JPH07302110A - Production of robot operation program - Google Patents

Abstract

PURPOSE:To extremely improve the coincidence between the track of a robot and its target one in a non-positioning control state. CONSTITUTION:The dividing points Ai are decided to divide a target robot track Qd (t) into many minute sections, and a teaching point generation intervenient point Bj is extracted out of these points Ai in response to the curvature. Each point Bj is extacted under the condition that the space between these points does not exceed a teaching point non-setting allowable section length index Lmax and every time the directional change of tangent vector <gi> exceeds a teaching point setting density index N at each of points Ai. Then a teaching point Pj is set at a position that is decided when the point BJ is shifted toward the center of the curvature according to the curvature of the track Qd (t) at the position of the point BJ. Then an interpolation system, a moving speed and a positioning rate are designated and a robot operation program is produced for execution of a trial reproduction operation. Meanwhile an actual robot track Qo (t) is recorded and the difference between both tracks Qo (t) and Qd (t) is evaluated. Then the trial operation is repeated while the parameter is changed to decide a shift extent toward the center of the curvature until the satisfactory result of evaluation is obtained.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for creating a robot operation program using an off-line programming device, and in particular, it is possible to control the robot in an operation mode in which it is not positioned, such as a sealing operation using the robot. The present invention relates to a method of creating a robot operation program suitable for a desired case.

[0002]

2. Description of the Related Art Robot control methods include a method of moving the robot while positioning the robot at each teaching point, and a method of moving the robot without positioning the robot at each teaching point (hereinafter referred to as "non-positioning control"). There is. For example, the former is usually used in spot welding, and the latter is usually used in painting-type work such as sealing work.

In the non-positioning control, since the robot is not positioned for each teaching, the robot trajectory during the reproduction operation of the robot operation program does not pass the teaching point, and the deviation of the robot trajectory from the teaching trajectory becomes large. Tend to be. Therefore, the trajectory that is intended for the robot trajectory during the regenerative operation (hereinafter referred to as "target robot trajectory").
In order to get closer to the above, it is necessary to take a large number of teaching points and perform a trial and error robot teaching operation over a long time.

[0004]

SUMMARY OF THE INVENTION An object of the present invention is to solve the above problems. That is, the invention of the present application is to provide a method of creating a robot motion program capable of significantly improving the degree of coincidence between a robot trajectory and a target robot trajectory realized during non-positioning control.

[0005]

According to the present invention, the basic configuration for solving the above technical problems is as follows: "(a) a step of designating a target robot trajectory on a robot motion program creating device; ) Determining a division point that divides the target robot trajectory into a large number of minute sections, and extracting teaching point generation intermediary points from the division points at a density according to the curvature in the vicinity of each division point, and (c). A step of correcting the position of each teaching point generation intermediary point according to the curvature in the vicinity of each teaching point generation intermediary point to determine a teaching point; and (d) a robot operation program including the determined teaching point data. A method of creating a robot operation program, characterized in that it includes a step of creating the program (the configuration described in claim 1).

Further, according to the present invention, by further expanding the above-mentioned basic configuration, it is possible to sequentially create a more desirable robot motion program, and "(a) specify a target robot trajectory on the robot motion program generation device. And (b) defining division points for dividing the target robot trajectory into a large number of minute sections, and extracting teaching point generation intermediary points from the division points at a density according to the curvature in the vicinity of each division point. And (c) determining the teaching point by correcting the position of each of the teaching point generation intermediary points according to the curvature in the vicinity of each of the teaching point generation intermediary points, and (d) the determined teaching point data. A step of creating a robot operation program to be included, and (e) a reproduction operation of the created robot operation program is executed while recording at least a reproduction robot trajectory. And (f) evaluating the deviation between the regenerated robot trajectory and the target robot trajectory, wherein the steps (c) to (f) include the correction amount in the step (c). While changing, the step (f)
The method for creating a robot operation program is characterized in that the above steps are repeated until a good evaluation result is obtained in (1) (the configuration described in claim 2).

[0007]

The present invention is premised on that the data of the target robot trajectory is prepared as offline data. This off-line data is, for example, on the off-line programming device, based on CAD data or the shape data of the work to be prepared prepared in another format, the shape data (line locus in case of sealing work) (line locus). Data) is specified.

Here, in the data of the target robot trajectory, the elements of straight lines, arcs or splines forming the robot trajectory are defined by a predetermined coordinate system (generally a three-dimensional coordinate system).
It is assumed to include the numerical expression data that can be specified above.

If the target robot trajectory is prepared as offline data, a division point for dividing the target robot trajectory into a number of minute sections is defined, and teaching is performed from the division points at a density according to the curvature in the vicinity of each division point. Extract point generation mediation points. Therefore, the following methods can be used. That is, first, the teaching point setting density index N, the teaching point non-setting allowable section length index Lmax, and the minute division index Δ are designated. The teaching point setting density index N is for adjusting the teaching point setting density according to the magnitude of the curvature of the target robot trajectory, and the teaching point non-setting allowable section length index Lmax is the section length for which teaching points are not set. It is set according to the maximum allowable value. The minute division index Δ is an index that specifies a unit of division (distance) when the specified line segment is divided, and is set to Δ = 0.1 mm, for example.

Further, an index α representing an interpolation method, a robot moving speed v (teaching speed), and a positioning ratio (deceleration ratio in acceleration / deceleration control when passing a teaching point) is designated. These settings are usually performed by manual input to an offline programming device. When the above preparation is completed, the robot operation program is automatically created by software processing on the offline programming device. The most important feature of the present invention is the teaching point generation process included in the automatic creation process of the robot operation program. This teaching point is determined by correcting the position of each of the extracted teaching point generation intermediary points according to the curvature in the vicinity of each teaching point generation intermediary point.

FIG. 1 is a conceptual diagram for explaining in principle a specific example of the teaching point generation process adopted in the present invention. Hereinafter, with reference to FIG. 1, an outline of a method example used in the teaching point generation process will be described. In the following description, the symbol <> represents a vector.

[1] A target robot trajectory is represented by a function Qd (t) at time t. From the above, the trajectory Qd (t) is a connection of three-dimensional straight line, arc, or spline elements. In Figure 1,
For convenience of explanation, this is conceptually shown by a general line from the start point P0 to the end point Pn.

When the trajectory Qd (t) is designated, first, the trajectory Qd (t) is divided along the moving direction of the robot by the minute division index Δ. Each division point is A1, A2.
..A small section of starting point Ai-1 and ending point Ai as An (i = 1,
2 .... n) is Gi. Here, the sections G1 to Gn-1
May be divided by an accurate section length Δ, and the section length of the final section Gn may be equal to or smaller than Δ (remainder section), but n is set as the minimum numerical value that gives the evenly divided minute section section length that does not exceed the index Δ, and all sections are May be divided into n equal parts. Also, for convenience, A1
, A2 ... An, in addition to the starting point A0 of the trajectory Qd (t)
, The end point An is also called a division point.

[2] Next, in order to grasp the transition of the bending of the trajectory Qd (t), as shown in the figure, the dividing points Ai-1 to Ai (i = 1, 2, ... We define a unit vector, the tangent vector <gi>, along the direction towards n).

[3] The inner product β01 of the tangent vectors <g0> and <g1> corresponding to the adjacent minute intervals is calculated. if,
Inner product β01> (or ≧, hereinafter omitted) Teaching point setting density index N
Greater than or the section long distance L between A0 and A1
01> (or ≧, hereinafter abbreviated) When the teaching point non-setting allowable section length index Lmax is set, the division point A1 is set to the teaching point generation intermediary point B1.
And (Hereinafter, the inner product of the tangent vectors <gi>, <gj> is represented by βij and the section long distance between Ai and Aj is represented by Lij.) In other cases, that is, β01 ≦ (or <, hereinafter omitted).
If N and L01 ≤ (or <, abbreviated below), the teaching point setting density index N and the inner product β02 and the section length L02 (= L01 + L12) are set without using the dividing point A1 as the teaching point generation intermediary point. The size relationship with the teaching point non-setting allowable section length index Lmax is checked. The inner product β02> N or L
If 02> Lmax, the dividing point A2 is the teaching point generation intermediate point B.
If it is 1, otherwise, the same calculation and magnitude comparison are performed in relation to the next division point A3.

In the same manner, the teaching point generation mediation points B1, B2, ... Are sequentially determined, and both end points of the trajectory Qd (t) are added to them as B0 = A0, Bm = An, and all teaching point generation mediation points are added. Designate points B0 to Bm. As shown in the figure after the extraction of the teaching point generating intermediary points also shown in FIG. 1, the teaching point generating intermediary points are densely distributed in a portion having a relatively large curvature on the trajectory Qd (t). However, it is sparsely distributed in the portion where the curvature is relatively small. Within the long straight line portion, they are sparsely distributed at intervals of approximately Lmax.

[4] Designated teaching point generation intermediate point B0-
The operation represented by the following equation (1) is performed on Bm to generate teaching points P0 to Pm. <Pj> = (k × κj) · <hj> + <Bj> (1) where <Pj>; a position vector representing the teaching point Pj <Bj>; a teaching point generation intermediary point Bj Position vector k; 0 or a positive coefficient parameter κj; Curvature at teaching point generation mediation point Bj (reciprocal of curvature radius) <hj>; Normal vector set up toward the center of curvature at teaching point generation mediation point Bj (straight line Part (part of κ j = 0)
Can be set in any direction perpendicular to the corresponding tangent vector <gi> (it can be a zero vector).

The reason that the coefficient k is 0 or positive is that the robot trajectory of the robot during non-positioning control has a general property that it tends to deviate from the center of curvature at each position with respect to the taught trajectory. Is.

Regarding the starting point <P0> of the trajectory Qd (t), κ0 = κ1, <h0> = <h1> or <
Appropriate end point processing such as κ 0> = 0 may be performed.

[5] In the above process, all teaching points have been set for the time being, but in reality, the coefficient parameter k
If the value of is not appropriate, the robot trajectory during regenerative operation is
It deviates from the target trajectory Qd (t). In general, the optimum value of k is influenced by the specified interpolation method, moving speed v, and positioning rate α. Therefore, when the value of k that gives a satisfactory result in relation to these conditions is known by estimation from similar cases, empirical rules, etc., the teaching point P0 determined by setting k to that value is set. .About.Pm is determined as the final teaching point as it is, and a robot operation program is created according to the specified interpolation method, moving speed v, and positioning ratio α, and this is used as the final program.

[6] However, since it is generally difficult to know the optimum k in advance, an appropriate initial value k0 is set as k and the teaching points P0 to Pm are obtained according to the above equation (1).
A robot operation program is created in accordance with the specified interpolation method, moving speed v, and positioning ratio α, and this is transferred to the robot controller of the actual machine. The initial value k
For example, 0, an optimum predicted value, a value obviously larger than the optimum predicted value, or the like can be set to 0 (see Examples described later).

[7] A trial regeneration operation is executed according to the first off-line teaching. Then, during this time, the actual robot trajectory Qo (t) is recorded. As the recording data, for example, current position data corresponding to all interpolation points, current position data corresponding to interpolation points only in the peripheral portion of the teaching point, and the like can be considered.

[8] The deviation between the target robot trajectory Qd (t) and the actual robot trajectory Qo (t) is evaluated, and if not satisfied, the k value is corrected and a robot operation program is created again. The second trial operation was performed and Qd (t)
And the deviation of Qo (t) are evaluated.

If not satisfied again, the k value is recorrected, the robot operation program is created again, the third trial operation is performed, and the deviation between Qd (t) and Qo (t) is evaluated again. To do.

In the same manner, this process is repeated until a satisfactory result is obtained, and a robot motion program having a higher degree of coincidence is determined.

There are various possible methods for evaluating the deviation between the target robot trajectory Qd (t) and the actual robot trajectory Qo (t). For example, the deviation function D (t) = | Qd
A method of defining (t) -Qo (t) | and setting k that minimizes the maximum value of D (t) as the optimum k value, Qd (t) and Qo.
There is a method of finding k that minimizes the area surrounded by (t) (the integrated value of D (t)).

[0027]

FIG. 2 is a block diagram showing the outline of the construction of an off-line programming apparatus that can be used when carrying out the method of the present invention. In the figure, 1 is a microprocessor (hereinafter referred to as CPU), 7 is a ROM storing a control program of the automatic programming device, and 2 is a system program loaded from a floppy disk or a hard disk (hereinafter simply referred to as a disk) 11. , A RAM for storing data representing the work shape, 3
Is a keyboard, 4 is a CRT display device (hereinafter, referred to as CRT), 9 is a tablet device, 8 is a disk controller, 10 is an XY plotter for outputting a drawing,
A printer 5 is connected to each of these elements via a bus 6. An input / output device (I / O) 12 as a general-purpose signal interface is connected to the robot and other external devices.

The tablet device 9 has a screen-corresponding area 9a and a menu table 9b, and the tablet cursor 9c is moved within the screen-corresponding area 9a to move the graphic cursor on the CRT 4 to perform an instruction operation.
Various menu items can be selected from the system program prepared on the disk 11 by designating an arbitrary position on the screen of No. 4 or moving the tablet cursor 9c on the menu table 9b and performing an instruction operation. Has become. Hereinafter, a process of creating a robot operation program according to the present invention using the above-described offline programming device will be described. [1] Work flow until robot operation program creation FIG. 3 shows a flow of work up to the robot operation program creation in the present embodiment in the form of a flowchart, the outline of which is described in the section of "action". It is common with the one described in.

First, the operator operates the tablet device 9 to download the system program of "preparation processing for non-positioning control robot operation program" incorporating the drawing tool from the menu table 9b, and the CPU 1
To start the condition setting screen on the CRT4,
Input and specify the target robot code (step OP
1).

Furthermore, the target robot trajectory setting screen is displayed on the CRT.
4 to specify the target robot trajectory Qd (t) (step OP2). Target robot trajectory Qd (t)
Is specified by reading the shape data of the target work (here, the work to be sealed) from the accumulated data of the disk 11 and applying some processing to it. Here, it is assumed that the work W has the shape shown in FIG. The shape of the work W is assumed to be composed of the following elements.

P1 to P2; straight line P2 to P3; straight line P3 to P4; straight line P4 to P5; spline (illustrated by a curve) P5 to P7; straight line (P6 is a midpoint thereof) Now, as a working line (ceiling line), It is assumed that the routes P1 to P6 are required. Therefore, the cursor, the key, or the mouse is operated on the CPT 4 to extract the section of P1 to P6, and the work line shown in FIG. 4B is generated. In this example, in order to create the robot operation program more smoothly, the bending points P2 and P3 are replaced with quarter arcs Q2 and Q3 of the small radius (automatically processed by an appropriate software. You can also run
You can omit the insertion of the arc.)

The line shown in FIG. 4 (2) is changed to P1.
, P6 which is the starting point (the direction of movement of the robot) is added and input as the target robot trajectory Qd (t).

Next, the teaching point non-setting permissible section length index Lmax, the teaching point setting density index N, and the minute division index Δ described in the section "Action" are designated and input (step OP3).
The values of these various values are selected by design in consideration of required accuracy and the like.

Further, the interpolation method, the interpolation method, the robot moving speed v (teaching speed), and the index α representing the positioning ratio (the deceleration ratio in the acceleration / deceleration control when passing the teaching point) are designated and input (step OP4).

When the above preparation is completed, the robot operation program is automatically created by software processing (step OP5). Hereinafter, the automatic creation process of the robot operation program will be described with reference to the flowchart of FIG. For an overview of the processing contents, see "Action".
It is common with the one described in the explanation column of.

First, the operator operates the tablet device 9 again, downloads the system program of "non-positioning control robot operation program generation process" from the menu table 9b, activates the CPU 1, and starts the process.

When the processing is started, the above step OP2
The dividing points A0 to An for dividing the input target robot trajectory Od (t) by the minute division index Δ are calculated (step S1). Then, tangent vectors <g0> to <gn> are calculated for each division point (step S2). Further, in accordance with the rule described in the "Action" section, the inner product between the tangent vectors is sequentially calculated, and the teaching point generation mediation points B0 to B
m is extracted and determined (step S3). Figure 4 (3) shows
The distribution tendency of the teaching point generation mediation points thus extracted and designated is conceptually represented. The teaching point generation intermediary points indicated by X are sparsely distributed in the straight line portion and densely distributed in the small radius portion. In the relatively gently curved spline portion, the teaching point generation intermediary point is specified with a medium line density. It should be noted that C1 and C2 indicate the positions of the centers of curvature in the respective arc insertion portions, and the arrows indicate the position correction direction when the teaching point is generated from the teaching point generation intermediary point.

In subsequent steps S4 and S5, the above (1)
The coefficient parameter k of the equation is set to an initial value k = 0, and the teaching point generation intermediary point number index j is set to j = 0. Then, in step S6, the curvature κj at the teaching point generation mediation point Bj is calculated according to the formula for calculating the curvature of the trajectory Qd (t), and in step S7, the teaching point Pj is calculated according to the equation (1).
Calculate the position of. The processing of steps S6 and S7 is
It is repeatedly executed for all the teaching point generation mediation points Bm (circulation of steps S5 to S9).

When the teaching points P0 to Pm have been obtained for all the teaching point generation mediating points B0 to Bm, the above step OP is performed.
An operation program for the robot specified in step OP1 is created in accordance with the conditions input in step 4 (step S10). In this embodiment, since the coefficient parameter k is started from 0, the taught trajectory taught by the operation program created the first time is almost the same as the target robot trajectory Od (t).

If a robot operation program has been created for one k value, it is transferred to the robot (robot controller) via the input / output device 12 (step S11), and then the robot is started to teach the program. The regenerating operation is executed on a trial basis (step S1).
2). During this time, the robot controller records the robot trajectory Qo (t) as the movement history of the robot.

When the trial operation of the robot is completed, the recorded data of the robot trajectory Qo (t) is transferred from the robot controller to the RAM 2 of the offline programming device.
To the evaluation function Dk [Qd (t), Qo
(T)] is calculated (step S13). The evaluation function is
As described in the section of “action”, for example, | Qd
It is possible to integrate (t) -Qo (t) | along all trajectories.

After the evaluation function is calculated, the superiority or inferiority of the evaluation function with respect to the previous k value is checked by, for example, the magnitude of the integrated value (step S14). Since there is no comparison target in the first time, the initial value of the evaluation function is set so that the determination in step S14 is unconditionally NO.

In step S15, the k value is updated to k + ε (ε is a small value set in advance), and then, in steps S5 to S5.
Step S14 is repeated. Step S14 of several times
Then, it is determined that the evaluation function has deteriorated and the peak of the coincidence between the trajectories Qd (t) and Qo (t) has passed. Then C
The PU 1 proceeds to step S16, determines the optimum k value, calculates the equation (1) based on it, and determines the optimum teaching point.

There are various possible ways to determine the optimum k value.
For example, the previous k value, the current k value, or the arithmetic mean value of the two can be taken. There is also a method of selecting, for example, the midpoint between the previously calculated teaching point position and the currently calculated teaching point position without determining the k value. When the optimum teaching point position is determined in this way, a robot operation program is created based on it, and the program is registered in the disk 11 together with the program name and the processing is terminated.

Although one embodiment has been described above,
The method of creating the robot operation program of the present invention is not limited to this. For example, as a method of evaluating the curvature in the vicinity of the division points of the target robot trajectory, there is a method of using the inner product of the normal vector other than the method of using the inner product of the tangent vector.

Further, the process of extracting the teaching point generation intermediary points from the division points can be modified, and the position of each of the teaching point generation intermediary points is determined according to the curvature in the vicinity of each of the teaching point generation intermediary points. There are various possible ways to modify. For example, there are modifications such as a method of determining an optimum k value for each curvature level, and a method of checking only the evaluation function value at a specified specific point to determine the k value.

[0047]

According to the present invention, teaching point generation intermediary points are set at a density corresponding to the curvature transition along the target robot trajectory, and the teaching point generation intermediary point position is corrected and shifted according to the curvature at each position. Since the teaching point is determined in the form of
It is possible to reduce the deviation between the actual robot trajectory and the target robot trajectory during non-positioning control.

Further, while changing the correction shift amount from the teaching point generation intermediary point position, the trial operation of the created robot operation program is performed, and the robot trajectory recorded during the trial operation is compared and evaluated with the target robot trajectory. , It is now possible to create a robot motion program that realizes a robot trajectory with a high degree of agreement with the target robot trajectory during non-positioning control.

[Brief description of drawings]

FIG. 1 is a conceptual diagram for explaining in principle a specific example of a teaching point generation process adopted in the present invention.

FIG. 2 is a principal block diagram showing an outline of the configuration of an off-line programming device that can be used when performing the method of the present invention.

FIG. 3 is a flow chart showing the flow of work up to the creation of a robot operation program in this embodiment.

FIG. 4 (1) is a CR of an off-line programming device
Shows the shape of the work W displayed on the T screen, and (2)
Is the target robot trajectory Qd obtained by processing this
(T) is represented. Further, (3) is a diagram conceptually showing the distribution of the teaching point generation intermediary points set in the target robot trajectory Qd (d) and the position shift direction at the time of creating the teaching points.

FIG. 5 is a flowchart showing an outline of processing contents of robot operation program automatic creation processing during non-control.

[Explanation of symbols]

1 Microprocessor (CPU) 2 RAM 4 CRT display device 6 Bus 7 ROM 8 Disk controller 9 Tablet device 11 Disk (floppy disk or hard disk) 12 Input / output device A0-An division point B0-Bm Teaching point generation mediation point C1, C2 Curvature center of arc portions Q2 and Q3 Qd (t) Target robot trajectory W Work (shape) <gi> Tangent vector <hj> Normal vector

Continuation of the front page (51) Int.Cl. ⁶ Identification number Office reference number FI technical display location G05B 19/4103 G05B 19/42 P

Claims (2)

[Claims] 1. A step of: (a) designating a target robot trajectory on a robot motion program creating device; and (b) defining division points for dividing the target robot trajectory into a large number of minute sections, in the vicinity of each division point. Extracting a teaching point generation intermediary point from the division points at a density according to the curvature, and (c) modifying the position of each teaching point generation intermediary point according to the curvature in the vicinity of each of the teaching point generation intermediary points And the step of: (d) creating a robot operation program including the specified teach point data. 2. A step of: (a) designating a target robot trajectory on a robot motion program creating device; and (b) defining division points for dividing the target robot trajectory into a number of minute sections, and in the vicinity of each division point. Extracting a teaching point generation intermediary point from the division points at a density according to the curvature, and (c) modifying the position of each teaching point generation intermediary point according to the curvature in the vicinity of each of the teaching point generation intermediary points And (d) creating a robot operation program including the specified teach point data, and (e) performing a replay operation of the created robot operation program by at least a replay robot trajectory. And (f) evaluating the deviation between the regenerated robot trajectory and the target robot trajectory, the steps (c) to (f) being performed. A method of creating a robot operation program, wherein the steps are repeated until the favorable evaluation result is obtained in the step (f) while changing the correction amount in the step (c).