JPS58169203A - Adaptive control system of traveling object - Google Patents

Abstract

PURPOSE:To permit a traveling object to follow up an indicated track without any deviation by varying the status feedback gain of a compensator according to the position of the body to be controlled and providing a system with desired characteristics all the time. CONSTITUTION:A gain adjusting part 23 inputs a displacement angle signal thetak read from the encoder 9 of a motor system 8, timer signal kDELTAT (DELTAT; sample period) from an arithmetic processor 24, detection signal Lk from a load detector 21 fitted on a robot, and status feedback gain value fx stored in a gain storage part 22 to calculate a status feedback gain f(kDELTAT) which is expected at time kDELTAT by linear interpolation, thereby setting it as the status feedback gain of the compensator. Thus, the compensator 2 is adjusted automatically.

Description

Detailed Description of the Invention ta+ Technical Field of the Invention The present invention relates to a control method for a moving object, and in particular, compensation for a control system such as a robot control system whose control characteristics change due to the movement of a controlled object or the application of a load. Adjustment method of device state feed bank gain (b) Conventional technology and problems Conventionally, in control systems where the control characteristics of the controlled object change depending on movement or load, such as control systems for robots and material handling, the control circuit In most cases, the state feedbank gain of the compensator has been fixed to a constant value due to implementation constraints. Therefore, as the control characteristics change, the characteristics of the closed-loop system deviate from the originally designed values, resulting in poor coordination and follow-up to command values, which may lead to a decline in work quality and problems.

In such cases, it is desired to develop a system that can change the gain of the compensator while identifying the state of the system one by one. However, it has not yet come to fruition.

(C1 Object of the Invention The object of the present invention is t) In order to make the moving object follow the indicated trajectory without deviation even if the characteristics of the controlled object change, the state feed bank gain of the compensator is adjusted according to the position of the controlled object. An object of the present invention is to provide an adaptive control method for a mobile body in which a closed-loop system always has desired characteristics by changing the characteristics one by one.

(di) Structure of the Invention In order to achieve these objects, the present invention first assumes that the control characteristics of a moving body are linear with respect to the position or displacement angle within a certain range. Then, the state fever dock gain of the compensator is set in advance, and the value is stored in the gain storage section.Then, the gain adjustment section adjusts the gain based on the position or displacement angle of the moving object and the information from the load detector. is determined by linear interpolation, and the gain is adjusted according to changes in control characteristics.

(el Embodiments of the Invention In the following, as specific embodiments of the present invention, an example of a robot control system will be explained with reference to FIGS. I will explain,
In reality, a system similar to this is constructed for each axis.

Figure 1 shows the mobile adaptive control system (
Fig. 2 is a curve diagram showing the relationship between the deviation angle and feedback gain of the control system, and Figs. 3 and 4 are gain memories of the control system, respectively. 3 is a flowchart showing a storage method in the gain adjustment section and an arithmetic algorithm in the gain adjustment section.

In FIG. 1, 1 is a function generator, 2 is a compensator, an integrator 31, a mechanical system state observer 4. Motor system condition observation device 5. It consists of an integrator 6, and 7 indicates a state feedback gain.

Furthermore, 8 is a motor system, 9 is an encoder, IO is a mechanical system, 1
1 is an acceleration needle, and the above configuration is the same as the servo control system 12 previously proposed by the applicants of the present application in Japanese Patent Application No. 101847-1983.

In this embodiment, the servo control system 12 further includes a negative detection section 21. Gain storage section 22. A gain adjustment section 23 was provided. Note that 24 is an arithmetic processing unit.

As is well known in robots, the length and posture of the arm.

Alternatively, the control characteristics change depending on the weight of the load gripped by the hand. Therefore, it is assumed that the control characteristics are linear within a range in which the displacement angle monotonically changes (this assumption generally holds). Select points A, B, C... (see ta+ in Figure 2) on the indicated trajectory at intervals where this assumption holds,
Similar to the teaching work of coordinate points, the robot is made to learn the coordinates of these points, and the control characteristics of the robot at these points are measured in advance. Here, the mechanical system of the robot G01s) -Y, (sl / Y, (sl
is the second-order vibration system G, +S) = Kω. s”/ (s2
+2ζω. S+ω:) However, Knigain; ζ:
Attenuation constant ω. : Assuming that it can be approximated by the transfer function of the natural angular frequency, and applying the above model to the data measured at each point, K, ζ
, ω0 are obtained as shown in FIG.

This compensator's state feed hack gain f^, "p.

fe... is time t3. on the indicated trajectory. Since the deviation angle θ is closely related to the presence or absence of load α, f8 can be expressed as r(t':
', θtsu, self) along with the time and coordinates as shown in FIG. 3 are stored in the gain storage unit 22. Here, the subscript x means alphabets A, B, (1,...), and f ('') is C7y+ far, 7131711
It is a hector consisting of 11 rMs'.

The above values are obtained in advance off-line using a measuring device, a computer, etc.

With the above, preparations for actually using the control system of this embodiment have been completed.

In this embodiment, the gain adjustment unit 23 receives input information, that is, the koji position angle signal read from the encoder 9 (or potentiometer) of the motor system 8 is OK.

ΔT (ΔT: sample period) is applied to the timer signal from the arithmetic processing unit 24, and the load detector 21 installed on the robot
and the state feed pack gain f stored in the gain storage section 22.

The state feedback gain f(kΔT) (see FIG. 2(e)) that should be taken at time ΔT is calculated and output by linear interpolation from the value of ΔT, and this is set as the state feedback gain of the compensator 2. The compensator 2 is automatically adjusted by adjusting the throat. Its basic formula is: between point α and point β, f(kΔ'r)-(, + (to θ-~)/(θp-θd
)x (f, 6 fJ.

Next, the actual control operation will be explained with reference to the flowchart shown in FIG.

When the control system of this embodiment is started, initialization is first performed as shown in FIG.
I will explain the black people in 1 as a starting point). That is,
1=2. = 0, and the current time 151 is the time tB of the next data.
, and the next time ta, and at that time. L:l
, d;1. self+fy +f;, that is, the data No. 05 No. 1 and No. 2 in FIG. 3 are entered as initial values [block 31], and the system starts operating.

Then, for each period ΔT, when we add 1 to k, we get ΔT
, θ, L are read [pro-tsuk 32, 33), k
It is determined whether ΔT has reached a next time tP (block 34).

If NO' here, the current load state is determined in block 43, and it is determined whether there is no load (Lx=1) or "
(LK#1), the current desired state feedback gain f (kΔT) is calculated using the initial value and each read value [block 44 or 45], and is output to the compensator 2 [block 44 or 45]. block 46].

When time progresses and kΔT reaches the next time point t, M [when kΔT≧tβ(=1B> in block 34)]
is the above-mentioned t++, with α-B, β-C, r=D,
, l:), l't,), ,1. G (D change FT (flow 35).

That is, the data at points 41, B, 2, and C [data of No., 2+ No., 3, and No. 014 in FIG. 3] are entered. Then in block 36 load 1 of the next point
．． ,.

In other words, the load LD of the light can be checked.

Since Lp=1 at the point, the value when there is no load is set as f [block 37], and then the load status at point β (i.e., point C) and the next point γ (i.e., point D) is calculated. Depending on whether they match [block 39], the process branches to block 40 or 43. In this case, there is no load at both points C and D, and since the load is "no degree" at the present time, the current state feedback gain f(kΔT) is output to the compensator 2 via blocks 43 and 44.

In this way, each time t = t^1t5 +k ,,,
The data is updated every time, and when time te (point E) is reached, t5
．． θ".

No. 6 data is set in Ly [block 3
5]. Here, since Lr''L;= 1, the block 36 branches to the block 38, and the feedback gain f(tp, θρ, −1
) is cent. Then block 41 is reached and point F is reached.

Compare the load conditions at F. As is clear from FIGS. 2 and 3, since LpL = 1 here, the value when there is no load is used as f (block 42).
, proceed to blocks 43, 44, and 46, because the gain in the previous state must be used until just before time tt (dim sum) when the load is applied.

Furthermore, when time tr (point F) is reached, tp, a,, I,
p, f.

Data of No. 6 is set in 11. θ,, t
, 1 is set with the data No. 7 [Pro-Tsuku 35]. Therefore, in block 36, the flow branches to the right in the figure, and in block 38, the feedback gain 1<1=, θQ, -1) with load at point a is sent to f9.

Block 41 then branches to the left in the figure and the process proceeds to block 43, where the current load state is examined.

Since L<-1 at the current time, that is, just before the load is applied,
The process branches to block 44, where the feedback gain under no load is outputted to compensator 2 [block 46].

Up to this point, no load has been applied yet, so everything passes through block 44 and the feed pack gain when there is no load is output.

Next, at time tA immediately after time 1F, when a load such as grasping an object is applied at <point i at the same position as point F, each data is updated in block 35, and then branched from block 36 to block 38. , leading to block 41.

Here, both HI and L are -1, and the current load state LK''' is -1, so it passes through blocks 41 and 43 and reaches block 45. Using the data, the optimal feedbank gain under load is calculated and output to the compensator 2 [block 46].

In this way, in this embodiment, even if the characteristics of the controlled object change, the feedback gain of the compensator is changed one by one according to the position of the controlled object and the load condition at that position, so that the system always maintains the desired state. It is possible to make the moving object follow the indicated trajectory without deviation.

Note that here, LK is only used to determine whether a load is applied or not, so to detect the presence or absence of a load, a positive or negative signal is generated using the reaction force generated when the hand holds an object. This can be done using a detector with a simple structure that outputs .

Furthermore, the adaptive control method for a moving object according to the present invention is effective not only for controlling a robot, a manipulator, etc., but also for controlling a controlled object whose control characteristics change due to movement or application of a load.

(fl) As described in detail, the present invention enables a closed-loop system to always maintain the desired characteristics even if the control characteristics change depending on the position of the controlled object, and allows the moving object to follow the indicated trajectory without deviation. An adaptive control strategy for the body is provided.

[Brief explanation of drawings]

1 to 4 are diagrams for explaining one embodiment of the present invention. FIG. 1 is a system configuration diagram showing an adaptive control system for a mobile body used in the above embodiment, and FIG. A curve diagram showing the relationship between the displacement angle and the feedback gain of the control system, and FIGS. 3 and 4 are flowcharts showing the storage method in the gain storage unit and the calculation algorithm in the gain adjustment unit, respectively, of the control system. In the figure, l is a function generator, 2 is a compensator, 3 and 6 are integrators, 4 is an I/IA mechanical system state observer, and 5 is a motor (Q).
2nd vlJ θ Fig. 3

Claims (1)

[Claims] A mobile object control system comprising a compensator designed based on control characteristics measured in advance at a predetermined position on a designated trajectory of a moving object, storing a state feed hack gain of the compensator at the predetermined position. a gain adjusting means for changing the gain in response to a change in the state feedback gain control characteristic of the compensator; and a load detecting means for detecting that a load is applied, In accordance with the information regarding the load from the detection means, the gain adjustment means calculates the optimum state feedbank gain of the compensator using the state feedback gain stored in the storage unit, and calculates the state feedback gain of the compensator. An adaptive control method for a moving body characterized by controlling a gain.