JPH01205207A - Off-line teaching system - Google Patents

Abstract

PURPOSE:To improve a work efficiency while the actual orbit of a robot is approached to a target orbit by executing the learning control once or plural times and successively preparing new teaching data. CONSTITUTION:A computer 3 prepares new teaching data from the position feedback data of a robot 8 obtained based on the teaching data and next, successively prepares further new teaching data from the position feedback data of the robot 8 obtained based on the new teaching data. A so-called learning control system is adopted. Consequently, the actual orbit of the robot 1 can be made close to the target orbit. Thus, the highly accurate teaching data can be easily obtained and the work efficiency can be improved.

Description

[Detailed Description of the Invention] "Field of Industrial Application" This invention relates to industrial robots, and in particular to an offline teaching system that is capable of obtaining highly accurate teaching data by adopting a learning control method. be.

``Prior art J'' Conventionally, in this type of offline teaching system, a computer (computing device) installed separately from the main body of the robot was used.
The teaching data is created as appropriate.

That is, in such an offline teaching system, workpiece shape data indicating the shape of the workpiece is stored in advance, and further, based on the workpiece shape data, the operator appropriately inputs teaching data for operating the robot (
For example, an operator manually uses a tablet or the like while looking at a display device.

Based on the teaching data input to the computer, the robot is made to repeatedly perform the same motion.

"Problem to be solved by the invention" By the way, in the offline teaching system as described above,
The robot is made to perform motions based on teaching data created on a computer, but in reality, the robot's actual trajectory data as a result of such motions and the corresponding target trajectory are The teaching data is quite isolated.

This is because there are differences in the load, servo gain, etc. of the robot, and as a result, the actual trajectory of the robot deviates from the target trajectory indicated by the teaching data, making it impossible to obtain the desired actual trajectory. could not.

If such a difference occurs between the target trajectory and the actual trajectory, the line on which the robot is placed is stopped for a long time, and the teaching data is corrected using the robot's control panel, for example. , This resulted in a decrease in work efficiency.

This invention was made in view of the above circumstances, and a computer creates new teaching data from robot position feedback data obtained based on the teaching data, and then uses the new teaching data to create new teaching data. A so-called learning control method is adopted in which new teaching data is sequentially created from the robot's position feedback data obtained based on the robot's position feedback data, which allows the robot's actual trajectory to approximate the target trajectory with high accuracy. It is an object of the present invention to provide an offline teaching system in which teaching data can be easily obtained and work efficiency is improved.

"Means for Solving the Problem" In order to achieve this purpose, an offline teaching system that reproduces the robot based on teaching data indicating a target trajectory created in a calculation device provided separately from the robot. , a data collection function that samples position feedback data during robot playback from a control panel provided on the robot; the position feedback data sampled by this data collection function; and teaching as a target trajectory created by the calculation device. The arithmetic unit is provided with a teaching data organizing function for creating new teaching data based on the teaching data.

"Operation" According to the present invention, first, when a robot is driven based on teaching data indicating a target trajectory created in a calculation device, data indicating an actual trajectory obtained by driving the robot, that is, position feedback. The data is sampled as appropriate, and furthermore, the arithmetic unit sequentially creates new teaching data based on the sampled position feedback data and the teaching data indicating the target trajectory.

Then, based on this new teaching data, position feedback data indicating the actual trajectory obtained when driving the robot is sampled as appropriate, and the sampled position feedback data and teaching data indicating the serious crime target trajectory are combined. Based on this, the arithmetic unit sequentially creates new teaching data.

Then, by performing such learning control once or several times and sequentially creating new teaching data, the actual trajectory of the robot can be brought closer to the target trajectory.

"Embodiment" An embodiment of the present invention will be described with reference to FIGS. 1 to 5.

First, what is indicated by the symbol M in FIG. 1 is the entire computer system, and what is indicated by the symbol N is the entire robot system.

The computer system M includes a display device IA.
, a computer 3 consisting of a keyboard IB, a computing device 2, etc., a tablet 4 as an input device, a printer 5 and a block 6 as recording devices, and a floppy disk device as a storage device, each connected to the computing device 2. It is composed of 7.

Then, the computer system M creates teaching data that will become the basic target trajectory, as indicated by "o" in FIG. 3 (described later).

The robot system N also includes a robot 8 that performs work such as painting on a workpiece W by servo-controlling an arm 8B on a base 8A and a working device (path shown) around six axes, and the robot Robot control panel 9 that controls the drive of 8
The robot control panel 9 is composed of:
The teaching information (teaching data, etc.) stored in the floppy disk device 7 is transmitted via the floppy disk D (to the robot control panel 9).
The function will be described later in step (2) below. ) Then, the robot system N can obtain actual trajectory data (position feedback data) based on the teaching data as shown by "X" in FIG. 3 (described later).

In FIG. 3, the horizontal axis indicates time t1, and the vertical axis θ1 indicates the positions of points "○" and "x", for example, the coordinates of the i-axis.

Furthermore, in the robot system N, the arrangement of the robots 8 is not limited to one, but a plurality of robots may be arranged.

Next, the functions of the computer system M1 and the robot system N will be explained in detail with reference to FIGS. 2 to 5. Step S shown in the following description corresponds to the flowchart in FIG.

(1) In the computer system M, the tablet 4
The workpiece shape data indicating the shape of the workpiece W is recognized in advance by input from , and further, based on the workpiece shape data, the operator creates a basic target trajectory for operating the robot 8 (described later in (2)). ) is input (for each point) using the tablet 4 or the like while looking at the display device 1, for example. The teaching data inputted for each of these points is indicated by "o" in FIG.

Further, the teaching data is written by the arithmetic unit 2 to the area E of the floppy disk D in the floppy disk device 7 (step 2).

(2) The floppy disk D is transferred to the floppy disk device 9A in the robot control panel 9 on the robot system N side (step 2), and the teaching data is read from the area E1 described above.

This teaching data is stored in a teaching data storage memory shown by reference numeral 9B in FIG. (hereinafter,
The teaching data (1) interpolated by this interpolation calculation means 9C will be referred to as the basic target trajectory V). The interpolation period is generally smaller than Ts and the third waveform is virtually continuous in time.
It looks like a solid line connecting the teaching points "○" shown in the figure.

Then, based on the basic target trajectory V created by the interpolation calculating means 9C, the robot driving means 9D performs servo control to drive the robot 8 along six axes (step 3).

Then, the locus of the actual trajectory obtained as a result of the driving of the robot 8, that is, the position feedback data, is written into the position feedback storage memory 9F at regular intervals Ts by the sampling means 9E, as shown by "X" in FIG. , furthermore, an area E of the floppy disk D,
will be written to.

Note that the period Ts of the points indicating the actual trajectory is set to be approximately the same as the dynamic delay time of the robot 8.

Also, the actual trajectory data to be written into the feedback storage memory 9F is stored in the feedback storage memory 9F.
When the capacity of F is exceeded, the files are automatically sent sequentially to the floppy disk device 9A.

(3) The actual trajectory data written on the floppy disk D is transferred to the floppy disk device 7 on the computer system M side and read out (step 4).
), and a certain calculation is performed in the calculation device 2 (
Step 5).

First, the teaching data is read out from the area E1 of the floppy disk D, and interpolated and calculated as shown in FIG.
Target trajectory data is generated at regular intervals TS, as shown by "." Next, area E of floppy disk D,
The position feedback data is read out at a constant period T.
The target trajectory “mouth” and the position feedback data “×
", and trajectory data (new teaching data) consisting of points "." and set at the period indicated by Ts is created (step 7).

Note that steps 6 and 8 will be described later.

Next, the calculation rule performed in the above step (3) will be explained.

First, let the teaching data as the basic target trajectory set in process (1) and process (2) be θ'd(k),
Assuming that the position feedback data as the actual trajectory obtained in step (2) is θ1(k), these teaching data θ'd(k) and position feedback data θ1(
k), trajectory data θ'v(k) is calculated based on the learning control algorithm below.

However, in the above formula, i; axis number (i = 1 = ie, in the case of 6 axes, i
= 1 to 6) k; Point number (k = 0 to ke, where point number ke indicates the total number of points in one trajectory data group.

) Furthermore, Ts-ke indicates the playback time based on the teaching data (trajectory data).

j; number of calculations Kj; correction calculation coefficient In the above learning control algorithm, k=0, k=1, =
2. = 3, ..., k = ke, the teaching data points indicating the basic target trajectory (indicated by circles in Figure 4) and the position feedback data points indicating the actual trajectory (Figure 4) (indicated by x), and further calculate these differences using the reference timing as the starting point (first, starting from k=o, then k=1.2, etc.).
, ke), multiplied by a constant Kj, and sequentially added for the number of times indicated by je, and the value obtained thereby is added to the teaching data as the basic target trajectory at each timing.

Then, by using such an algorithm, "
As shown in ``, k=0, k=1, k-2, k=3,
..., at the timing shown by k=ke, the robot 8
The points as trajectory data obtained by approximating the actual trajectory to the basic target trajectory are k=0, k=1 . ni = 2, k = 3
It can be calculated as appropriate in the order indicated by .

That is, by performing such learning control once or several times, it is possible to sequentially create trajectory data that approximates the actual trajectory of the robot 8 to the target trajectory (
Step 7).

Note that from the second time onwards, the first term θ'd(k) on the right side of the equation of the learning control algorithm uses trajectory data from − times before.

In addition, in the above learning control algorithm, the number of operations j
e and the correction calculation coefficient Kj are appropriately determined based on the individual dynamic characteristics of each robot 8.

In addition, when creating the trajectory data indicated by "・", the error average Ei (hereinafter referred to as (calculated by the formula shown in ) becomes less than or equal to a certain constant (step 6) (loop shown in steps 2 to 7).

Ke kJ Then, when the error average Ei calculated by the above formula becomes less than or equal to the constant , if the loop shown in Steps 2 to 7 is not passed, the teaching data created in Step 1 is set as is (Step 8).

In creating such trajectory data, the trajectory data and position feedback data are received and passed through the floppy disk D, as shown in steps 2 and 4.

In addition, in the offline teaching system described above, teaching data, trajectory data, and position feedback data are exchanged with the robot stem N via the floppy disk D, but the present invention is not limited to this. Data may be received and delivered by communication means using infrared rays, electromagnetic waves, or the like.

As explained above, according to the offline teaching system described above, first, when the robot 8 is driven based on the teaching data (see ○ in FIG. 3) indicating the target trajectory created in the arithmetic device 2, Position feedback data obtained by driving the robot 8 (see x in Figure 3)
Further, based on the sampled position feedback data and the teaching data indicating the basic target trajectory, the arithmetic unit 2 generates the trajectory data (new teaching data) (as shown in FIG. 4). reference) are now created sequentially.

Then, based on the trajectory data, position feedback data indicating the actual trajectory obtained when the robot is driven is sampled as appropriate, and the sampled position feedback data, teaching data indicating the target trajectory, and the basic goal are The arithmetic device sequentially creates new trajectory data (new teaching data) based on the trajectory.

In other words, by performing such learning control once or several times and sequentially creating new trajectory data, it is possible to bring the actual trajectory of the robot closer to the basic target trajectory.
This allows extremely highly accurate teaching data to be obtained.

Furthermore, since the new trajectory data can be created by performing learning control several times, the time required to stop the line is substantially reduced, and work efficiency can be improved.

In addition, the creation of trajectory data using such learning control is
Since it can be performed for each robot, the load on the robot,
The influence of dynamic characteristics due to differences in servo gain etc. can be eliminated.

In addition, when creating trajectory data by learning control as described above, the minimum functions required for the robot control panel 9 are a function to read teaching data from the floppy disk D, a function to sample position feedback data, and a function to read teaching data from the floppy disk D.
Since the function is to store the position feedback data on the floppy disk D, the normally used robot control panel 9 can be fully applied.

Further, the robot control panel 9 is provided with a memory for temporarily storing sampled position feedback data, and the storage capacity of this memory is such that the position feedback data is sequentially stored on the floppy disk D. Alternatively, it is possible to reduce the number of data by sequentially transferring them to a computer using a communication means.

"Effects of the Invention" As explained in detail above, according to the present invention, when the robot is driven based on the teaching data, the actual trajectory data indicating the actual trajectory obtained by driving the robot, that is, the position The feedback data is sampled as appropriate, and furthermore, the arithmetic unit sequentially creates new teaching data based on the sampled position feedback data and the teaching data indicating the target trajectory.

By performing such learning control once or several times and sequentially creating new teaching data, it is possible to bring the robot's actual trajectory closer to the target trajectory, which results in extremely highly accurate teaching. data can be obtained.

Further, since the new teaching data can be created by performing learning control several times, the time required to stop the line is substantially reduced, and work efficiency can be improved.

In addition, the creation of teaching data using such learning control is
Since it can be performed for each robot, the load on the robot,
The influence of differences in servo gain etc. can be eliminated.

[Brief explanation of the drawing]

FIGS. 1 to 5 are diagrams showing one example of the uJ of the present invention,
Figure 1 shows the entire offline teaching system, Figure 2
The figure is a block diagram showing the robot control panel, Figure 3 is a graph showing the relationship between teaching data as a target trajectory and actual trajectory data, and Figure 4 is a graph showing the relationship between teaching data as a target trajectory, actual trajectory data, and learning control. FIG. 5 is a graph showing the relationship with new teaching data created by . FIG. 5 is a flowchart for creating new teaching data. M...Computer system (teaching data organization function) N...Robot system (data collection function) 8...Robot,

Claims (1)

[Scope of Claim] In an offline teaching system that reproduces a robot based on teaching data indicating a target trajectory created in a calculation device provided separately from the robot, the robot is regenerated from a control panel provided in the robot. A data collection function that samples position feedback data during playback; and a new method for the calculation device based on the position feedback data sampled by this data collection function and teaching data as a target trajectory created by the calculation device. 1. An offline teaching system comprising a teaching data organizing function for creating teaching data.