JPH0766291B2 - Linear time-series command signal generation method - Google Patents

Description

TECHNICAL FIELD OF THE INVENTION The present invention relates to a linear time-series command signal generating method for servo control of an industrial robot.

[Conventional technology]

The conventional time-series command signal is generated by using floating point arithmetic. In the case of a straight line shape, it is calculated by the formula Y = αt + B, and is converted into an integer when Y is output. It was However, this requires floating-point arithmetic, which requires an expensive processor to be installed and an expensive processor with high processing speed.

On the other hand, as a method of generating the linear time-series command signal only by integer calculation, it is conceivable to use the following technique used when drawing on the XY plane. For example, linear interpolation is used for linear positioning control of a motion path. The principle of this linear interpolation is, for example, XY
This is done by distributing the pulses on the X-axis and the Y-axis on the plane according to the inclination of the straight line with the passage of time.

Nobuo Saida, the presenter of "Creating a new functional concept" in Yasukawa Denki Technical Report (Vol. 147), has introduced a simple linear interpolation method. The linear interpolation method uses X with high pulse density.
Using the axis as the reference for pulse distribution, the other side Y
By sequentially determining the presence or absence of the axis pulse, the command pulse train signal corresponding to the straight line having the predetermined inclination is sequentially distributed.

By the way, when a time-series command pulse is generated for servo control of an industrial robot, a command pulse is generated by interpolating an X-axis or a Y-axis to be controlled for each sampling cycle with a time axis as a reference.

FIG. 1 is a straight line passing from the origin (0, 0) to the point P (T ₀ , Y ₀ ) with the time T on the horizontal axis and the control amount Y on the vertical axis based on the linear interpolation method. Is shown.

Here, the equation of the straight line is expressed by the following using the slope (Y ₀ / T ₀ ).

Y = (Y ₀ / T ₀ ) T Therefore, the above equation can be defined as follows.

T ₀ Y = Y ₀ T = L Next, FIG. 2 shows the distribution status of pulses on the time axis as pulse intervals T ₀ and Y ₀ .

At a certain time point T = i, L = Y ₀ i, and in that L, T
_{If 0} is included in the number j, Y = j. Therefore, Y ₀ is added every time the pulse interval (time) T ₀ is incremented by one unit for each sampling period on the time axis, and by counting how many T ₀ are included in it, , The target straight line is obtained.

FIG. 3 shows a procedure for obtaining points P ₁ , P ₂ , P ₃ ... On the straight line. The points P ₁ , P ₂ , P _3, ... On the straight line can be sequentially calculated by a loop operation.

[Problems of conventional technology]

In the case of such control, when the time axis becomes the axis of high pulse density, in other words, when the direct slope (Y ₀ / T ₀ ) becomes larger than 1, the loop calculation for each time step is completed once. (Y ₀ / T ₀ ) times, the calculation time becomes longer as the inclination becomes larger, the quick calculation at each sampling cycle becomes difficult, and the motion speed of the controlled object is also hindered.

[Object of the Invention]

Therefore, an object of the present invention is to propose a method for efficiently generating a linear-shaped time series command signal by using only integer arithmetic without using floating point arithmetic and without loop arithmetic.

[Means for Solving the Invention]

Based on the above object, the present invention, in obtaining a linear time-series command signal based on a linear equation, divides the linear equation into a quotient and a remainder by division in advance, and the remainder term has a small slope with a small slope. It is defined as a linear expression, and the quotient is added for each sampling period, and the pulse distribution amount can be obtained with only one calculation of the remaining term, that is, without loop calculation.

That is, the linear time-series command signal generation method of the present invention,
In order to obtain a linear time series command signal of the linear equation Y = (Y ₀ / T ₀ ) T, the quotient Q ₀ = Y ₀ / T ₀ and the remainder R ₀ are substituted, and the above equation is Y = Q ₀ as _{T + (R 0 / T 0} ) T, for each sampling period of T side, by examining the presence or absence of T ₀ at Y side, when T ₀ there, Y _n
= Y _n-1 + Q ₀ +1 and without T ₀ , Y _n = Y _n-1 +
By outputting the signal of Q ₀ , a linear time series command signal is generated.

Such time-series command generation processing is performed for each axis to be controlled. In addition to the Y-axis, it is executed as the required amount of rotation of the X-axis, Z-axis, joint axis of the controlled object and the like.

[Details of Invention]

Similar to FIG. 1, FIG. 4 shows the origin (0, 0) and the point P.
A straight line passing through (T ₀ , Y ₀ ) is shown.

As described above, the above equation of the straight line is written as follows.

Y = (Y ₀ / T ₀ ) T Here, T is the number of time-series steps, and Y is a command signal to be generated.

Further, if the division Y ₀ ÷ T ₀ = Q ₀ , that is, the quotient is Q ₀ and the remainder at that time is R ₀ , the above straight line equation can be defined as follows.

Y = Q ₀ T + (R ₀ / T ₀ ) T Here, the previous term corresponds to counting the value of Q ₀ for each sampling cycle. Further, the remaining latter term is obtained by the conventional linear interpolation method.

After all, as shown in FIG. 4, the equation of the straight line to be obtained is expressed as the sum of the equation of the straight line Y = Q ₀ T and the equation of the straight line of the remainder term y = (R ₀ / T ₀ ) T. . Here, since the remainder R ₀ is smaller than the pulse interval T ₀ , the linear expression y is always a linear expression with a small inclination. Therefore, y is always obtained by one calculation without a plurality of loop calculations.

FIG. 5 shows the situation of pulse distribution between y and T.

Next, FIG. 6 shows a flowchart of pulse distribution.

At the beginning of the start, the initial value D is set to 0, and the current value D is obtained from the previous value D by the following calculation.

D = D−T ₀ + R _{0 The} obtained value D is compared with 0, and when it is equal to or larger than 0, the increment dy = 1 is set, and the control amount Y is calculated by the following calculation. Desired. Where the code S ₀
Is given as "1" and is set to + when the slope is positive, and is set to-when the slope is negative.

Y _n = Y _n-1 + Q ₀ +1 Further, as a result of the comparison, when the value D is smaller than 0,
The following operations are performed in the next two steps.

dy = 0 D = D + T ₀ Y = Y + Q ₀ Then, the value of any D is advanced to the next step.

In this way, the pulse distribution amount for the Y axis is obtained for each sampling cycle with the time axis as a reference. Of course, the above calculation is sequentially repeated at every sampling cycle.

As is apparent from this flowchart, only one calculation is always performed in the sampling cycle, and a plurality of loop calculations as in the conventional art are not required.

Such control is executed as the amount of rotation of the control target Y-axis, and other necessary control target axes X-axis, Z-axis, joint axis θ, and the like. Therefore, Y in the claims is
It should be interpreted as X, Z, θ, etc.

〔The invention's effect〕

In the present invention, since the pulse density of the axis on the base side, that is, the time axis is always lower than that of the axis on the output side (command signal), the pulse distribution amount can be obtained without loop calculation. Therefore, a command pulse can be generated in a short time even in a servo system having high pulse resolution and high speed. Further, in the pulse calculation process, the necessary pulse distribution can be performed only by a simple addition / subtraction calculation, and the pulse is always generated on the high-density axis side, so that a smooth motion path can be obtained.

If such a linear signal generation method is used as a command generator of a servo control system, a real number operation in real time becomes unnecessary, and the execution time becomes extremely short, so a special operation chip is not required, In addition, the sampling cycle is shortened, and smooth path control is possible.

[Brief description of drawings]

FIG. 1 is a graph of a linear interpolation method, FIG. 2 is an explanatory diagram of pulse distribution of a linear interpolation method, FIG. 3 is a flow chart diagram, FIG. 4 is a graph of linear interpolation of the present invention, and FIG. FIG. 6 is a flow chart of a linear time-series command signal generating method according to the present invention.

Claims (1)

[Claims] 1. When obtaining a linear time-series command signal of a linear equation Y = (Y ₀ / T ₀ ) T, the above equation is replaced by the quotient Q ₀ = Y ₀ / T ₀ and the remainder R _0. Assuming that Y = Q ₀ T + (R ₀ / T ₀ ) T, the presence or absence of T ₀ is checked on the Y side for each sampling cycle on the T side,
When T _{0 is} present, Y _n = Y _n-1 + Q ₀ +1 is set, and when T ₀ is not present, a signal of Y _n = Y _n-1 + Q ₀ is output. Time series command signal generation method.