JPH0833762B2 - Learning control method for industrial robots - Google Patents

Description

TECHNICAL FIELD The present invention relates to a learning control method for an industrial robot, a repetitive positioning apparatus, or the like.

[Prior Art] Conventionally, a controlled object is reproduced according to a teaching value stored in a memory, a deviation between a target value and a reproduction locus is measured at the time of the reproducing operation, and the deviation is taught at the time of the next reproducing operation. A learning control method for an industrial robot is known in which a teaching value for performing a reproduction operation on a controlled object is sequentially corrected by adding the value to a value to improve the trajectory accuracy of the reproduction operation.

According to this learning control method, the teaching value is sequentially corrected according to the deviation between the target value and the actual reproduction trajectory,
The teaching value converges within the allowable error range of the target value as the reproducing operation is repeated several times. For this reason,
The work locus of the regenerating operation can be controlled according to the target value.

[Problems to be Solved by the Invention] However, in general, an arm or the like to be controlled in an industrial robot or the like has a certain delay time from being given a target value until it actually starts operating. Therefore, as described above, in the method of correcting the teaching value used next time by the deviation of the reproduction locus with respect to the teaching value given at the current time,
Since the teaching value is corrected by the deviation of the previous time by the delay time, it is necessary to repeat many playback operations until the teaching value converges within the allowable error range of the target value, and learning efficiency is improved. There is a problem of being bad.

The present invention has been made to solve such problems, and an object of the present invention is to provide a learning control method for an industrial robot that can improve learning efficiency.

[Means and Actions for Solving Problems] The present invention measures a delay time from when a teaching value is given to when a controlled object starts to operate, and a reproduction locus at a time delayed by a time corresponding to this delay time. And the target value,
Based on this deviation, the teaching value for the next reproducing operation is corrected.

[Embodiment] FIG. 1 is a block diagram showing an embodiment of a control device for an industrial robot to which the present invention is applied. Position information of each point on the motion locus of an arm (not shown) to be controlled. Target value memory 1 for storing the target value as a target value, teaching value memory 2 for storing the target value corrected by the teaching correction as a teaching value, and arm position information for the target value detected by position detector 3 as a reproduction value. The reproduction value memory 4 to be stored and the deviation between the reproduction value and the target value stored in the reproduction value memory 4 are obtained, and the content (teaching value) of the teaching value memory 2 is corrected by the deviation, and the target value memory 1 or The actuator 6 is read by reading the stored contents of the teaching value memory 2.
And an arithmetic unit 5 for operating the arm via the.

FIG. 2 is a flow chart showing the procedure for performing the teaching correction. First, the delay time of the arm is measured in the steps 10 to 15, and the teaching value of the teaching value memory 2 is set to the delay time of the arm based on the measurement result. It is corrected by the deviation between the reproduction value and the target value in consideration of.

In this case, the teaching value memory 2 initially stores the same position information as the contents of the target value memory 1 by the teaching work by the operator.

First, in step 10, the target value of each point on the motion locus stored in the target value memory 1 is read by the arithmetic unit 5, and each target value is given to the actuator 6, whereby the arm is set to each target value. Driven correspondingly.

At this time, the actual operation position is detected by the position detector 3 for each target value and sequentially stored in the reproduction value memory 4 as a reproduction value (step 11). Then, when the reproduction value for the current target value is obtained, a feedback calculation for adding the deviation to the next target value is performed (step 12),
The next target value is given to the actuator 6 by this calculation result. As a result, the locus of the arm changes according to the target value stored in the target value memory 1.

When the first reproduction operation with all target values is completed in this way, the arm operation start time is detected depending on which address the reproduction value is stored in the reproduction value memory 4 next (step 14).

That is, the target value memory 1, the teaching value memory 2, and the reproduction value memory 4 are designed so that their memory addresses are designated by the same address signal that changes sequentially in a predetermined cycle. For example, the reproduction value for the target value read from the target value memory 1 is
Is stored at the same address as the target value. But,
When the arm operation start time is delayed, it is stored at an address deviated by an amount corresponding to this delay time.

Therefore, the delay time of the arm can be detected by detecting the shift amount of this address.

If the arm delay time can be measured in this way, the teaching value stored in the teaching value memory 2 is corrected by the deviation between the reproduction value at the time delayed by the arm delay time and the target value. To be done.

That is, the delay time is T and the target value at time t is M
₀ (t), a reproduction value at time t M2 (t), the teachings value at time t M1 (t), if the K a constant, M1 (t) '= M1 (t) + K {M 0 (t + T ) -M2 (t +
T)} corrects the teaching value M1 (t) in the teaching value memory, and the correction result M1 (t) 'is used as the teaching value corrected by the delay time of the arm.

Next, when such correction is completed for all the teaching values at each time, the corrected teaching values are sequentially read thereafter, and the deviation from the reproduction value detected by the position detector 3 is obtained. The feedback calculation of adding the deviation to the teaching value is performed for the teaching values at all the operating positions (steps 16 to 19). If the motion locus of the arm driven by the resulting teaching value is not within the allowable error range, the processing returns to steps 16 to 19 again, the teaching value is corrected by teaching, and the allowable error range is returned. If it converges within, the teaching correction is finished.

[Effects of the Invention] As is apparent from the above description, in the present invention, since the teaching value is corrected by teaching in consideration of the actual delay time of the controlled object, the number of teaching corrections is reduced, and learning is performed. The efficiency can be improved.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of a control device for an industrial robot to which the present invention is applied, and FIG. 2 is a flow chart showing a procedure for teaching correction. 1 ... Target value memory, 2 ... Teaching value memory, 3 ... Position detector, 4 ... Reproduction value memory, 5 ... Computing device, 6 ...
Actuator.

Claims (1)

[Claims] 1. A teaching value stored in a teaching value memory is sequentially read to perform a reproducing operation on a controlled object, and the reproducing value is sequentially stored in a reproducing value memory to obtain a deviation between a target value and a reproducing value. In a learning control method for an industrial robot in which a deviation is added to the teaching value in the next reproducing operation to sequentially correct the teaching value, in the reproducing operation, the teaching value is changed by an address signal that sequentially changes in a predetermined cycle. By sequentially designating the addresses of the same number in the memory and the playback value memory, the teaching value is sequentially read from the first address, and the playback value is sequentially stored from the address corresponding to the time when the controlled object starts the operation. The first address number of the teaching value memory where the reading of the value started and the address number of the playing value memory where the storing of the reproduced value started Based on the above, the delay time until the controlled object starts to operate is calculated, and the teaching value at the time of the next playback operation is deviated from the playback value at the time delayed by the time corresponding to the calculated delay time and the target value. A learning control method for an industrial robot, which is modified based on the above.