JPH0592345A - Control device for polishing tool - Google Patents

Abstract

PURPOSE:To simply execute the control of a polishing tool including its working direction on a photographing screen by analizing the photographed picture image of a workpiece through a three dimensional picture image processing device, and polishing the fixed portion of the workpiece by means of a polishing tool on the analized positional data. CONSTITUTION:The three dimensional data of the shape of a workpiece are previously stored, and the surface of the workpiece is divided and set into picture image frames corresponding to one photographing screen. A control device for a polishing tool comprises a model memory means 2 to set up the direction of the normal on a workpiece surface in each of the picture image frames and to store it, a photographing means 3 to photograph the workpiece through the picture image frame coinciding with each picture image frame set for a carried workpiece B, an analizing means 4 to analize the picture image of the workpiece for obtaining the positional data of a polished portion, and a posture control means 5 to control the posture of a polishing tool 50 relative to the surface of the workpiece based on the positional data of the polished portion and the stored normal data of the polished portion.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a polishing tool control device, and more particularly to a polishing tool control device for polishing and removing defective portions of a coating surface of a work such as an automobile body.

[0002]

2. Description of the Related Art Conventionally, for the purpose of improving the leveling of the painted surface and the adhesion of the overcoat paint, the entire painted surface of a work such as an undercoated or middle-coated automobile body is polished before overcoating. In order to move the polishing tool along the curved surface of the workpiece when polishing is performed by a robot equipped with a hand or an arm tip,
A technique in which an actual work is carried in and the above-mentioned polishing tool is taught along the work coating surface and operated based on this teaching is disclosed in, for example, JP-A-60-2084.
It is known as seen in Japanese Patent Publication No. 9.

Further, in order to inspect the coated surface of the above work and polish and remove the defective coating portion with a polishing tool, as a technique for inspecting this coated surface, the coated surface is irradiated with light and the reflected light is reflected on the screen. A technique has been proposed in which a surface defect is projected and the surface defect is automatically detected from the sharpness of the projected image.

On the other hand, in Japanese Patent Laid-Open No. 58-64517, in order to correct a coating defect that has occurred on the coated surface of the work by polishing, the worker visually inspects the conveyed work and finds out the detected defect. A technique has been proposed in which, by inputting a part and a defective state, respectively, with an indicating device, a water-polishing device including a robot or the like automatically polishes on the downstream side.

[0005]

However, in the case of polishing a predetermined portion of a work with the above-mentioned polishing tool, the operation pattern of the polishing tool is stored by teaching, or a predetermined portion is instructed by position data. In the control of polishing, there is a problem that accurate and high-quality polishing cannot be performed, and the teaching process becomes complicated.

That is, when polishing a work with a polishing tool, a good polishing surface cannot be obtained unless work is performed in a direction perpendicular to the surface of the work. In addition to the complicated work of memorizing the data, it is necessary to call the working direction of the relevant portion from a large number of teachings and to perform the polishing based on the working direction.

Further, when performing based on the position data,
The position in the planar image picked up by the image pick-up means does not match the position of the surface of the three-dimensional work, and there are various deviations,
Moreover, it is difficult to set the operation of the polishing tool from the direction normal to the surface of the work.

In view of the above-mentioned circumstances, an object of the present invention is to provide a polishing tool control device capable of easily performing control including the operating direction of the polishing tool on a photographing screen.

[0009]

In order to achieve the above-mentioned object, a polishing tool control apparatus of the present invention has a basic structure shown in FIG. The photographed image of the work B is analyzed by the three-dimensional image processing apparatus 1, and the polishing tool 50 is analyzed based on the position data.
The three-dimensional data of the surface shape of the work B is stored in advance, and the surface of the work B is divided into a plurality of image frames having a size corresponding to one photographing screen. The model storage means 2 is divided and set, and the normal direction of the work surface is set and stored in each image frame portion, and the model storage means 2 is set in the model storage means 2 for the work B that is carried in and is actually polished. An image pickup means 3 for picking up a surface image of the work B in an image frame corresponding to each image frame, and an analysis means 4 for analyzing the image of the work B picked up by the image pickup means 3 to obtain position data of a polishing portion for polishing. An attitude control means 5 for controlling the attitude of the polishing tool 50 with respect to the work surface based on the position data of the polishing part by the analyzing means 4 and the normal line data of the polishing part stored in the model storage means 2. Provided by those configured.

Further, the analyzing means 4 has a model storage means 2 according to a photographing angle of the polishing portion by the image pickup means 3.
It is preferable to perform conversion correction on the position data of the plane parallel to the coordinate axes in.

[0011]

In the polishing tool control apparatus as described above, three-dimensional data of the surface shape of the work is stored in advance in the model storage means of the three-dimensional image processing apparatus, and the surface of the work in this three-dimensional data is photographed once. Divide and set into multiple image frames of a size corresponding to the screen, teach the shooting order, and set and store the normal direction perpendicular to the surface of the three-dimensional data in each image frame portion. For the work to be polished on the actual line, the surface image of the work is sequentially photographed by the image pickup means in correspondence with the image frame of the model storage means, and the work image photographed by the image pickup means is taken. In the image, the coating state of the surface is analyzed by the analyzing means, the position data of the polishing portion to be polished is obtained by the polishing tool, and the position data of the polishing portion is determined by the analyzing means. The normal data of the polishing portion is input from the storage means to the posture control means, and the posture control means causes the polishing tool to move the polishing tool from the proper polishing direction based on the normal direction to the polishing portion. Position control is performed so as to act on the surface of the work, and a predetermined polishing process is performed. It is possible to easily perform the processing based on the setting of the normal direction based on the dimensional data, and to obtain accurate and good polishing accuracy.

Further, when the work is photographed by the image pickup means, the position data of the polished portion on the plane of the photographed image and the three-dimensional data in the model storage means are taken from the relationship between the image pickup direction and the work surface shape. There may be a deviation from the position data of the corresponding polishing area, but if the conversion data is corrected to the position data of the coordinate axis in the model storage means and the parallel surface according to the imaging angle and the work shape, more accurate position data will be obtained. Therefore, the polishing accuracy can be improved.

[0013]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 2 shows a schematic plan configuration of a water polishing processing line of an automobile manufacturing plant equipped with a polishing tool control device according to one embodiment. The front, rear, left, and right directions of the vehicle body as the work B will be described as a reference.

The water-polishing processing line L for automobiles is provided on the downstream side of the intermediate coating line (not shown). From the upstream side, the imaging station L1, the polishing station L2, and the repair station.
L3 and flush station L4 are provided.

First, the imaging station L1 is provided with a pair of front and rear imaging robots R1 and R2 for photographing the surface of the work B, and the work B coated with the intermediate coating is conveyed to the imaging station L1 and, for example, the front side. The imaging robot R1 images the front half of the work B, and the imaging robot R2 behind the image captures the second half of the work B. An image captured by a camera 43 of the imaging robots R1 and R2, which will be described later, is input to the three-dimensional image processing apparatus 1 and a defective portion of coating is detected. Although the detailed structure is not shown, the three-dimensional image processing apparatus 1 is composed of a main body portion for performing arithmetic processing and storage, a CRT display for displaying an image, a keyboard for input operation, a tablet, and the like.

Imaging robots R1 and R are installed in the polishing station L2.
A polishing robot R3 of the same type as that of 2 is provided, and the polishing robot R3 polishes a coating defect portion, that is, a polishing portion, based on position data based on the analysis result of the captured image from the imaging station L1. Further, the repair station L3 is provided in order to polish the remaining polishing site by the operator when the polishing robot R3 cannot polish all the coating defect parts within a predetermined line tact due to the large number of polishing sites. ing. Further, the washing station L4 is provided with a plurality of washing showers 27 and washing brushes 28, 29 for washing the work B after polishing.

Next, an example of mechanical structures of the imaging robots R1 and R2 and the polishing robot R3 will be described. Since each robot is basically the same type of robot and has the same structure, the front imaging robot R1 will be described with reference to FIG. The same members of each robot are designated by the same reference numerals.

Columns 10 are erected at four corners in the front, rear, left and right of the imaging station L1, and beams 11 extending in the front and rear direction are fixed to the upper ends of the two columns 10 on the left and right, respectively. A guide rail 12 is provided at the upper end in the front-rear direction, and a movable frame 13 extending in the left-right direction across the beams 11 on both sides is erected, and the engaging portions 13a with bearings at both ends of the movable frame 13 have the above-mentioned guides. The rail 12 is engaged with the rail 12 and is movable in the front-rear direction. A ball screw shaft 14 that is rotationally driven by a servo motor (not shown) is provided at the upper end of the right beam 11 so as to extend in the front-rear direction, and the right end portion of the movable frame 13 is screwed onto the ball screw shaft 14. A nut is provided, and the movable frame 13 is moved in the front-rear direction via the nut by rotationally driving the shaft 14 with a servomotor.

On the other hand, the movable frame 13 has a left and right ball screw shaft 16 which is driven to rotate by a servomotor 15.
And a guide rail 17 provided above and below the movable rail 18 having an engaging portion (not shown) slidably engaged with the guide rail 17 and a nut screwed with the ball screw shaft 16 Servo motor 15
By rotating the shaft 16 by means of, the movable table 18 is moved in the left-right direction via the nut.

Further, the movable table 18 is provided with a ball screw nut 19 and a pair of rod guides 18a which are rotatably driven by a vertically oriented servo motor, and a ball screw shaft 20 and a rod guide 18a which are screwed through the ball screw nut 19 are provided. At the lower end of the guide rod 21 inserted in the
Is fixed, and the hand holding member 22 is moved up and down by rotating the ball screw nut 19. The hand holding member 22 includes a drive connecting portion 23 for rotating around a vertical axis by a servo motor, a drive connecting portion 24 for rotating around a horizontal axis by a servo motor, and a hand connecting portion around a hand axis by a servo motor. Drive connection 25 for rotational drive
The hand 26 is attached via and. Further, an image pickup CCD camera 43 and an illumination projector 42 are attached to the tip of the hand 26 by a support member 41.

In the case of the polishing robot R3, the polishing tool 50 as shown in FIG.
Is attached. The polishing tool 50 is a grindstone at the tip.
51 and an actuator for laterally driving the whetstone 51 in an oscillating manner
52 and.

The imaging robots R1 and R configured as described above
2 and the polishing robot R3 have six degrees of freedom: x-axis (front-back direction), y-axis (horizontal direction), z-axis (vertical direction), vertical axis, horizontal axis, and hand axis.

Next, a control device for the imaging robots R1 and R2 and the polishing robot R3 will be described. Imaging station
A robot control device 32 for controlling the positions of the imaging robots R1, R2 is provided in L1, and a three-dimensional image processing device 1 for processing an image taken by the CCD camera 43 of the imaging station L1 is installed. In addition, the polishing station L2 controls the position of the polishing robot R3 and
A polishing control device 34 for controlling 50 polishing conditions is provided, and the position data from the three-dimensional image processing device 1 is input to the polishing control device 34.

Here, in order to detect a defective portion on the painted surface of the work B carried into the water-cutting processing line L to obtain a polished portion, when the surface of the work is sequentially photographed by the CCD camera 43 and an image is captured, The robot CAD simulation is used to teach the imaging robots R1 and R2. In this teaching, the image frame and the shooting direction vector (indicating the distance and direction between the camera origin and the point on the work B) are displayed on the CRT display. Shooting points are set sequentially while outputting. And before doing this teaching,
The image processing apparatus 1 stores three-dimensional data of the surface shape of the work B in advance, and an image width (capture frame) obtained by one shooting when teaching the imaging robots R1 and R2.
Is set to a range in which an effective image is obtained in a range smaller than the size, that is, an image frame is set, and the entire work model image based on the three-dimensional data is partitioned so as to be sequentially arranged in the image frame. Teaching is performed by setting a vector. In addition, a normal direction perpendicular to the surface of the model image at a plurality of portions of each image frame, for example, a plurality of points on the frame line, is obtained and stored.

Then, in response to the actual work B after the intermediate coating being applied to the automatic water research line L, the imaging robots R1 and R2 of the imaging station L1 are operated based on the teaching.
Is controlled so that the same image frame as the image frame set on the model image is photographed, the position of the image frame, the distance set by the photographing direction vector, and the photographing direction are sequentially set on the work B. The image is taken and the image is captured. Then, a coating defect portion is detected from this input image by visual observation or sharpness, and this portion is designated as a polishing portion to be polished by the polishing tool 50 of the polishing station L2, and data for performing this polishing is output. To do. The command data for the polishing robot R3 of the polishing station L2 is the three-dimensional position data of the polishing site and the normal direction data of that region, and based on this, the predetermined polishing site for the work B moved to the polishing station L2. While moving to, the grindstone 51 of the polishing tool 50 is pressed against the surface from the normal direction to polish.

Here, the processing in the simulation will be described with reference to FIGS. FIG. 5 shows an example of creating the three-dimensional data by the surface model BM when storing the three-dimensional data of the work shape. The surface model BM used here is a three-dimensional patch shape and the work B is A grid-like intersection a is set as a coordinate point separately for each planar portion, and the x, y, and z values of this coordinate point are input respectively to store the surface data. Is for obtaining approximate coordinates by a three-dimensional surface interpolation formula. In a work B such as an actual car body, it is difficult for the surface model BM to represent the entire complex shape due to its nature. Therefore, the shape to be expressed is divided and surface data in groups is used. Create and express the model shape and memorize it.

The work surface model BM created as described above is CR in any of the xy, yz, and zx planes depending on the teaching points of the imaging robots R1 and R2.
Display on the T display. For example, as shown in FIG. 6, in the case of teaching the side surface of the work, the surface model BM of the zx plane is displayed. In this surface model BM, data is set based on the coordinate origin O. Further, in the case where the teaching data of the work B and the surface data are set for each group, the target data group is selected.

Next, the captured image frame F (frame) of the camera 43 is displayed on the CRT display for the surface model BM of the work B displayed on the CRT display. This image frame F is, as described above, the CCD camera 43.
Thus, the effective image range is set within the image input range, which corresponds to the range of one captured image. The surface model of the work B corresponding to the image frame F
The BM is divided and teaching points are automatically generated.

The above automatic generation of teaching points
First, at the start, the image frame F is brought to an appropriate position, for example, the position closest to the coordinate origin O, and only the first teaching is set. However, at this point, only the x and z components are set, and the y component is automatically set later. Further, the scanning direction for automatically generating the teaching point is determined, and in the example of FIG. 6, x and z are scanned in the + direction. The y component of the teaching point is set to the + direction or − direction (− direction in the example of FIG. 6) with respect to the work B. Further, the ranges in the x direction and the z direction in which the teaching is performed according to the size of the surface model BM are set by respective coordinate values so as to surround the outside of the model BM. Next, the image frame F of the camera 43 is set in the x direction and the z direction. The image frames F are set sequentially, automatically or manually so that the frame lines of the adjacent image frames F overlap. To do. Finally, for the image frame F set in the entire operation range, it is not necessary to teach the image frame F that does not overlap the surface model BM, and the image frame F to be taught is selected. However, this selection may be performed at the same time when the image frames F are sequentially set in the case of manual operation, and when the image frames F are automatically set, the projected image frame F may be picked, This is performed by a method such as assigning a number to the image frame F to select it.

By setting the image frame F, the x and z values on the work B where the center point T of the frame F (see FIG. 7) is located are the x and z points where the CCD camera 43 is located, A predetermined distance (eg 30
cm) points x, y, z where the camera 43 is actually located
It becomes a point, and this point becomes a teaching point, and the center point T of the image frame F is set (a shooting direction vector indicating a shooting direction and a distance).

Subsequently, the teaching point and the normal vector of the surface of the work B from which the image is to be captured are automatically set. This setting is obtained by performing the following processing for each image frame F.

First, as shown in FIG. 7, the points b at a constant distance, for example, 1 cm, are passed along the frame of the image frame F,
An intersection point d that contacts the three-dimensional surface model BM of the work B in the negative direction is obtained by the line segment c parallel to the y axis. In addition, when a fraction occurs at the end at the set interval, the data at the four corners of the frame is always obtained. In addition, when the surface model BM is set in group units and the surface model BM of the part to be obtained overlaps, the intersection d is calculated for the data of all the groups, and the average value of that point is calculated later. To do. On the other hand, in the part where the frame is out of the surface model BM, the intersection point d with the surface model BM does not exist,
I don't count.

For the line segment c for which the intersection point d is obtained as described above, the average value of y coordinates on the surface model BM is obtained and the reference plane G is set as shown in FIG. Further, the normal vector V at the intersection point d between the line segment c and the surface model BM is obtained.

The normal vector V is a unit vector,
Although various methods can be considered for obtaining the normal vector V, for example, as shown in FIG. 9, it is obtained by using the patch-like surface model BM. That is, the intersection point d corresponds to the patch-shaped surface model BM of FIG.
The patches a at the four corners in the vicinity of the portion where is located are defined as one plane p, and the direction perpendicular to the plane p is determined and set as the normal vector V. The intersection point d on the surface model BM is obtained by surface interpolation. The data of the normal vector V is stored according to a predetermined file format.

The portion for obtaining the normal vector V does not have to be a portion above the frame of the image frame F. For example, the image frame F
It may be obtained at a predetermined position such as the image center T or the n-divided center.

The order of teaching is set for the image frame F set as described above. The setting is performed by, for example, picking an image displayed on the CRT display. Be seen. Also,
The teaching program is generally created for each module. For example, in the example of FIG. 6, teaching is performed so that the front fender portion (8 image frames) is set to the first module and shooting is performed in order. The middle door part (8 image frames) is set to the second module, and teaching is performed so that images are taken in order, and the rear fender part (6
Teaching may be performed such that the image frame) is set in the third module and the images are sequentially shot.

FIG. 10 shows a correction for improving the accuracy of the position data of the image taken by the camera 43. As a premise of this correction, the offset on the image frame F and the work surface model BM is shown. Ask for. This offset value t is calculated from the average y-coordinate value (reference plane G) at the intersection point d with the line segment c, and the yz value of the surface data
The distance to the intersection point d on the cross section is obtained and recorded. Then, when the work surface BM for measuring the position data of the polishing portion to be polished is offset from the reference plane G with respect to the camera 43, the position on the photographing screen is displaced from the accurate position. Make a correction.

That is, when the camera 43 captures an image, since it is performed in a plane on the reference plane G with a scale factor of SF ₀ , the work surface has a curved shape and is displaced from the reference plane G. The position data (y ₀ , z ₀ ) photographed by the camera 43 does not match the actual position (y, z) when the position is offset by t, and the scale factor SF ( The mm / pixel) similarity conversion correction formula is as follows, where L is the distance between the camera 43 and the reference plane G. SF = SF ₀ · L / (L + t) By this conversion, the accurate position of the polishing portion is obtained, and by sending this position data to the control device of the polishing station L2 and moving the polishing tool 50, accurate polishing can be performed. It is a thing. In the conversion correction, scale conversion is performed for each area in accordance with the normal vector V obtained for each area obtained by subdividing each section. Similar correction is performed in the x direction.

Then, the normal vector V and scale conversion obtained as described above are used for controlling the attitude of the polishing tool 50 as described above. That is, an image of the work B taken by the camera 43 is taken in along the taught image frame F, the coating defect portion is detected by analyzing the image, and the polishing portion is located at the position (x ₀ , y _{0) in the image.} ，
z ₀ ), the posture of the polishing tool 50 is controlled based on the normal vector V in each region and the position data (x, y, z) corrected by scale conversion, and the polishing tool 50 is set at the predetermined polishing position. The polishing is performed by pressing in the polishing direction.

In the above embodiment, the imaging robots R1 and R2 and the polishing robot R3 have a structure in which they are moved by being installed in the column 10 installed in the station, but they are indirect from the main body. The camera 43 or the polishing tool 50 may be installed at the tip of a robot having an extended arm, and other types of robots can be appropriately adopted.

Further, the camera is used to detect the coating defects.
In addition to the detection by the image captured by 43, the inspector visually inspects the surface of the work B to put a predetermined defect mark on the coating defect portion, and the defect mark is detected by the camera 43.
Alternatively, the position data may be detected by the image captured in step S3 and sent to the polishing station L2 side.

Further, a camera 43 for photographing the surface image of the work
The shooting position, that is, the shooting direction, is set from the direction parallel to the coordinate axes as in the above-described embodiment, and the teaching point is set so that the shooting is performed from the direction perpendicular to the work surface in the image frame F as much as possible. You can

[Brief description of drawings]

FIG. 1 is a block diagram showing a basic configuration of a polishing tool control device of the present invention.

FIG. 2 is a schematic plan view of a water polishing processing line equipped with a polishing tool control device according to an embodiment of the present invention.

3 is a sectional front view of the imaging station of FIG.

FIG. 4 is a perspective view of a polishing tool.

FIG. 5 is an explanatory diagram showing a setting example of a three-dimensional surface model of a work shape.

FIG. 6 is an explanatory diagram showing a display example of a surface model and an image frame teaching setting example.

7 and 8 are explanatory diagrams showing an example of setting a normal vector.

FIG. 9 is an explanatory diagram showing an example of calculating a normal vector.

FIG. 10 is an explanatory diagram showing correction of position data.

[Explanation of symbols]

 1 Image Processing Device B Work 2 Model Storage Means 3 Imaging Means 4 Analyzing Means 5 Attitude Control Means R1, R2 Imaging Robot R3 Polishing Robot 43 Camera 50 Polishing Tool

Claims (2)

[Claims] 1. A polishing tool control device that analyzes a photographed image of a work by a three-dimensional image processing device, and polishes a predetermined portion of the work by the polishing tool based on the position data, and a surface shape of the work in advance. 3D data is stored, and the surface of the work is divided and set into a plurality of image frames of a size corresponding to one shooting screen, and the normal direction of the work surface is set and stored in each image frame portion. Model storage means, image pickup means for picking up a surface image of the work in an image frame corresponding to each image frame set in the model storage means, and image of the work taken by the image pickup means. Analyzing means for obtaining position data of a polishing portion for polishing and polishing, position data of the polishing portion by the analyzing means and normal line data of the polishing portion stored in the model storing means. A polishing tool control device, comprising: a posture control means for controlling a posture of the polishing tool with respect to a work surface. 2. The polishing tool according to claim 1, wherein the analysis unit performs conversion correction on the position data of the plane parallel to the coordinate axis in the model storage unit according to the imaging angle of the polishing site by the imaging unit. Control device.