JPH09222913A - Teaching position correcting device for robot - Google Patents

Abstract

PROBLEM TO BE SOLVED: To precisely correct a teaching position without generating interference of a work, limiting the attitude that the robot should take, or causing deterioration in the quality of welding, etc., by mounting a visual sensor which photographs the teaching position and providing means for measurement, operation, etc., and correcting the teaching position. SOLUTION: The visual sensor 3 is fitted in such position relation that its image pickup surface is perpendicular to the length of a tool 2, and picks up an image of the teaching position to which the tool tip 2a of the robot 1 should to move. Then, this device is equipped with a means which corrects one teaching position along the length of the tool 2 by a quantity corresponding to the error in reference distance between the visual sensor 3 and a work 4 when the tool tip 2a is set at the one teaching position. Further, the device is equipped with a means which corrects one teaching position at right angles to the tool 2 on the basis of the relative position relation between a sensor coordinate system and a tool coordinate system by a quantity corresponding to a position shift on the tool coordinate system corresponding to a position shift on the sensor coordinate system.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for correcting a teaching position of a tool tip of a robot.

[0002]

2. Description of the Related Art In an automatic processing apparatus for automatically performing work using a working robot such as a cutting robot or a welding robot (hereinafter, simply referred to as a robot), when teaching an operation program of the robot, the teaching program is used for teaching. Jigs and CAD
Off-line teaching using links etc. is generally performed. According to the off-line teaching, the teaching work can be simplified and the time can be shortened as compared with the case of actually teaching by moving the robot.

However, since there is an error between the position taught offline and the position actually moved by the robot due to the influence of the parameter difference of the robot or the deflection due to its own weight, an accurate position is taught. You cannot do it.

Therefore, it is necessary to correct the above error after teaching.

As a method of correcting such an error in the teaching position, a three-dimensional visual sensor composed of two cameras is installed in a robot as taught in Japanese Unexamined Patent Publication No. 7-84631, and the teaching is performed offline. Move the robot to the desired position, recognize the image of the correction mark that was made in advance, or the feature point that the work originally has, measure the correction amount, and correct the teaching position by this correction amount There is something to fix.

[0006]

In the above-mentioned conventional teaching position correcting apparatus, it is necessary to install two cameras.

Therefore, not only the calibration such as optical axis alignment becomes complicated, but also the error factor increases, and as a result, the error contained in the measurement value of the correction amount becomes large.

Further, since the three-dimensional visual sensor is composed of two cameras and they are installed in the tool part, the structure of the tool part becomes large, and the robot and the work are not interfered with each other severely. In this case, there arises a problem that the camera interferes with the work.

Further, when trying to recognize a teaching point along a three-dimensional shape such as a teaching point on a welding line, the correction mark may not be visible due to the shadow of the work depending on the line of sight of the camera. . Therefore, to avoid this, 2
It is necessary to change the posture of the robot so that the cameras can simultaneously capture the marks, but this limits the postures that the robot can take. Further, if the changed robot posture is an unreasonable posture for the robot, vibration may occur during the operation of the robot, resulting in deterioration of quality such as welding.

The present invention has been made in view of the above circumstances, and it is possible to achieve accuracy without interfering with a work, limiting a posture that a robot can take, and deteriorating quality such as welding. It is an object of the present invention to provide an apparatus capable of well correcting a teaching position.

[0011]

In order to solve this problem, in the present invention, the tip of the tool attached to the arm of the robot is taught the position on the work to be moved so that the tip of the tool is In a robot configured to drive the robot so as to be positioned at the teaching position, a visual sensor attached to the tool with a predetermined positional relationship, for imaging a teaching position where a tool tip of the robot should move, A tool coordinate system that moves with the tool and displays the coordinate position on a plane perpendicular to the longitudinal direction of the tool is set, and a sensor coordinate system that displays the coordinate position within the imaging visual field of the visual sensor is set. The coordinate system setting means and the visual sensor are moved by a predetermined distance, and an image of one teaching position is picked up by the visual sensor at each moving position. From this imaging result, first measuring means for measuring the coordinate position of the one teaching position on the sensor coordinate system for each moving position, and the one teaching position measured by the first measuring means. Based on each coordinate position and a predetermined distance moved by the visual sensor, the distance between the visual sensor and the work is calculated, and the calculated distance and the tool tip when the tool tip is positioned at the one teaching position. A first calculation unit that calculates an error between a reference distance between the visual sensor and the work, and the one teaching position is imaged by the visual sensor, and from the result of this imaging, the one of the one on the sensor coordinate system is imaged. Second measuring means for measuring the coordinate position of the teaching position, the coordinate position of the one teaching position measured by the second measuring means, and the tool tip at the one teaching position. Based on the positional shift between the one teaching position and the reference coordinate position on the sensor coordinate system and the relative positional relationship between the sensor coordinate system and the tool coordinate system, the positional shift on the sensor coordinate system is dealt with. Second computing means for computing a positional deviation on the tool coordinate system, and the one teaching position are corrected in the longitudinal direction of the tool by an amount corresponding to the error computed by the first computing means. And a correction means and a tool for correcting the taught position so that the one taught position is corrected in the direction perpendicular to the tool by an amount corresponding to the positional deviation calculated by the second calculating means. I am trying to get it.

That is, according to such a configuration, it is not necessary to dispose a three-dimensional visual sensor composed of two cameras in the tool section as in the prior art, and basically,
Only one visual sensor needs to be provided in the tool section.

As a result, the structure of the tool portion becomes small, and the error factors are reduced, so that interference with the work occurs, the posture that the robot can take is limited, and the quality of welding or the like deteriorates. The teaching position can be accurately corrected without doing so.

[0014]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of a teaching position correcting apparatus for a robot according to the present invention will be described below with reference to the drawings.

In this embodiment, a cutting robot for cutting work is assumed.

That is, FIG. 1 is a perspective view showing a multi-joint cutting robot 1. As shown in FIG. 1, the tip of an arm 1a of the robot 1 is visually connected to a tool (torch) 2 via a bracket 21. The sensor 3 is attached.

Here, the visual sensor 3 is attached to the tool 2 in such a positional relationship that its imaging surface is perpendicular to the longitudinal direction of the tool 2 (see FIG. 2), and the tool of the robot 1 is used. An image is taken of the teaching position to which the tip 2a should move.

In this embodiment, the visual sensor 3 is arranged in a non-interference area (dead space) between the tip of the arm 1a and the tool 2 to avoid interference with the work 4. There is. The positional relationship between the visual sensor 3 and the tool 2 is not limited to that of this embodiment, and is arbitrary.

On the other hand, a work to be worked or a master work (hereinafter, simply referred to as a work) 4 is fixed in a working space by a fixing means (not shown), and a cutting work line 41 is written on the work 4 by a stroke. Has been done. Each teaching point 42 is inscribed so as to intersect with the cutting work line 41.

In this embodiment, the teaching point 42 is displayed at the intersection of the writing lines, but it can be displayed in any form as long as it can be recognized by the visual sensor 3. For example, it may be a seal or a ridgeline of the groove portion if it is a welding work.

FIG. 2 shows the system configuration of this embodiment.

The robot controller 11 shown in FIG.
Is to drive and control each axis of the robot 1 according to the operation program to control the operation path of the robot 1 and the attitude of the tool 2, and send a start command or an end command to a cutting machine control device / power supply (not shown). Then, a series of cutting work is performed.

On the other hand, the sensor controller 31 inputs a video signal picked up by the visual sensor 3, and a teaching point 42 in the picked-up visual field by an image processing program described later.
Is to be detected. Furthermore, the sensor controller 3
1 transmits / receives the detected position data and operation program to / from the robot controller 11 and edits / corrects the robot operation program. The image processing program is executed in parallel with the robot operation while taking timing with parallel input / output as described later. Next,
The operation program correction processing (teaching position correction processing) executed by the sensor controller 31 and the robot controller 11 will be described with reference to the flowchart shown in FIG.

As shown in FIG. 4, a sensor coordinate system (origin Oo) as an Xo-Yo coordinate plane is set in the imaging visual field 3a of the visual sensor 3, and a teaching point 42 on this sensor coordinate system is set. The coordinate position of is detected and displayed. Further, as will be described later, the tool 2 displays the coordinate position on the XRT-YRT coordinate plane (origin ORT) that is a coordinate system that moves together with the tool 2 and is perpendicular to the longitudinal direction of the tool 2. The tool coordinate system to be set is set.

Therefore, it is first determined whether or not the relative positional relationship (see FIG. 4) of the tool coordinate system with respect to the sensor coordinate system is set within the imaging field of view 3a of the visual sensor 3 (step 101).

Here, if the tool coordinate system is not set, the following tool coordinate system setting program (steps 102-1 to 102-6) is executed in the sensor controller 31 to set the tool coordinate system. Processing is performed (step 102). The tool coordinate system is
It may be set again only when the relative positional relationship between the tool 2 of the robot 1 and the visual sensor 3 is changed due to replacement of the visual sensor 3 or when the magnification of the lens of the visual sensor 3 is changed.

The tool coordinate system setting program is stored in the sensor controller 31, and by executing this program, the process of setting the tool coordinate system by processing the image of the visual sensor 3 while operating the robot 1 is automatically performed. Is done in a regular manner.

First, before the above program is executed, the distance between the tool tip 2a and the work 4 becomes the distance between the tool and the work during cutting by manually operating the robot 1 or the like, and the posture of the tool 2 is changed. The tip 2a of the tool 2 is made to accurately coincide with a reference teaching point (hereinafter referred to as a reference point) 42 so that the tool has a posture during cutting. Note that the tip 2a of the tool 2 may be aligned with a reference mark provided separately instead of the teaching point 42 itself.

The program is executed in this state, and an image of the reference point 42 is captured. The visual sensor 3
It is assumed that the captured image has a field of view sufficient to show the reference point 42 (or mark).

In the following, referring to FIG. 4, the captured image taken in is processed, and as a result, the reference point 4 is obtained.
The position P0 of 2 is measured as the coordinate position P0 (Xo p0, Yo p0) in the sensor coordinate system Xo-Oo-Yo. Here, the reference point P0 corresponds to the origin ORT of the tool coordinate system XRT-ORT-YRT. That is, the tool tip 2a is the reference point 4
When accurately positioned at 2, the reference point P0 coincides with the origin ORT of the tool coordinate system on the captured image, and the origin ORT is the coordinate position P0 (Xo p0,
(Yop0) (step 102-1).

Next, the tool 2 is moved by a known predetermined amount L in the XRT axis direction of the tool coordinate system, and the tool tip 2a is moved to the point P1 (step 102-2).

At this time, in the imaging visual field 3a of the visual sensor 3, the reference point P0 appears to move to the point P1 '. Therefore,
The image at this time is captured, and the coordinate position P1 '(Xo p1', Yo p of the point P1 'in the sensor coordinate system Xo-Oo-Yo
1 ') is measured.

Since the point P1 and the point P1 'are point-symmetrical with respect to the point P0, the point P1 (Xo p1, Yo p1)
Is calculated by the following equation (1).

[0034] (Step 102-3) Then, based on the coordinate positions of the points P0 and P1 obtained as described above, the rotation angle θ of the tool coordinate system XRT-ORT-YRT with respect to the sensor coordinate system Xo-Oo-Yo is as follows (2). It is calculated like an expression.

[0035] (Step 102-4) Further, the magnification B is calculated by the following equation (3).

[0036] (Step 102-5) Then, the reference distance between the visual sensor 3 and the workpiece 4 provided with the reference point 42, that is, the tool tip 2a is the reference point 4
Visual sensor 3 and work 4 when accurately positioned in 2
The reference distance S0 between and is obtained by the following equation (4) using the principle of triangulation.

[0037] In the above equation, f is the focal length of the lens used in the visual sensor 3 (step 102-6).

The coordinate position P0 (Xop0, Yop0) of the origin ORT of the tool coordinate system, the rotation angle θ, the magnification B, and the reference distance S0 obtained as described above are set data of the tool coordinate system and are set as the sensor controller 31. Be stored in.

In this way, when the setting data of the tool coordinate system is acquired and the relative positional relationship of the tool coordinate system with respect to the sensor coordinate system is set, an arbitrary point (on the sensor coordinate system Xo-Oo-Yo ( Xo, Yo) is the following conversion formula (5)
Can be converted into a coordinate position (XRT, YRT) on the tool coordinate system XRT-ORT-YRT.

[0040] When the setting of the relative positional relationship of the tool coordinate system with respect to the sensor coordinate system is completed in this way, the process proceeds to step 103, and the correction amount of the teaching position, that is, the correction amount of the tool 2 in the longitudinal direction, the correction in the plane perpendicular to the tool 2 are performed. A measurement job is created to automatically measure the quantity.

The measurement job is automatically created by a program stored in the sensor controller 31.

When the operation program number to be corrected and the teaching point to be corrected are designated, the sensor controller 31
The corresponding operation program is read from the robot controller 11, and a measurement job as shown in FIG. 5 is created using the robot operation program editing function (step 103).

Then, this measurement job is executed. Processing for obtaining correction amount in tool direction First, two points A and B in a plane perpendicular to the tool 2 are determined within a range in which the teaching point 42 to be corrected can be imaged by the visual sensor 3, and from the point A to the point B. A process of moving the visual sensor 3 (tool 2) is performed. Here, points A and B are teaching points 4
It is assumed that the two are separated by a known distance L1, and
The direction between points A and B is not particularly limited.

However, when it is set in a direction orthogonal to the tool advancing direction (cutting work line direction), interference between the tool 2 and the work 4 is unlikely to occur. Further, when a teaching point having a three-dimensional shape is imaged like a teaching point on the welding line, the visual field of the visual sensor 3 is visually recognized in a direction along the welding line so that the work 4 does not block the visual field. The sensor 3 may be moved.

Further, either one of the two points A and B may be matched with the teaching point 42.

The processing on the left side of the measurement job in FIG.
It is the operation of the robot and is the content executed by the robot controller 11. The processing on the right is an image processing program, and is the content executed by the sensor controller 31.

Therefore, in order to execute the robot operation and the image processing program in parallel, a signal for timing a movement command for moving the robot 1 to each point and a measurement command for measuring an image at each point. Is input to each of the robot controller 11 and the sensor controller 31, and the processing shown in FIG. 5 is sequentially executed based on this timing signal.

That is, of the set two points A and B, each axis of the robot 1 is drive-controlled so that the visual sensor 3 is moved together with the tool 2 to the point A (step 201).

Next, the sensor coordinate system Xo-Oo-Yo of the teaching point 42 when the visual sensor 3 is located at the point A.
The coordinate position A (Xo A, Yo A) on the top is measured based on the imaging result of the visual sensor 3 (step 202).

Then, from the point A to the point B, the known distance L
Each axis of the robot 1 is drive-controlled so that the visual sensor 3 is moved by 1 with the tool 2 (step 20).
3).

Next, the sensor coordinate system Xo-Oo-Yo of the teaching point 42 when the visual sensor 3 is located at the point B.
The coordinate position B (Xo B, Yo B) on the top is measured based on the imaging result of the visual sensor 3 (step 204).

Therefore, the principle of triangulation is used in the same manner as the equation (4), and the coordinate positions A (Xo A, Yo A) and B (Xo B of the teaching point 42 at the two points A and B obtained above are used. ,
The distance S1 between the visual sensor 3 and the work 4 based on
Is calculated by the following equation (6).

[0053] Therefore, as shown in the following expression (7), the visual sensor 3 when the tool tip 2a is accurately positioned at the reference point 42, which has already been obtained and stored in step 102 from the obtained distance S1. By subtracting the reference distance S0 from the work 4, the distance error ΔS, that is, the amount ΔS to be corrected in the longitudinal direction of the tool 2 is obtained.

ΔS = S1−S0 (7) (Step 205) Processing for Obtaining Correction Amount in Plane Vertical to Tool Longitudinal Direction Next, the teaching point 42 is changed by the correction amount ΔS acquired in the above Step 205. A movement command for moving the visual sensor 3 together with the tool 2 to the point C corrected in the tool longitudinal direction is
It is output to the robot controller 11.

In response to this, each axis of the robot 1 is drive-controlled so that the visual sensor 3 is moved together with the tool 2 to the point C (step 206).

Next, the coordinate position C (Xoc, Yoc) on the sensor coordinate system Xo-Oo-Yo when the visual sensor 3 is located at the point C is measured based on the imaging result of the visual sensor 3. (Step 207).

Thus, when the coordinate position C (Xo c, Yo c) of the point C on the sensor coordinate system Xo-Oo-Yo is obtained,
This coordinate position C and the coordinate position P0 (Xo of the origin ORT of the tool coordinate system already obtained and stored in step 102)
p0, Yo p0), rotation angle θ, and magnification B above (5)
A conversion formula similar to the formula, Position shift on the sensor coordinate system (Xoc-Xo
The positional deviation (ΔXRT, ΔYRT) on the tool coordinate system corresponding to p0, Yo c-Yo p0) is obtained. here,
When the tool tip 2a is accurately positioned at the teaching point 42, the teaching point 42 should coincide with the origin ORT of the tool coordinate system, as described above. Therefore, in order to make it coincide with the origin ORT, if the correction is made by the amount of the positional deviation (ΔXRT, ΔYRT) on the tool coordinate system, the tool 2
A correction is made in a plane perpendicular to (step 208). When the step 208 ends, the measurement job ends (step 104).

As described above, after the movement to the point C and the correction in the longitudinal direction of the tool are performed, the correction amount (ΔXRT, ΔYRT) in the plane perpendicular to the tool 2 is calculated. This is to improve the accuracy by keeping the distance from the visual sensor 3 constant.

Of course, the point C is not provided and the point A or the point B is set.
Instead of using the teaching point 42, the tool longitudinal direction correction amount ΔS obtained from the measurement values of these two points and the correction amount (ΔXRT, ΔYRT) in the tool vertical plane directly based on the measurement value at the teaching point 42. You may calculate.

In this way, the correction amount Δ in the longitudinal direction of the tool
When S and the correction amount (ΔXRT, ΔYRT) in the tool vertical plane are acquired, the teaching point 42 is corrected in the longitudinal direction of the tool 2 by the correction amount ΔS and the teaching point 42 is perpendicular to the tool 2. The correction amount (ΔXRT, ΔYRT) is corrected in the plane (on the tool coordinate system). By correcting the teaching position in this way, the robot operation program is corrected, and the corrected robot operation program is transferred to the robot controller 11. When the robot 1 is drive-controlled in accordance with this modified robot operation program, the tool 2 of the robot 1 can be used even if there is a difference between robot parameters or a deflection due to its own weight.
Is accurately moved along each teaching position, and the cutting work is performed accurately (step 105).

In this embodiment, the image measurement is performed for each teaching point to be corrected and the correction amount is obtained. However, for example, when the work is misaligned, one teaching point is corrected. It is also possible to correct a plurality of other teaching points by using the correction amount obtained from the measurement result of 1. In this embodiment, a robot that performs cutting work is assumed, but of course any work such as welding robot can be performed. It is applicable to robots.

Further, in this embodiment, the visual sensor 3
However, it is assumed that the image pickup surface is attached to the tool 2 in such a positional relationship that the image pickup surface is perpendicular to the longitudinal direction of the tool 2, but the image pickup surface is attached in any other positional relationship. Also in this case, a point on the sensor coordinate system can be converted to a point on the tool coordinate system by the same conversion formula as the above formula (5) as follows.

As shown in FIG. 6, when the tool coordinate system is ORT-XRT-YRT-ZRT, the sensor coordinate system is Oo.
-Xo-Yo-Zo is set by three-dimensional coordinates. (1) The origin Oo is offset from the origin ORT by H, L, and α.

(2) It rotates about the Zo axis by θ.

(3) Origin Oo in the plane including ZRT axis and Zo axis
Rotate the Zo axis by β around. Is given to indicate that the visual sensor 3 has an arbitrary positional relationship with the tool 2.

Then, the conditions of (2) and (3) above are: (2) 'rotate about the Zo axis by α.

(3) 'Rotate by β around the Yo axis.

(4) 'Rotate about the Zo axis by (θ-α).

Since it is equivalent to that, the conversion equation is obtained as in the following equation (9).

In the following, in order to avoid complication, sin
(Sine) is represented by “s” in the formula, and cos (cosine) is represented by “c” in the formula. Further, θ-α is represented by “γ”.

[0071] The leftmost matrix on the right side of the above equation (9) is a matrix for rotating by -α, the matrix on the right side is a matrix for rotating by -β, and the matrix on the right side is- This is a matrix for rotating by γ. The rightmost matrix is for the offset (the above condition (1)).

The above equation (9) can be summarized as the following equation (10).

[0073] Therefore, the point (Xo, Yo, 0) on the sensor coordinate system is converted to the point (XRT,
YRT, ZRT) can be converted.

[0074]

[Brief description of drawings]

FIG. 1 is a perspective view showing an appearance of a robot applied to an embodiment of a teaching position correcting apparatus for a robot according to the present invention.

FIG. 2 is a diagram showing a configuration of a system assumed in the embodiment.

FIG. 3 is a flowchart showing a procedure of processing executed by the sensor controller and robot controller shown in FIG.

FIG. 4 is a diagram showing an imaging field of view of the visual sensor shown in FIGS. 1 and 2;

5 is a flowchart illustrating details of the contents executed in step 104 of FIG.

FIG. 6 is a perspective view showing a three-dimensional geometrical relationship between a tool coordinate system and a sensor coordinate system.

[Explanation of symbols]

 1 Robot 2 Tool 3 Visual sensor 4 Work piece 11 Robot controller 31 Sensor controller

Claims (1)

[Claims] 1. A position on a work to which a tip of a tool attached to an arm of a robot is to be moved is taught,
A robot adapted to drive the robot so that the tool tip is located at the teaching position, wherein a visual image of a teaching position to which the tool tip of the robot is to be moved is attached to the tool in a predetermined positional relationship. A sensor and a sensor coordinate that moves with the tool and sets a tool coordinate system that displays the coordinate position on a plane perpendicular to the longitudinal direction of the tool, and that displays the coordinate position within the imaging field of view of the visual sensor. A coordinate system setting means for setting a system and the visual sensor are moved by a predetermined distance, and one visual instruction position is imaged by the visual sensor for each moving position, and from this imaging result, the sensor is set for each moving position. First measuring means for measuring the coordinate position of the one teaching position on the coordinate system, and the first measuring means for measuring the coordinate position. The distance between the visual sensor and the work is calculated based on each coordinate position of the teaching position and the predetermined distance moved by the visual sensor, and the calculated distance and the tool tip become the one teaching position. A first calculation means for calculating an error between a reference distance between the visual sensor and the work when it is positioned; and an image of the one teaching position by the visual sensor, and based on the result of the imaging, the sensor coordinate system Second measuring means for measuring the coordinate position of the one teaching position on the above, the coordinate position of the one teaching position measured by the second measuring means, and the tool tip at the one teaching position On the sensor coordinate system, based on the positional deviation between the reference coordinate position of the one teaching position on the sensor coordinate system when positioned and the relative positional relationship between the sensor coordinate system and the tool coordinate system. And second calculating means for calculating a positional deviation on the tool coordinate system corresponding to the kicking position shift, the longitudinal direction of the one taught position is the tool, the first
According to the positional deviation calculated by the second calculating means, the one teaching position is corrected in the direction perpendicular to the tool so as to be corrected by the error calculated by the calculating means. A teaching position correcting device for a robot, comprising a correcting means for correcting the teaching position so as to be corrected by the amount.