JPH0247761B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an industrial robot that performs work such as welding, cutting, or painting in accordance with command signals, and in particular to a control device for smooth operation when changing the posture of a processing tool. It is related to.

In general, industrial robots use coordinate axes (hereinafter these coordinate axes are referred to as main coordinate axes) for moving the processing tool and the workpiece relative to each other, for example, along three mutually perpendicular coordinate axes, X, Y, and Z. ,
The posture control axis is a sub-coordinate axis for determining the posture of the wrist that holds this processing tool, for example, Θ rotates in a plane parallel to the X-axis and Y-axis, that is, the XY plane.
The Ψ-axis rotates in a plane parallel to the Z-axis (hereinafter, the Θ-axis and Ψ-axis are collectively referred to as wrist axes), and these axes can be moved as instructed to move within a three-dimensional space. The process is controlled to faithfully reproduce the processed lines formed in the process.

FIG. 1 is a perspective view showing an example of an industrial robot configured in this way. In the figure, 1 is a base, 2 is a column mounted on the base 1 so as to be able to move freely in the X-axis direction, and 3 is a column. An arm support 4 is attached to the column 2 so that it can move up and down in the Z-axis direction, and an arm 4 is supported on the arm support 3 so that it can move back and forth in the Y-axis direction. A wrist 5 having a torch is attached, and this wrist 5 can control the posture of the processing tool 6 by rotating it around the Θ axis in the XY plane and around the Ψ axis in a plane parallel to the Z axis. It's getting old. These column 2, arm support 3, and arm 4 each constitute orthogonal axes as main coordinate axes, and are each driven and positioned to a commanded position by a driving means (not shown) such as an electric motor or a hydraulic cylinder. Furthermore, these robot bodies are installed in correspondence with a mounting jig 102 for the workpiece 101 that is prepared separately. The workpiece mounting jig is configured to be rotatable around the U-axis and V-axis in the figure and movable in the W-axis direction as necessary.

In this type of industrial robot, when machining a curved or polygonal machining line, the machining tool must always be kept in a constant position relative to the machining line of the workpiece. To simplify the explanation, in the XY plane as shown in Figure 2, B
Consider a case where there are processing lines A, B, and C bent at a point, and the processing tool 6 also moves within the XY plane. In the same figure, since the line between points A and B is direct, the posture of the processing tool can remain constant, and since it is bent at point B, the wrist axis can be rotated by θ and the position between B and C is also relative to the line of processing. It is necessary that the processing tool 6 forms the same angle α as between A and B. At this time, without changing the position of the wrist support part, that is, the tip of the arm 4 in FIG.
When rotated, since the center of rotation is separated from the tip of the processing tool, the tip of the processing tool is located at a position far away from the original pointing position, as shown by the dotted line in the figure.

Conventionally, when machining such a bending machining line or in order to prevent the pointing position of the tip of the machining tool from changing even if the attitude of the machining tool with respect to the machining line changes, there is a step of memorizing the machining line, that is, teaching. At the same time, the change in the position of the tip of the processing tool that occurred due to the change in wrist posture is moved to the X-axis and Y-axis by a manual command, and after confirming that the tip of the processing tool has returned to its normal position, a teaching command is issued. Target values for the wrist axis, X-axis, and Y-axis are memorized, and during execution, the amount of movement required for each of the X and Y axes when changing the wrist posture is calculated by a computer from the previously memorized target values. A method was used in which this was done while When using such a method, not only does the teaching work become very complicated, but also the rotation of the wrist axis and the correction operation of the arm tip position are performed independently during execution. At this time, it is necessary to match the dynamic characteristics of each axis in order to keep the position of the tip of the processing tool completely immovable even when the wrist rotates. By the way, mechanisms with different weights are attached to the X, Y, and Z axes, which are the main coordinate axes, and the loop gain and response speed are naturally different from the Θ and Ψ axes for wrist rotation, which are the polar coordinate system. It is quite difficult to adjust these accurately to match the dynamic characteristics. Moreover,
The dynamic characteristics of each of these axes are not constant because the load weight and moment of inertia change depending on the current position of each axis, and it is almost impossible to match these varying dynamic characteristics at all positions. close.

On the other hand, in order to facilitate teaching work, the rotation command of the wrist axis is divided, and in order to hold the tip of the processing tool in its original position according to this command signal, the amount by which the tip of the arm should be moved is calculated. This calculation result can be added to the position command signals for each of the X, Y, and Z axes to perform control while correcting the position of the main coordinate axes, or the rotation angle of the wrist axis can be detected and the X, Y, and Y axes can be controlled by , Z. A method of calculating and controlling the amount of correction for each coordinate axis has been proposed. However, in the former case, the rotation of the wrist and the driving of the main coordinate axes for correction are performed independently based on the wrist axis rotation command signal, so even if a common clock pulse is used between the wrist axis and the main coordinate axes, Even if synchronization is attempted by supplying a Since the correction operation of the coordinate axes takes precedence, it is not possible to change the posture by rotating the wrist axis without accurately changing the pointing position of the tip of the processing tool. On the other hand, in the latter case, the amount of rotation of the wrist axis is detected, and then the correction amount of the main coordinate axis is calculated based on this detected value, and the main coordinate axis is moved based on the result of this calculation, so a considerable amount of delay in operation can be avoided. I can't. Furthermore, in order to accurately correct the position of the principal coordinate axes, it is necessary to detect the rotation angle of the wrist axis at as fine an interval as possible, but the calculation of this correction amount involves trigonometric functions as described later. Therefore, when performing such calculations at small intervals, it is necessary to calculate a large number of trigonometric functions at high speed, which is completely impossible to achieve with the performance of a microcomputer generally used in industrial robots. Therefore, the detection interval is restricted within the range of computing power, an increase in delay is inevitable, and accurate correction operation cannot be expected.
Of course, if you memorize the function table without calculating the trigonometric functions and then read out the necessary numbers and perform only the four arithmetic calculations, the calculation speed will be faster, but the storage capacity will be increased to memorize the function table. A large portion will be occupied and other work will be hindered.

Further, in the conventional example described above, the values of the main coordinate axes to be corrected by rotation of the wrist axis are immediately output and added to the command signals of each coordinate axis. However, in a servo system, there is always a response delay in response to a command signal, and since this delay amount is not constant for each axis, a correction amount is immediately calculated for the rotation command of the wrist axis. When the position correction operation of the main coordinate axes is performed, there is a possibility that the main coordinate axes are corrected in advance, which may induce unnecessary movement of the processing tool.

The present invention improves the drawbacks of the conventional devices described above, and in an apparatus that changes the attitude of a processing tool by rotating the wrist axis, the attitude control axis, that is, the wrist axis, is driven in response to an attitude control command signal. At the same time, an arithmetic circuit is provided that receives the attitude control command signal and predicts and calculates in advance the amount of change in the pointing position of the tip of the processing tool that will occur due to the attitude change. This detection signal is used to read out the predicted amount of change in the pointing position of the tip of the processing tool, which was previously calculated and stored, and the position command signal for the main coordinate axes is corrected using this predicted value to control the position of the main coordinate axes. This structure enables accurate and timely position control without changing the directional position of the tip of the processing tool at all when changing the attitude of the processing tool by rotating the wrist axis.

FIG. 3 shows the implementation of the present invention when applied to an industrial robot having three orthogonal axes X, Y, and Z as principal coordinate axes and two rotational axes Θ and Ψ as wrist axes, as shown in FIG. It is a block diagram which shows an example. In the figure, 7 is a controller that generates command signals for each axis, 8 is a teaching operation box for manually commanding the movement of each axis, and 9 is an automatic controller that commands start and stop of operation during automatic operation. It is an operation box. 11a is an X-axis drive control circuit, 11b is an X-axis drive motor, and 11c is an X-axis position detection encoder, and these 11a to 11c constitute an X-axis positioning servo control circuit. Similarly 12a, 1
3a, 14a, and 15a are the Y axis, Z axis, and Θ, respectively.
Axis and Ψ axis drive control circuits, 12b, 13
b, 14b, 15b are electric motors for driving each axis, 12
c, 13c, 14c, 15c are encoders for detecting the position of each axis, 12a to 12c are for the Y axis,
13a to 13c are for Z axis, 14a to 14
c constitutes a positioning servo control circuit for the Θ axis, and 15a to 15c constitute a positioning servo control circuit for the Ψ axis.
16 and 17 are Θ-axis and Ψ-axis current position storage registers that integrate and store the Θ-axis position command signal and Ψ-axis position command signal from the controller 7, respectively, and when each axis returns to its home position, It will be reset. Here encoders 14c and 15c
alone and these registers 16 and 17 constitute a detector that detects the amount of movement of the Θ and Ψ axes of the wrist 5, which are posture control axes. 18 is the controller 7
Incremental position command θc of both axes for the Θ and Ψ axes from
19 is generally a waiting queue or
This is a storage circuit called FIFo memory, which stores the calculation results in the order in which they arrive using the calculation end signal m of the calculation circuit 18 as a write command, and stores the calculation results in the order in which they arrive as a read command with the interpolation completion signal r of the interpolation circuit 20, which will be described later. This is a circuit that reads from. This memory circuit 19 is an important element of the present invention together with the arithmetic circuit 18 and the interpolation circuit 20, and as will be explained in detail later, extremely smooth and accurate operation can be obtained by providing these circuits. It is.

Encoders 14c and 15c 20 receive the output of the memory circuit 19 and detect movement of the Θ and Ψ axes.
Linear interpolation circuits 21 to 23 are correction circuits that correct the output of the controller 7 using the output of the linear interpolation circuit 20 to obtain a total output.

In Fig. 3, the controller 7 outputs a position command signal _xc to _φc for each axis in each command unit at a pulse interval corresponding to a separately determined moving speed, and each axis receives this position command signal. It is driven by a servo control circuit in response to an input signal.

The positioning results are detected by the encoders 11c to 15c provided for each axis and fed back to the control circuits 11a to 15a for the respective axes, and the axes are stopped at the position where there is no deviation from the input signal. The position command signals θc and φc for the wrist axes are also integrated by the Θ-axis and Ψ-axis current position recording registers 16 and 17, and are also supplied to the arithmetic circuit 18. The registers 16 and 17 integrate the rotation angles θ and φ of the wrist axis from the origin and output the result to the arithmetic circuit 18. The arithmetic circuit 18 calculates the amount of correction for each of the X, Y, and Z axes from the θ and φ, and outputs it to the storage circuit 19, as will be described in detail later. The storage circuit 19 receives the calculation results upon completion of the calculation by the calculation circuit 18 and sequentially stores them. On the other hand, in response to command signals θc and φc from the wrist axis, the Θ and Ψ axes begin to rotate, and the encoder 1 accordingly begins to rotate.
4c and 15c generate output pulses. The interpolation circuit 2 uses the output pulses from this encoder.
0 reads the contents of the memory circuit 19 in the order in which they are stored, distributes the stored contents in pulses according to the output pulses from the encoders 14c and 15c, and sequentially outputs correction amounts for each of the X, Y, and Z orthogonal axes by linear interpolation. This correction amount is determined by the correction circuit 21 or 2.
3, the command signals x _c , y _c , z _c from the controller 7 are added and synthesized, and the signal is added to the orthogonal axis drive control circuits 11a to 1.
3a. As a result, the tip of the processing tool always faces the same pointing position regardless of changes in the positions of the wrist axes Θ and Ψ axes.

Here, the arithmetic circuit 1 occupies an important part of the present invention.
8. The memory circuit 19 and the interpolation circuit 20 will be explained in detail with reference to FIG.

FIG. 4 is an explanatory diagram showing the relationship with the coordinate axes in which only the processing tool portion of the apparatus shown in FIG. 1 is extracted. Fourth
In the figure, l ₁ indicates the length of the wrist from the rotation center of the Θ axis of the wrist to the attachment position of the processing tool, and l ₂ indicates the effective length of the processing tool. Now, consider the case where the wrist attachment position is at point Q and the tip of the processing tool is at point _P0 .
Let the coordinates of point P ₀ be P ₀ (x ₀ , y ₀ , z ₀ ), and from this, Θ
If the position of the tip of the processing tool when rotated by Θ around the axis and φ around the Ψ axis is P ₁ (x ₁ , y ₁ , z ₁ ), then
At this time, point P ₁ is from point P ₀ as follows: Δx ₁ = l ₁ (cosθ−1) − l ₂ sinφsinθ …(1) Δy ₁ = l ₁ sinθ+l ₂ sinφcosθ …(2) Δz ₁ = l ₂ (1−cosφ) It will move by (3). Therefore, when the position of the wrist axis is (θ, φ), the position of the tip of the processing tool is
P ₁ (x ₀ +Δx ₁ , y ₀ +Δy ₁ , z ₀ +Δz ₁ ). In order to maintain this at the original position P ₀ , the wrist attachment portion Q only needs to be moved by (−Δx ₁ , −Δy ₁ , −Δz ₁ ) along the orthogonal coordinate axes. From the state corrected in this way, the wrist axis is further adjusted to θ _c , φ _c which is one command unit.
The position of the tip of the processing tool (Δx ₂ , Δy ₂ , Δz ₂ ) at this time is determined by setting θ as (θ + θ _c ) and φ as (φ + φ _c ) in equations (1) to (3) above. However, in reality, the wrist attachment part is already (−
Since the amount to be corrected this time is (Δx ₂ −Δx ₁ , _{Δy 2 −Δy 1 ,}
Δz ₂ −Δz ₁ ). Therefore, the arithmetic circuit 18
is the above (1) or
Calculate the formula (3) and memorize it. Δx, Δy, Δz one command unit ago that was previously calculated and stored.
It is sufficient if the difference is subtracted from the value of , and the difference is output as correction signals for the orthogonal axes X, Y, and Z to the storage circuit 19 together with the computation end signal.

By the way, in microcomputers used in industrial robots, it is convenient for the above-mentioned calculations to be performed using a method in which digital signals are normally input and digital signals are output. And above (1)
When outputting the calculation results in digital quantities in the calculations of formulas (3) and 3), values less than one unit amount are rounded down for each calculation, so simply the digital calculation results (Δx, Δy, Δz ) and the calculation result one command unit before is used as the correction amount, fractions are rounded down each time calculation is performed, and this may be sequentially accumulated, resulting in a large error. To prevent this, when the arithmetic circuit 18 receives the n-th command signal for the Θ and Ψ axes, Δx _o , Δy _o ,
The first arithmetic circuit that calculates Δz _o and the (nth
−1) Total sum of the correction amounts Δx′, Δy′, and Δz′ of the orthogonal coordinate axes calculated for each command unit up to the th point _o-1 〓 ⁱ⁼¹ Δx′ _i , _o-1 〓 ⁱ⁼¹
Δy′ _i , _o-1 〓 ⁱ⁼¹ Subtract the addition result of this adder from the calculation result of the first calculation circuit and the adder that calculates Δz _i Δx′ _o = Δx _o − _o-1 〓 ⁱ⁼¹ Δx′ _i …(4) Δy′ _o =Δy _o − _o-1 〓 ⁱ⁼¹ Δx′ _i …(5) Δz′ _o =Δz _o − _o-1 〓 ⁱ⁼¹ Δx′ _i …( 6) and a second arithmetic circuit that calculates . By configuring the arithmetic circuit 18 in this manner, the above-mentioned error when performing arithmetic operations using digitized values can be compensated for each command unit of the Θ and Ψ axes as needed. As a result, the occurrence of errors is extremely reduced, and accuracy can be dramatically improved.

Next, the role and operation of the memory circuit 19 will be explained. As mentioned above, in a servo control system having a feedback loop, a delay with respect to a command signal always occurs at the time of startup and during steady state. In other words, the operation of the servo system is possible due to the presence of this delay. Therefore, when each axis receives a command signal, it moves in a direction toward the target with a delay unique to each axis. As explained in FIGS. 2 and 4, when the wrist axes Θ and Ψ axes receive a command signal, a predetermined calculation is performed on this command signal, and the result of this calculation is used to calculate the main coordinate axis (this In cases where correction commands are issued for the X, Y, and Z axes, if these correction operations are performed by handling only the command signal, the axes that originally caused the correction (i.e., the Θ and Ψ axes) and the axes (i.e., X, Y,
The order of the operations may be reversed with respect to the Z-axis operations. In other words, if the response of the axes receiving the correction command is fast, the X, Y, and Z axes will execute correction operations first while the wrist Θ and Ψ axes have not yet rotated enough to require correction of the principal coordinate axes. Stowing away occurs. In order to prevent this, the memory circuit 19 is provided, which only calculates the correction amount of the principal coordinate axes with respect to the command value of the wrist axes (Θ, Ψ axis), temporarily stores this calculation result, and then When the Θ and Ψ axes are rotated to a value that requires correction, the calculation results are sequentially read out and the principal coordinate axes are corrected. In this way, the amount of correction calculated in advance for the rotation commands of the Θ and Ψ axes can be applied to the actual Θ,
When the Ψ axis rotates, it is output to the main coordinate axes, so regardless of the amount of delay in the operation of the Θ and Ψ axes, that is, the dynamic characteristics, only the necessary amount of correction is always executed, making the operation extremely stable and reliable. The interpolation circuit 20 reads out the memory content at the beginning of the memory circuit 19, temporarily stores it, and distributes the memory content in pulses by linear interpolation using the outputs of the encoders 14c and 15c that detect the rotation of the attitude control axes Θ and Ψ. . Here, when the outputs of the encoders 14c and 15c emit a pulse of 1 unit for each 1 command unit of the Θ and Ψ axes, the interpolation circuit 20 is naturally unnecessary, and the stored contents of the memory circuit 19 are directly transmitted to the encoder 14c. It is preferable to read out the output of 15c and 15c as a read command and supply it to the correction circuits 21 to 23. However, in this case, it is sufficient that the correction amount obtained by the above-mentioned equations (4) to (6) for one command unit of the Θ and Ψ axes is approximately one command unit of each of the X, Y, and Z axes, but When the hand length l ₁ and processing tool length l ₂ in the figure become large values, X,
The amount of correction for each of the Y and Z axes can reach several tens to hundreds of command units. If such a large amount of correction is made at once, the operation of the tip of the processing tool becomes irregular, and the initial purpose may not be achieved. To solve this problem, encoder 14c
The output of 15c is assumed to generate n times the number of commands for the Θ and Ψ axes, and the correction amount read from the memory circuit 19 is linearly interpolated using the n times the number of commands for the Θ and Ψ axes. All you have to do is distribute it. In this way, for one command unit of the Θ and Ψ axes, n
It becomes possible to correct the position of each of the X, Y, and Z axes with twice the resolution. In this case, command signals x _c , y _c , z _c for the X, Y, and Z axes and command signals θ _c for the Θ and Ψ axes,
It is necessary to have a pulse ratio of n:1 or more between φ _c , but in general, the amount of movement in each of the X, Y, and Z axes is several meters per minute or more, and the corresponding pulse is a multidirectional pulse. reach On the other hand, Θ, Ψ
The amount of movement of both axes is on the order of several revolutions per minute, and the pulses required for this are several to several pulses. Therefore, it is sufficient to set n in the above ratio to about 10.

Of course, the interpolation circuit 20 may be omitted if the storage circuit 19 obtains an output obtained by linearly interpolating the output in accordance with an external pulse distribution command.

Note that in the above explanation, in order to simplify the explanation, an industrial robot that positions a processing tool along three orthogonal axes of X, Y, and Z was explained.
The number of these principal coordinate axes may be two or less, and the control device of the present invention can also be applied to a robot having only one of the Θ axis and the Ψ axis for the wrist axis. In this case, as can be easily understood from equations (1) to (6) above, in contrast to the device shown in Figure 1 in which the orthogonal axes are only the X and Y axes, only the Θ axis and only the Z and X axes are used. For such devices, it is applicable only to the Ψ axis. Furthermore, since machining is performed by relative movement between the processing tool and the workpiece, it goes without saying that any axis among these principal coordinate axes may be used as the axis for controlling the position of the workpiece. . For example, in the apparatus shown in FIG. 1, the apparatus of the present invention is applied to control the position of the W-axis of the workpiece mounting jig instead of the X-axis. When doing this, there is no need to perform delicate correction operations on the X-axis, where it is difficult to control accurate positioning since both the load amount and the moment of inertia are maximized. It is the sum of the ranges that can be used to measure functional improvement.

Furthermore, the present invention is not only applicable to cases where the principal coordinate axes are constituted by a rectangular coordinate system as shown in the above embodiments, but also to cases where the principal coordinate axes are constituted by a polar coordinate system and a cylindrical coordinate system. It is easy to understand that the present invention can be applied to industrial robots configured with any coordinate systems, such as cases where these coordinate systems are partially mixed, or cases where a mixture of these coordinate systems is used. In these cases, it goes without saying that the calculation contents of the calculation circuit and the type of command signals for the coordinate axes are determined according to each coordinate system.

As described above, when the device of the present invention is used, the command signals for the wrist posture control axes Θ and Ψ are received, and the amount of position correction of the main coordinate axes that occurs due to the posture change is calculated, and this is sequentially stored. The stored correction amount is sequentially read out using the detection pulse that detects the movement according to the movement of the main coordinate axis, and the position command value of the main coordinate axes is corrected to drive each axis. This can be done smoothly while keeping the pointing position of the tool tip at a predetermined position.
In addition, since the calculation of the correction amount itself is performed for each command unit of the Θ and Ψ axes when the attitude command signal is transmitted, the calculation time can be relatively long.The calculation results are also temporarily stored in a memory circuit and sequentially stored. Since the principal coordinate axes are corrected by reading them out, there are no errors due to response delays or unnecessary correction operations, and corrections can be made stably and accurately. Furthermore, when correcting the memorized correction amount by linearly interpolating the movement amount of the wrist axis using the output of an encoder that generates an output pulse number n times the command unit, the position of the principal coordinate axes can be corrected with high resolution. This can be done at

[Brief explanation of the drawing]

FIG. 1 is a perspective view showing the external shape of an industrial robot to which the present invention is applied, FIGS. 2 and 4 are explanatory views showing how the processing tool changes its posture, and FIG.
The figure is a configuration diagram showing an embodiment of the apparatus of the present invention. 2...Column, 3...Arm support, 4...Arm, 5...Wrist part, 6...Processing tool, 7...Controller, 11a to 15a...Drive control circuit for each axis, 1
1b to 15b...drive motor, 11c to 15c...
... Encoder, 18 ... Arithmetic circuit, 19 ... Memory circuit, 20 ... Interpolation circuit, 21 to 23 ... Correction circuit, X, Y, Z ... Orthogonal coordinate axes, Θ, Ψ ... Attitude control axes.

Claims (1)

[Claims] 1. A controller that has a coordinate axis for relatively moving a processing tool and a workpiece, and an attitude control axis that determines the attitude of the processing tool, and that commands the position of each of the axes. and a servo control circuit for positioning each axis according to a position command signal from the controller, further comprising a detector for detecting the amount of movement of the attitude control axis, and a servo control circuit for positioning each axis according to a position command signal from the controller. an arithmetic circuit that receives a command signal for the attitude control axis among the signals and calculates a change in the processing tool tip pointing position that occurs when changing the attitude and calculates a position correction signal for the coordinate axis; and an arithmetic circuit that stores the output of the arithmetic circuit and An industrial device comprising: a memory circuit that reads out data corresponding to the output of the detector; and a correction circuit that corrects a command signal for the coordinate axes among the position command signals using the output signal of the memory circuit and transmits the corrected signal to the servo control circuit. Robot control device. 2. The detector includes an encoder that outputs a signal obtained by dividing the basic movement amount corresponding to one command unit of the attitude control command signal into n (n≧1), and the storage circuit stores the stored value in the attitude based on the output of the encoder. 2. The control device for an industrial robot according to claim 1, wherein the circuit includes a circuit that distributes and outputs one command unit of a control command signal by a linear interpolation method divided into n parts. 3. The arithmetic circuit includes a first arithmetic unit that calculates a position correction signal for the coordinate axes according to a rotation angle from the origin of each axis when receiving an attitude change command signal for the attitude control axes; and a second computing unit that subtracts the sum of the correction amounts of the coordinate axes calculated before receiving the attitude change command signal of the attitude control axis from the output of the computing unit. Robot control device.