11, a display device 13 that displays an image captured by the camera 12, a memory unit 14 that stores teaching data and the like, a reception unit 15 that receives instructions from an operator to correct a welding position, a robot control unit 16 that controls the welding robot 11 based on the teaching data and the like, a storage unit 17 that stores the positions of points on a welding trajectory, and a prediction unit 18 that predicts a welding path using the stored positions of the points, and may further include a welding power source 19 that supplies a high voltage used in welding to the welding torch 11a and the workpiece 5, as necessary.

[0019] In this embodiment, as an example, as shown in FIG. 1 , a case will be mainly described in which a welding robot 11 and a camera 12 are located at a work site 1 where welding is performed, a display device 13, a memory unit 14, a reception unit 15, and the like are located at a remote location 2 where remote operation is performed, and the two are connected by a cable 3. The cable 3 may include, for example, a communication line for transmitting signals of images captured by the camera 12, a control line for transmitting control signals for the welding robot 11, and a welding cable for transmitting a high voltage required for welding. Note that as long as the welding robot 11 and the camera 12 are located at the work site 1 and the display device 13 and the reception unit 15 are located at the remote location 2, the other components may be located at the work site 1, the remote location 2, or another location. This is because remote operation can still be performed in this case.

[0020] The welding robot 11 has a welding torch 11a and moves the welding torch 11a to weld the workpiece 5. The welding robot 11 may be, for example, a vertical articulated robot with the welding torch 11a attached to its end, or may be any other robot capable of moving the welding torch 11a in any direction in three dimensions. As shown in FIG. 2 , the welding robot 11 itself may be moved by a moving device such as an elevator 21. The elevator 21 may have, for example, a rail arranged vertically, a slider slidably mounted on the rail, and a moving means for moving the slider vertically. The moving means may include, for example, a rack and pinion and a driving means for driving the pinion, or a ball screw and a driving means for rotating the screw shaft. The welding robot 11 may be fixed to the slider. Note that Figure 2 shows a case in which four welding robots 11 are used to weld four locations on the workpiece 5, and four workers 7, checking the status of each welding location on a display device 13, adjust the welding target position of each welding robot 11. However, the number of welding robots 11 included in the remote-controlled welding robot system 10 and the number of workers remotely adjusting the target position of each welding robot 11 are not limited to this.

[0021] The welding performed by the welding robot 11 may be, for example, arc welding. In the remote-controlled welding robot system 10, TIG welding, which is an arc welding method, is generally performed, but other arc welding methods such as MIG welding may also be performed. The arc welding performed by the welding robot 11 may be, for example, non-consumable electrode welding such as TIG welding, or consumable electrode welding such as MIG welding.

[0022] The camera 12 is attached to the welding robot 11 and captures an image of an unwelded portion of the workpiece 5. The welded portion of the workpiece 5 may be, for example, a groove portion in groove welding, a fillet portion in fillet welding, or a butt portion in butt welding. The camera 12 acquires video as a moving image. It is preferable that the video captured by the camera 12 indicates the degree of misalignment between the tip of the welding torch 11a and the unwelded portion of the workpiece 5. Therefore, the camera 12 may capture, for example, a position of the unwelded portion of the workpiece 5 that is close to the tip of the welding torch 11a. Furthermore, the subject captured by the camera 12 may include, for example, the tip of the welding torch 11a as well as the unwelded portion of the workpiece 5.

[0023] The display device 13 displays the video captured by the camera 12. For example, the display device 13 may receive video transmitted from the camera 12 via a communication line and display the received video. Strictly speaking, the transmitted and received video is a video signal, but for convenience of explanation, it may be simply referred to as video. The display device 13 may be, for example, a liquid crystal display or an organic EL display.

[0024] The memory unit 14 stores teaching data for the welding robot 11. This teaching data may indicate, for example, the movement path (e.g., position, posture, etc.) of the tip of the welding torch 11a. Furthermore, for example, the teaching data may include information regarding welding conditions such as the start and end points of welding, welding current, welding voltage, etc. Furthermore, the storage unit 17 may store the positions of points on the welding trajectory in the memory unit 14.

[0025] The process by which the teaching data is stored in the storage unit 14 is not important. For example, the teaching data may be stored in the storage unit 14 via a recording medium, or teaching data transmitted via a communication line or the like may be stored in the storage unit 14, or teaching data acquired during teaching may be stored in the storage unit 14. The storage unit 14 is preferably realized by a non-volatile recording medium, but may also be realized by a volatile recording medium. The recording medium may be, for example, a semiconductor memory or a magnetic disk.

[0026] The reception unit 15 receives an instruction to correct the welding position from the worker in accordance with the image displayed on the display device 13. That is, the instruction to correct the welding position is received from the worker who is watching the image displayed on the display device 13. The worker corrects the welding position while watching the image displayed on the display device 13. This correction of the welding position may be, for example, a correction of the aim position of the welding torch 11a. The instruction to correct the welding position may include, for example, the direction of correction of the welding position and the degree (distance) of the correction. Note that the direction of correction may be selected from, for example, predetermined directions. The reception unit 15 may receive an instruction input from, for example, an input device (e.g., a teaching pendant, a joystick, etc.) or may receive an instruction transmitted via a wired or wireless communication line. Note that the reception unit 15 may or may not include a device for receiving the instruction (e.g., an input device, a communication device, etc.). The reception unit 15 may be realized by hardware or by software such as a driver that drives a predetermined device.

[0027] The robot control unit 16 controls the welding robot 11 to perform welding based on the teaching data stored in the memory unit 14. The robot control unit 16 also performs control to correct the welding position in response to an instruction received by the receiving unit 15. That is, when an instruction to correct the welding position is received by the receiving unit 15, the robot control unit 16 controls the welding robot 11 to change the welding position in response to the instruction. When the prediction unit 18 predicts a welding path, the robot control unit 16 controls the welding robot 11 to move the welding torch 11a along the predicted welding path. In this way, when a predicted welding path is available, the robot control unit 16 may control the welding robot 11 using the predicted welding path, and when a predicted welding path is not available, the robot control unit 16 may control the welding robot 11 using the teaching data. When an instruction to correct the welding position is received, the robot control unit 16 may perform control to correct the welding position in response to the instruction. The control for correcting the welding position may be, for example, control for moving the aim position of the welding torch 11a in a direction indicated by the received instruction by an amount indicated by the instruction. Note that, for example, when an instruction to correct the welding position indicates a corrected welding position, the robot control unit 16 may control the welding robot 11 so that the aim position of the welding torch 11a becomes the welding position indicated by the received instruction. Furthermore, moving the welding torch 11a along the predicted welding path may mean, for example, that the tip of the welding torch 11a moves along the predicted welding path, i.e., that the welding position becomes on the predicted welding path. Alternatively, as described below, if the approximation line and the welding path are identical, it may mean that the tip of the welding torch 11a moves in the direction of the welding path (i.e., the approximation line). In this embodiment, the former case will be mainly described. Note that the tip of the welding torch 11a may be, for example, the aim position of the welding torch 11a.

[0028] The accumulation unit 17 accumulates the positions of multiple points on the trajectory of welding performed by the welding robot 11 in the memory unit 14. The welding trajectory may be, for example, a welded weld line. Furthermore, the points to be accumulated may be, for example, positions corrected in response to a received instruction to correct the welding position. In this case, each time an instruction to correct the welding position is received, the position of a point on the welding trajectory that is the corrected welding position may be accumulated in the memory unit 14. Furthermore, for example, the positions of points on the welding trajectory other than the corrected welding position may be accumulated in the memory unit 14. In this embodiment, a case where the position of a point on the welding trajectory that is the corrected welding position is accumulated each time a correction instruction is received will be mainly described. The position of a point on the welding trajectory may be, for example, a coordinate value in a predetermined coordinate system such as a robot coordinate system. The accumulation unit 17 may accumulate the positions of multiple points so that the order of accumulation is clear. As an example, the accumulation unit 17 may accumulate the position of a point in association with the time of accumulation.

[0029] The prediction unit 18 predicts the welding path using the positions of the multiple points accumulated by the accumulation unit 17. The prediction unit 18 may predict the welding path using, for example, the positions of a predetermined number of the most recent points among the positions of the multiple points accumulated. The prediction unit 18 may, for example, identify the position of the approximation line so as to optimize an objective function corresponding to the difference between the accumulated positions of the multiple points and a predetermined linear approximation line, and predict the welding path corresponding to the position of the identified approximation line. The approximation line may, for example, be a regression line in regression analysis. Furthermore, the welding path may, for example, be a path along which the tip of the welding torch 11a will move in the future. Therefore, the welding path may be considered to be, for example, a path corresponding to an unwelded welding line.

[0030] The predetermined linear shape may be, for example, a straight line or a curved line without an inflection point. A curved line without an inflection point means that there is no inflection point along the curve, and may be, for example, an arc or a parabolic shape. In this embodiment, the case where the approximation line of the predetermined linear shape is a straight line will be mainly described. The difference between each accumulated point of position and the approximation line may be, for example, the distance between each point and the approximation line. This distance may be, for example, the distance between the two in a predetermined direction. Specifically, the difference between the approximation line L10 in the y-axis direction in FIG. 3 described later and each point P １ ~P Ｎ The distance may be between .

[0031] The objective function corresponding to the difference between the accumulated multiple points and a predetermined linear approximation line may be, for example, a function whose value increases as the difference (error) between the multiple points and the approximation line increases, or a function whose value decreases as the difference increases. In the former case, optimization minimizes the objective function, while in the latter case, optimization maximizes the objective function. In either case, optimizing the objective function determines the position of the approximation line so as to minimize the difference between the multiple points and the approximation line. In other words, the position of the approximation line is identified so as to fit the multiple points. In this embodiment, the case of minimizing the objective function will be mainly described.

[0032] The prediction unit 18 may also optimize an objective function in which a weight is assigned to the difference between a plurality of points and the approximation line, the weight decreasing as the difference between the current time and the welding time increases. In this case, the objective function may multiply the difference between a certain point and the approximation line by a weight that decreases as the difference between the current time and the welding time corresponding to that point increases, i.e., the smaller the difference, the greater the weight. By doing so, when determining the position of the approximation line, points closer to the current time have a greater influence on determining the position of the approximation line. Therefore, the position of the approximation line is determined to match the shape of a more recently welded portion of the welding trajectory. Even if the shape of the welding trajectory gradually changes, the position of the approximation line can be determined according to the latest shape. Furthermore, even if the shape of the welded portion of the workpiece 5 changes as welding progresses, the position of the approximation line can be determined to match the shape of a portion close to the current welding position.

[0033] Here, as shown in FIG. 3, N accumulated points P １ ~P Ｎ Here, a method for specifying the position of the approximation line L10 using the position of the point P will be described. Note that N is an integer of 2 or more and may be determined in advance. １ In other words, in FIG. 3, the position of point P Ｎ is the position on the oldest weld line, and point P １ is the position on the newest weld line. Also, point P ｋ The coordinates of (x ｋ , y ｋ ) where k is an integer from 1 to N. For example, when the teaching line indicated by the teaching data is linear, the x-y Cartesian coordinate system may be set so that the x-axis is in the direction of the teaching line.

[0034] Furthermore, when the equation of the linear approximation line L10 is y=a+bx, the objective function Q may be expressed by the following equation: where a and b are real numbers.

[0035] In the above equation, ω ｋis the weight given by the following equation: ０ , L are positive real constants. Note that the weighting of new points can be adjusted by changing the value of L. By setting a larger value of L, newer points can be weighted more heavily. It goes without saying that these weights are just an example, and other weights may be used as long as they decrease the further away from the current point, i.e., the larger k is.

[0036] In the above-mentioned formula for the objective function Q, the distance between the point and the line in the y-axis direction is squared, and the position of the approximation line L10 is identified by the weighted least squares method, but this is not necessarily the case. For example, the distance between the point and the line in the y-axis direction may be raised to the fourth power, and the absolute value of this distance may be used. In addition, the weight ω ｋ is set so that the weight is larger for newer points, i.e., for smaller values ​​of k. Therefore, this weight ω ｋ is a weight that decreases as the difference between the current time and the welding time increases. If no weight is used, for example, ω ｋ = 1.

[0037] The prediction unit 18 can specify the values ​​of a and b by optimizing the objective function Q, and can specify the approximate line y = a + bx. Note that the optimization in this case is minimization. The prediction unit 18 may, for example, use this approximate line as the welding path. In this case, the latest point P １ It is preferable that the weight of is set to be large compared to the other weights, that is, L is set to a larger value.

[0038] Further, the prediction unit 18 may set, for example, a straight line extending from the current welding position in parallel to the approximation line as the welding path. １ , the prediction unit 18 calculates, for example, the point P １The welding path may be a straight line that passes through the line t and extends in the direction of the gradient b determined by optimization. This case will be mainly described in this embodiment.

[0039] Welding power source 19 supplies a high voltage used in welding to welding torch 11a and workpiece 5. Welding power source 19 may supply the high voltage, for example, in response to an instruction from robot control unit 16. Furthermore, when consumable electrode arc welding is performed by welding robot 11, welding power source 19 may also control the supply of welding wire.

[0040] Next, the operation of welding robot system 10 will be described with reference to the flowchart in Fig. 4. (Step S101) Reception unit 15 determines whether an instruction to correct the welding position has been received. If an instruction to correct the welding position has been received, the process proceeds to step S102; if not, the process proceeds to step S106.

[0041] (Step S102) Robot control unit 16 performs control to correct the welding position in accordance with the received instruction. In accordance with this control, the target position of welding torch 11a of welding robot 11 is changed.

[0042] (Step S103) The accumulation unit 17 accumulates the positions of the points on the welding trajectory in the memory unit 14. The positions of the points to be accumulated may be, for example, the corrected welding positions.

[0043] (Step S104) The prediction unit 18 determines whether to predict a welding path. If a welding path is to be predicted, the process proceeds to step S105. If not, the process returns to step S101. Note that the prediction unit 18 may determine to predict a welding path when, for example, the positions of a predetermined number of points or more are stored in the storage unit 14.

[0044] (Step S105) The prediction unit 18 predicts a welding path using the positions of the multiple points stored in the storage unit 14. The prediction unit 18 may, for example, identify the position of the approximation line by optimizing an objective function according to the difference between the multiple points and the approximation line, and predict the welding path using the identified position of the approximation line. Information indicating the predicted welding path may be stored in, for example, the storage unit 14. Then, the process returns to step S101.

[0045] (Step S106) The robot control unit 16 determines whether or not a predicted welding path exists. If a predicted welding path exists, the process proceeds to step S107; if not, the process proceeds to step S108.

[0046] (Step S107) Robot control unit 16 controls welding robot 11 so that welding torch 11a moves along the predicted welding path. Then, the process returns to step S101. By repeating this control, welding torch 11a moves along the predicted welding path.

[0047] (Step S108) Robot control unit 16 controls welding robot 11 in accordance with the teaching data stored in memory unit 14. Then, the process returns to step S101. By repeating this control, welding robot 11 operates in accordance with the teaching data. That is, welding torch 11a of welding robot 11 moves as instructed.

[0048] In steps S102, S107, and S108 in the flowchart of FIG. 4 , the robot control unit 16 may control the welding power source 19 so that welding is performed under predetermined welding conditions. The welding conditions may be included in, for example, teaching data. The processing order in the flowchart of FIG. 4 is merely an example, and the order of the steps may be changed as long as similar results are obtained. Although not included in the flowchart of FIG. 4 , an image of the welding area may be captured by the camera 12 and displayed on the display device 13. A worker correcting the welding position may input a correction instruction via an input device such as a teaching pendant or joystick while viewing the displayed image. In the flowchart of FIG. 4 , the processing ends when the power is turned off or an interrupt is issued to end the processing.

[0049] Next, the operation of the welding robot system 10 according to this embodiment will be described using a specific example. In this specific example, a teaching line L1 is set as shown in Fig. 5A, and the welding portion L2 of the workpiece 5 gradually shifts from the teaching line L1 in the distortion direction, which is to the left in the figure, due to thermal distortion of the workpiece 5.

[0050] First, the welding torch 11a is moved along the teaching line L1 in accordance with the teaching data, and welding is performed along the teaching line L1 (steps S101, S106, and S108). An image of the welding area is captured by the camera 12 and displayed on the display device 13. When the worker confirms from the image that the welding position has shifted from the welding portion L2 of the workpiece 5, he or she operates the input device to input an instruction to correct the welding position. This input is accepted by the reception unit 15 and passed to the robot control unit 16 (step S101). Upon receiving the instruction to correct the welding position, the robot control unit 16 controls the welding robot 11 in accordance with the instruction to change the welding position (step S102). Assume that this control changes the welding position to position Pa in FIG. 5A . In response to this change, the robot control unit 16 passes the coordinate values ​​of position Pa in the robot coordinate system to the accumulation unit 17. Upon receiving the coordinate values, the accumulation unit 17 accumulates the coordinate values ​​of position Pa in the memory unit 14 (step S103).

[0051] In this example, it is assumed that the welding path is predicted using the three most recent accumulated positions. However, since only one position has been accumulated at this point, the prediction unit 18 determines not to predict the welding path (step S104). Then, welding continues along the direction of travel indicated by the teaching data (steps S101, S106, and S108).

[0052] Furthermore, when an instruction to correct the welding position is received again from the operator (step S101), the welding position is changed to position Pb in Fig. 5A in response to the instruction (step S102), and the coordinate value of position Pb is stored in memory unit 14 (step S103). At this point, since only two positions have been stored, no welding path prediction is performed (step S104).

[0053] Thereafter, when an instruction to further correct the welding position is received from the operator (step S101), the welding position is changed to position Pc in FIG. 5A in response to the instruction (step S102), and the coordinate value of position Pc is stored in memory unit 14 (step S103). At this point, three positions are stored in memory unit 14, so prediction unit 18 determines to predict the welding path (step S104), identifies the position of a linear approximation line using the positions of three points Pa, Pb, and Pc, predicts the welding path that passes through point Pc and extends in the direction of the approximation line, and passes the predicted welding path to robot control unit 16 (step S105). The welding path passed to robot control unit 16 may be, for example, information indicating the welding path in the robot coordinate system.

[0054] Upon receiving the predicted welding path, the robot control unit 16 controls the welding robot 11 to move the welding torch 11a along the predicted welding path (steps S101, S106, and S107). As a result, as shown in FIG. 5A , welding is performed from point Pc in the direction of arrow A1. By moving the welding torch 11a along the predicted welding path in this manner, the deviation between the welded portion L2 of the workpiece 5 and the welding trajectory L3 is reduced. This reduces the number of instructions to correct the welding position, thereby reducing worker fatigue. Furthermore, as shown in FIG. 5A , even if there are periods without images, the possibility of deviations exceeding the allowable range occurring during those periods can be reduced, thereby improving welding quality.

[0055] Furthermore, when an instruction to further correct the welding position is received from the operator (step S101), the welding position is changed to position Pd in ​​FIG. 5A in accordance with the instruction (step S102), and the coordinate value of position Pd is stored in memory unit 14 (step S103). Then, prediction unit 18 identifies the position of a linear approximation line using the positions of three points Pb, Pc, and Pd, predicts a welding path that passes through point Pd and extends in the direction of the approximation line, and transmits the predicted welding path to robot control unit 16 (steps S104 and S105). Robot control unit 16 then controls welding robot 11 to perform welding along the predicted welding path (steps S101, S106, and S107). In this way, welding is performed from point Pd in ​​the direction of arrow A2 in FIG. 5A.

[0056] In this specific example, the welding path prediction is started after a predetermined number of points are accumulated, but this is not necessarily the case. For example, a dummy welding line may be assumed in the direction opposite to the teaching line L1 from the welding start point, and multiple points may be set on the dummy welding line in advance, so that the welding path prediction can be started when the first point is accumulated. In this way, it becomes possible to predict the welding path even when the coordinate values ​​of point Pa are accumulated, for example.

[0057] As described above, with the welding robot system 10 according to this embodiment, even if the unwelded portion of the workpiece 5 changes due to thermal distortion, the welding torch 11a can be moved along a predicted welding path in response to the change, thereby reducing the deviation between the unwelded portion of the workpiece 5 and the welding position of the welding torch 11a. As a result, the frequency of remote corrections performed by the operator can be reduced, thereby reducing operator fatigue. Therefore, for example, remote corrections can be performed by fewer operators. Furthermore, because the welding torch 11a can be moved along a predicted welding path in response to the changed welded portion, even if there is a predetermined period of video skip due to communication delays, the possibility that the deviation of the welding torch 11a will exceed the allowable range during that time can be reduced, thereby providing a welding robot system 10 that is more robust against communication delays. Furthermore, when thermal distortion occurs in the workpiece 5, the welded portion of the workpiece 5 often shifts in one direction, as shown in FIG. 5A , for example. Therefore, predicting the welding path using the welding trajectory in this manner is believed to enable highly accurate prediction of the welding path.

[0058] Furthermore, by identifying the position of the approximation line so as to optimize an objective function according to the difference between multiple points on the welding trajectory and the approximation line, and predicting the welding path using the approximation line, it is possible to predict a welding path that fits the multiple points. Furthermore, by optimizing an objective function in which a weight is assigned to the difference between the multiple points and the approximation line so that the weight decreases as the difference between the current time and the welding time increases, it is possible to increase the influence of newer points on the welding trajectory when predicting the welding path, thereby making it possible to predict a welding path that fits the current welding location, for example. Furthermore, even if the shape of the unwelded portion of the workpiece 5 gradually changes due to thermal distortion or the like, it becomes possible to predict a welding path that matches the latest situation.

[0059] Although the present embodiment has been described primarily with reference to a case where the welding position is corrected in a direction perpendicular to the teaching line L1, this is not necessarily the case. For example, when the welding torch 11a moves along the welding path L3 predicted by the prediction unit 18, the reception unit 15 may receive an instruction to correct the welding position in a direction perpendicular to the welding path L3, and the robot control unit 16 may correct the welding position in accordance with the instruction. Furthermore, when the welding torch 11a moves along the welding path L3 predicted by the prediction unit 18, the display device 13 may display an image in which the direction of the welding path L3 is oriented in a predetermined direction. The predetermined direction may be, for example, a predetermined direction (e.g., left / right) on the display screen of the display device 13. More specifically, the image may be displayed so that the direction of the welding path is oriented in the left / right direction (horizontal direction) on the display screen. Such a display may be performed, for example, by changing the orientation of the camera 12 or by image processing. When a linear welding path is displayed in the horizontal direction, the reception unit 15 may receive an instruction to correct the welding position in the vertical direction, which is perpendicular to the horizontal direction. By performing such a correction, the operator can more intuitively correct the welding position. Furthermore, even when the welding path is oriented in various directions, the operator can perform the same correction, thereby reducing the operator's workload. In this case, for example, as shown in FIG. 5B , when welding is being performed in the direction of arrow A1, the welding position is corrected in a direction perpendicular to that direction. When welding is being performed in the direction of arrow A2, the welding position is corrected in a direction perpendicular to that direction. The direction of the welding path may be, for example, the direction of the linear welding path if the welding path is linear. Alternatively, when the welding path is curved, the direction may be the tangent direction of the welding path at the current welding position. Furthermore, the direction perpendicular to the welding path may be, for example, a direction perpendicular to the linear welding path if the welding path is linear, or a direction perpendicular to the tangent to the welding path at the current welding position if the welding path is curved.

[0060] Furthermore, in the above embodiments, each process or function may be realized by centralized processing by a single device or a single system, or may be realized by distributed processing by multiple devices or multiple systems.

[0061] Furthermore, the above-described embodiments are merely examples for specifically implementing the present invention, and are not intended to limit the technical scope of the present invention. The technical scope of the present invention is defined by the claims, not by the description of the embodiments, and is intended to include modifications within the literal scope of the claims and within the scope of equivalent meanings.

Claims

1. A welding robot system comprising: a welding robot which moves a welding torch; a camera attached to said welding robot which photographs unwelded portions of a workpiece; a display device which displays the image photographed by said camera; a reception unit which receives instructions to correct a welding position from an operator in response to the image displayed by said display device; a robot control unit which controls said welding robot to perform welding based on teaching data and controls to correct the welding position in response to the instructions received by said reception unit; an accumulation unit which accumulates the positions of a plurality of points on a trajectory of welding by said welding robot; and a prediction unit which predicts a welding path using the positions of the plurality of points accumulated by said accumulation unit, wherein said robot control unit controls said welding robot so that the welding torch moves along the welding path predicted by said prediction unit.

2. A welding robot system as described in claim 1, wherein the prediction unit identifies the position of an approximation line so as to optimize an objective function according to the difference between the multiple points and an approximation line of a predetermined linear shape, and predicts the welding path according to the position of the identified approximation line.

3. A welding robot system according to claim 2, wherein the prediction unit optimizes an objective function in which a weight is assigned to the difference between the multiple points and the approximation line, the weight being smaller the greater the difference between the current time and the welding time.

4. The welding robot system according to claim 2 or 3, wherein the predetermined linear shape is a straight line.

5. A welding robot system as described in any one of claims 1 to 3, wherein the display device displays an image in which, when the welding torch moves along the welding path predicted by the prediction unit, the direction of the welding path is in a predetermined direction.

6. A welding robot system as described in claim 5, wherein, when the welding torch moves along the welding path predicted by the prediction unit, the reception unit receives an instruction for correction in a direction perpendicular to the welding path.

Images