JP3063999B2 - Flexible structure flexibility controller - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial application field) The present invention relates to a flexibility control apparatus for a flexible structure.

(Prior Art) An industrial robot arm is generally designed so that sufficient positioning accuracy can be obtained in an end effector section.
However, when the work is a flexible structure, that is, a flexible elastic body typified by a metal sheet, the bending and vibration due to the elastic deformation of the work itself cannot be ignored, and thus the desired positioning accuracy and high-speed control cannot be expected.

For example, in the case of providing a work supply service to a die (punch and die) of a bending machine, it is necessary to insert the work tip into the main body of the machine according to the height of the die (die). Since the deflection and vibration due to the elastic deformation of the work itself during supply of the work cannot be ignored, the tip of the work cannot be controlled at a desired height, and may fall down from the upper surface of the mold and hit the mold.

For this reason, the working efficiency is greatly reduced, and sometimes there is even a risk of damaging the work. Conventionally, in the above-mentioned bending machine, when there is a possibility that the height of the work tip is lower than the die (die) surface, the amount of saliva descent of the work tip is input, and the height of the robot wrist is increased. Take measures to raise only y, but for this, it is necessary to measure the amount of salivation △ y for each work, or measure it manually, and input it, which is troublesome and hinders automation. ing. In addition, when the end effector is lifted by the amount of saliva △ y to provide the work supply service, since the height of the intermediate position between the end effector and the work is different from the expected height, there is a problem of interference of the part with the machine. Secondary problems such as the occurrence of Further, for example, in swinging a work, the control in consideration of the amount of drooling Δy becomes complicated, and an extra control amount is given in view of safety, so that operation time is wasted.

On the other hand, in recent years, workpieces in sheet metal processing and the like have become thinner and larger, and handling operations such as transporting flexible materials and inserting them into metal processing machines have been automated using robots and manipulators, thereby improving work and production efficiency. Expectations for a significant improvement are increasing.

In view of these points, the present inventors have made the following literature 1)
7), flexibility control has been proposed, and a method for actively controlling mechanical vibration of a flexible structure, a measure, and a method for identifying physical parameters have been described.

Reference 1) Fukuda, Kuribayashi, Hosugai, Yajima, Flexibility control of solar cell paddle (1st report, Vibration suppression and attitude control using solar sensor), Theory, 51-465, C
(Showa 60), 979.

Reference 2) Fukuda, Kuribayashi, Hosokai, Yajima, Flexibility control of solar cell paddle (2nd report, Vibration control and attitude control method using differential solar cell sensor and state estimation), Theory, 52-473, C (Showa 61), 344.

Reference 3) Fukuda, Hosogai, Yajima, Flexibility control of solar array paddle (3rd report, Paddle vibration / posture control method considering reaction wheel dynamic characteristics), Theory, 52
-473, C (Showa 61), 349.

Reference 4) Fukuda, Arai, Hosogai, Yajima, Flexibility control of solar cell paddle (4th report, Torsional vibration suppression and attitude control of paddle using differential solar cell sensor),
53-487, C (Showa 62), 744.

Reference 5) Fukuda, Arai, Hosugai, Yajima, Flexibility control of flexible structures (1st report, Modeling and control method of coupled bending and torsional vibration), Theory, 54-499, C (Showa 63), 630.

Reference 6) Fukuda, Arai, Hosugai, Yajima, Flexibility Control of Flexible Structures (2nd Report, Bending / Torsion Coupling Control Method), Theory, 55-510, C (Showa 64), 365.

Reference 7) Arai and Fukuda, Flexibility Control of Flexible Structures (3rd Report, Physical Parameter Identification for Handling Flexible Structures), Theory (Pending) (1990) Paper No. 90-0
335.

Among them, especially in Document 7), a method for identifying a physical model is considered by actively compensating and controlling the position / posture and vibration of a flexible structure whose physical parameters are unknown using a robot manipulator. Propose,
Simulation results show the effectiveness of the identification method.

(Problems to be solved by the invention) However, the handling control (compensation / vibration control) of a flexible structure using a robot / manipulator can be realized mainly in two points: the necessity of parameter identification and the restriction on sensor / actuator selection. It is considered difficult and there is no concrete research result on this feasibility.

Therefore, an object of the present invention is to provide a flexibility control apparatus for a flexible structure that can extend the result of the previous report 7) and actively compensate the deflection of the flexible structure from the identification result of the physical parameter. I do.

[Configuration of the invention]

(Means for Solving the Problems) In view of the above-described problems, the present invention provides a work holding portion provided at a tip end portion of a robot arm so as to be rotatable in a vertical direction. In the flexibility control device for a flexible structure for positioning to a height position, a displacement sensor for detecting a displacement of the work at a predetermined distance from a grip point of the work grip is provided in the work grip, and the work grip is provided. Natural frequency calculating means for inputting a detection signal of the displacement sensor when a rotational positioning is performed from a stationary state to a predetermined angle and exciting elastic work to the work, and calculating a primary natural frequency of the work, A primary natural frequency of the work calculated by the natural frequency calculating means, a distance from a base point of the work held by the work holding portion to a holding point, From the distance from the workpiece to the tip of the work, the deflection of the tip of the work is determined based on the gravitational acceleration, and the compensation angle for rotating the work gripper in the vertical direction to correct the determined deflection of the tip is determined. Compensation angle calculation means for calculating, and rotating the work gripper up and down by the compensation angle calculated by the compensation angle calculation means to control the position of the tip of the work at a predetermined height and to excite the elastic vibration. And a robot controller for rotating and positioning the work gripper to the predetermined angle.

(Example) Hereinafter, an example of the present invention will be described.

First, a first example of a robot to which the present invention is applied will be described.
In the figure, the robot 1 is shown as an example of five axes (S, L, U, B, T), and the work gripper 1 that grips the work W as an end effector unit has a single-plane position of the work W. Is provided with a non-contact displacement sensor 2 for detecting the distance in a non-contact manner. The S axis is a main rotation axis that rotates the main shaft 3 in a horizontal plane,
L axis is a rotation axis for rotating the first arm 4 in the salivary plane, U
The axis is a rotation axis for rotating the second arm in a vertical plane with respect to the first arm 4, and the B axis is the third arm 6 with respect to the second arm 5.
Is a rotation axis that rotates the gripper 1 with respect to the third arm 6. Therefore,
The gripper 1 can swing up and down while holding the workpiece W by rotation of the B axis.

As shown in FIG. 2, a detection signal of the sensor 2 is sent to an arithmetic unit (C) through an analog / digital (A / D) converter 7.
PU) 8. The CPU 8 may be a CPU of a normal NC device, or may be a computing device separate from the NC device.

The CPU 8 inputs a signal from the sensor 2 as a digital signal and calculates a natural frequency of the work W. The natural frequency of the work W calculated by the CPU 8 is output to the control device 10 of the robot 9, whereby each axis, especially the B axis, is rotated by a predetermined compensation angle, and the work W A predetermined operation such as a bending work supply service is performed while the deflection is corrected and the tip of the workpiece is slightly lifted. Note that the deflection of the workpiece W in FIG. 1 is illustrated extremely.

FIG. 3 is an explanatory view showing an example of the shape of the work W.
As illustrated, the width of the work W is h, the thickness is d, the material density is ρ, the cross-sectional area is A, and the longitudinal elastic modulus is E.

FIG. 4 is an explanatory diagram showing a state in which the gripper 1 grips the work W. The base point of the workpiece W is O, the length is L, the distance from the base point O to the grip point is Rh, and the distance from the grip point to the displacement detection point is xs. The workpiece supply direction is X, and the height direction orthogonal to this in the horizontal plane is Y. FIG. 5 is an explanatory diagram showing a state in which the work W is drooling.

First, the principle of the present invention in the robot 9 as described above will be described.

In FIG. 1, it can be seen that the tip is largely bent because the work W is an elastic body, and the desired positioning accuracy cannot be obtained as it is. Controlling the tip position of a flexible work in consideration of the deformation of the work due to the influence of gravity is called compensation control.

As a model for identifying physical parameters for compensation control, a model based on a constrained mode method is used as in the previous report 7).

According to the previous report 7), indirect handling of elastic displacement of a flexible structure is required for handling control of a flexible structure by a robot. Indirect displacement sensing methods can be broadly divided into two methods: a method using a non-contact displacement sensor and a method using a TV camera. However, if the main purpose is to measure the displacement of the tip of the work, directly measuring the displacement of the tip by the method is considered to be a shortcut, but the method has a disadvantage that the measurement system becomes large. Therefore, in order to realize a compensation control system using a simple measurement system according to the method of the present embodiment, a non-contact type displacement sensor 2 is set at a gripping portion which is a tip end portion of an end effector of a robot holding a flexible structure, Enables indirect vibration sensing.

In the model of FIG. 5, the fixed end of the flexible structure is x =
It is set to 0, and the elastic displacement w (x, t) of the flexible structure is defined as shown in the figure. Here, XY is used as the elastic displacement of the flexible structure.
Considering only those due to bending deformation in the plane, the elastic displacement w (x, t) of the flexible structure is very small.

The cross-sectional shape, cross-sectional area A, density ρ, and longitudinal elastic modulus E of the flexible structure are constant over the x-axis.

Ignore the effects of rotational inertia and shear deformation on bending vibration.

, The motion of the system is described by the following PDE

However, EI is bending rigidity, J _r is the end-around origin O
The moment of inertia of the effector and the flexible structure, g represents the gravitational acceleration.

In the constrained mode method, as shown in Reference 7), (t) ≡0 (3) Using the boundary condition and orthogonal condition as an expansion formula,
The following centralized system is obtained from the partial differential equations of equations (1) and (2).

Here, the deflection due to the elastic deformation at x = xs is set as ws, and ws is decomposed into a first-order mode component w1 and a residual mode component wR of the second-order mode or higher. That is, ws When the equations (5) and (6) are transformed using the first-order mode component w1, the following equation of motion is obtained.

_{J r (t) + P0 (} xs, t) 1 (t) + Res (xs, t) = u (t) -ρAgL 2/2 · cosθ (t) ... (8) Res is composed of vibration components of the second mode or higher. Becomes

However, when estimating the compensation angle based on equations (8) and (9), the calculation of the physical parameters becomes complicated because the coefficient parameters of equations (8) and (9) include integration. . Therefore, here, the equations (8) and (9) are transformed into a static deflection curve. Simplified using (x).

Where ψ0 = ρAL ⁴ / (8EI)… (14) F (x) = 2 * (x / L) ² −4 / 3 * (x / L) ³ + 1/3 * (x / L) ⁴ … (15 ). When c is determined so as to satisfy the direct condition,
Equations (8) and (9) can be modified as follows.

Jr (t) + P0a1 (t ) + Res (xs, t) = u (t) -ρAgL 2/2 · cosθ (t) ... (16) 1 (t) + Ω1a 2 w1 (t) + P1a (t) = -81 / 52Fsgcosθ (t) ... (17) where P0a = 13/45 · ρAL ² / Fs ... (18) P1a = 9/8 · L · Fs ... (19) Fs = F (xs) = 2 * (xs / L ) ² -4 / 3 * (physics ^{xs / L) 2 + 1/} 3 * (xs / L) 4 ... (20) Ω1a 2 = 12.36EI / (L 4 ρA) ... (21) Next, the flexible structure The parameters can be broadly classified into shape parameters (L, h, d, _Jr ) and material parameters (density ρ, longitudinal elastic modulus E).

Here, assuming that all the physical parameters of the flexible structure are unknown, a method using an extended Kalman filter described in the following document 8) is proposed as a method for estimating the compensation angle of the flexible structure.

Reference 8) Katayama, Applied Kalman Filter, Asakura Shoten,
(1983).

This makes it possible to estimate the compensation angle only from the position information of the work W.

First, the static deflection w (L) at the tip of the work W is obtained from the above assumption, w (L) = − ρAgL ⁴ / (8EI) * cos θ (22) Thus, in FIG. 5, if the rotation angle θ of the end effector when the tip position y of the workpiece W becomes 0 is called the compensation angle θ ^* , θ ^* is θ ^* sin ^-1 ｛1 / (Rh + L) * ρAgL ⁴ / (8EI)｝… (2
3)

Now, since the primary natural frequency Ω1 of the work W can be approximated by Ω1a expressed by equation (21), rewriting equation (23) using Ω1a results in θ ^* = sin ⁻¹ ｛1 / ( Rh + L) * 12.36g / ( 8Ω1a 2)}
… (24) Since Rh is known, it is possible to estimate the compensation angle θ ^* by identifying L and Ω1a.

In order to estimate the compensation angle θ ^* , in this example, it is considered that L and Ω1a are identified based on simplified equations of motion (16) and (17) using a static deflection curve.

From equations (16) and (17), the equation of motion of a flexible structure can be written as follows.

^{However, G0 = -ρAgL 2/2 G1} = -81 / 52Fsg (25) and decoupling the inertia term of the equation is ^{transformed, + a0 2 w1 = b0u +} c0cosθ + d0Res ... (26) 1 + a1 2 w1 = b1u + c1cosθ + d1Res ... (27) However, a0 ² = P0aΩ1a / β a1 ² = JrΩ1a / βb0 = 1 / βb1 = −P1a / βc0 = − (G0−P0aG1) / βc1 = (P1aG0−JrG1a) / β (28) d0 = −1 / β d1 = P1a / β β = Jr−P0aP1a Although it takes a little computation time, model parameters (26) and (27) are used to identify the coefficient parameters a0, a1, b0, b1, c0, and c1 using the extended Kalman filter as described in 7). It is conceivable to estimate L and Ω1a.

However, looking again at equation (27), equation (27) is a vibration system with a natural frequency a1, and the term (b1u), the gravity g, and the inclination angle θ due to the input from the end effector as an external force.
It can be seen that there is a term (c1cos θ) due to and a term (d1Res) due to the second or higher order residual mode. In general, considering that the moment of inertia (Jh) of the end effector around the point O is larger than that of the work W (Jp) and that the end effector is subjected to position servo (= 0), It is considered that the free vibration when the elastic vibration of the work W is intentionally excited by the input from the effector can be approximated by the free vibration of the cantilever. Therefore, comparing equations (17) and (27), and assuming that a1 ≒ Ω1a ... (29) c1 ≒ G1 = -81 / 52Fsg ... (30), equation (27) can be expressed as an extended Kalman filter model Identify the coefficient parameters a1 and 1c as
It is possible to estimate Fs, ie, L and Ω1a, based on equations (29) and (30). Therefore, it can be seen that the compensation angle θ ^* can be estimated from Expression (24). In this example, the compensation angle θ ^* is estimated using the equation (27) as a model of the extended Kalman filter under the above assumption, from the viewpoint of shortening the operation time as much as possible and examining the possibility of compensation control.

State vector And the observed value y Then, the state equation and the observation equation are as follows.

Where f = [u cos θ] ^T Equations (33) and (34) are discretized and unknown parameters Consider the following linear system containing

Here, W (i) and v (i) are mean 0, covariance matrix Are assumed to be mutually independent Gaussian white noises. Also, the initial value (0) is a Gaussian probability vector with mean 0] and covariance matrix Σ0, Assume that (0) is independent of W (i) v (i).
If an extended Kalman filter is used for this model, it is possible to estimate the compensation angle by exciting the elastic vibration of the flexible structure.

A simulation for estimating the compensation angle of the workpiece W is performed on the model shown in FIG. The system model used in the simulation places importance on strictness, is based on the unconstrained mode method, and considers the third-order elastic mode. The viscous resistance acting on the system is considered assuming that it can be handled independently for each mode. On the other hand, the model for identification is a model based on the constraint mode method shown in equation (27), and estimates L and Ω1a from equations (29) and (30), and estimates the compensation angle from equation (24). Shall be.

Even if all the physical parameters of the work W are unknown, it is theoretically possible to estimate the compensation angle. However, in consideration of general metalworking, material parameters are often known, so a case where shape parameters are unknown will be described here. That is, the length L of the work W is L = 300, 400, 500, 600 [m
m], thickness d = 0.42, 0.80 [mm], width h = 50 [m
m], eight types of samples are considered, and the length and cross-sectional area of the work W are unknown. The material parameters are all known, and the model parameters are determined as follows.

ρ = 7870 [kg / m ³ ] E = 2.06 × 10 ¹¹ [N / m ² ] Jh = 1 [kg · m ² ] Rh = 269 [mm] In the simulation, the sample was moved from a stationary state of 0 °.
Positioning control is performed to 10 ° to excite elastic vibration, a1 and c1 are identified, and the compensation angle of the work W is estimated by the procedure described in Chapter 3. However, the control input is unified for each sample,
It is large enough to excite the elastic vibration of the flexible structure. In the extended Kalmar filter, the covariance matrices Q and R
Is uniformly Q = 0 R = {w (0) / 10} ² (38) where w (0) is the initial deflection amount. Q = 0 is based on the assumption that the effect of Res in Equation (27) is so small that it can be neglected, and R considers the wR component of the vibration excited by the input here as a kind of noise, and the initial deflection differs for each sample. It was hypothesized as above using quantities. This is because the variance of the residual mode component excited when a plurality of beams having different softnesses are vibrated with the same input is generally considered to be larger as the flexible beam is larger.

FIGS. 6 (a) and (b) and FIGS. 7 (a) and 7 (b) show the results of fixing a1 and c1. However, in FIG. 6, the thickness is 0.80 mm and the length is 600 mm, and in FIG. 7, the thickness is 0.42 mm and the length is 600 mm. From the figure, it can be seen that the identification values of a1 and c1 almost coincide with Ω1a and G1, respectively. Thus, it can be understood that the approximation of the equations (29) and (30) is theoretically satisfactory.

Tables 1 and 2 show the results of estimating the compensation angle for the eight samples and performing compensation control. From Tables 1 and 2, it can be seen that the compensation error y falls within the range of 0 mm to -10 mm.

When performing compensation control using the compensation angle estimated as described above, it can be said that it is desirable to actively control elastic vibration using vibration control together. 8 to 10 show the results of compensation control when vibration control is used together. At this time, the sample of the work W has a length L of L = 600 [mm], a thickness d of d = 0.80 [mm], and a width h of h = 50 [mm]. The vibration control terms added to the input are It is. FIG. 8 shows an example in which positive damping is applied to the secondary mode to make the primary mode attenuate and the secondary mode becomes unstable. Fig. 9 shows the estimated values by the extended Kalman filter In this example, the instability of the secondary mode is prevented by feeding back (HAC: High Authority Control). This shows that the extended Kalman filter has a role as a high-frequency cut filter.

Fig. 10 stabilizes the second-order mode using ws (a kind of LA
C: Low Authority Control), This is an example in which the first-order mode is sufficiently stabilized using (a kind of LAC + HAC). From FIG. 10, it can be seen that by performing hierarchical feedback control as shown in the following document 9), compensation control can be performed while suppressing elastic vibration.

9) Aubrun, J., N., Theory of the Control of Structur
es by Low Authority Controllers, J of Guidance and
Control, 3-5 (1980), 444/451 As the non-contact displacement sensor, an eddy current sensor having a measurement range of 0 mm to 10 mm can be used. Here, since the purpose is compensation control, a compensation control experiment of a flexible structure is performed by moving only the B axis in FIG. Eight samples of the flexible structure are prepared as in Chapter 4, and the robot performs positioning control from a stationary state of 0 ° to 10 ° to excite elastic vibration to estimate a compensation angle. However, the control input was unified for each sample so as to sufficiently excite the elastic vibration of the flexible structure, and the covariance matrices Q and R in the extended Kalman filter were set to equation (38) as in the simulation.

FIG. 11 and FIG. 12 show the results of experiments for identifying a1 and c1.
However, in FIG. 11, a work W having a thickness of 0.80 mm and a length of 600 mm is used, and in FIG. 12, a work W having a thickness of 0.42 mm and a length of 600 mm is used. From the figure, it can be seen that the identification values of a1 and c1 almost coincide with Ω1a and G1, respectively. However, the difference between c1 and G1 in Fig. 12 is conspicuous because the bending deformation at this time exceeds the range of minute deflection, and the actual amount of deflection is smaller than the theoretical amount of deflection. Can be

Tables 3 and 4 show the results of performing compensation angle estimation experiments on eight types of samples and conducting compensation control experiments.

According to Table 3, the compensation error y is within the range of 0 mm to -5 mm even when L is changed in the sample of 0.80 mm thickness, but is 500 mm and 600 mm in the sample of 0.42 mm thickness. It can be seen that the compensation error has increased. Tables 1-4
Fig. 13 shows the relationship between the length and the compensation error by dividing the results up to the thickness of the workpiece W as a flexible structure.
It looks like Figure 14. 15 and 16 when the static deflection is compared between theory and experiment with θ = 0 °. No.
From Figs. 14 and 16, the sample with large compensation error in the experiment had a large amount of static deflection due to large deformation, and the difference between the theoretical and actual values was large. Is considered to have increased. From the above,
The approximation of the equations (29) and (39) is practically acceptable within the range of the small deflection, and it can be concluded that the method of estimating the compensation angle of the present invention is effective.

According to the present invention, when all the physical parameters of the workpiece W are unknown, the physical parameters are identified by a simple measurement system, and a compensation angle for actively compensating the deflection of the flexible structure is estimated from the identification result. Suggested. Through simulations and experiments, the effectiveness of the compensation control system proposed by the present invention and its limitations were shown.

As a model used for identification, a simplified model is sufficiently practicable, and even if the physical parameters are unknown, compensation control can be performed within the range of minute deflection.

The use of the simplified model can reduce the calculation time.

〔The invention's effect〕

As can be understood from the above description of the embodiment, in short, the present invention is directed to a tip of a work (W) gripped by a workpiece gripper (1) provided at a tip of a robot arm so as to be rotatable in a vertical direction. In a flexibility control device for a flexible structure for positioning a workpiece at a desired height position, a displacement sensor for detecting a displacement of the workpiece (W) at a predetermined distance (XS) from a grip point of the workpiece gripper (1). (2)
The displacement sensor (2) is provided when the workpiece gripping part (1) is provided, and the workpiece gripping part (1) is rotationally positioned from a stationary state to a predetermined angle to excite an elastic vibration on the workpiece (W). A natural frequency calculating means for calculating the primary natural frequency of the work (W) by inputting the detection signal of (a), and a primary natural frequency of the work (W) calculated by the natural frequency calculating means. And the distance (Rh) from the base point (O) of the workpiece (W) gripped by the workpiece gripper (1) to the gripping point.
And the deflection (L) from the gripping point to the tip of the work (W) and the gravitational acceleration (g) are used to determine the deflection of the tip of the work (W). Compensation angle calculating means for calculating a compensation angle for rotating the work gripper (1) in the vertical direction for correction, and raising and lowering the work gripper (1) by the compensation angle calculated by the compensation angle calculation means. A robot controller for rotating and controlling the position of the tip end of the work (W) at a predetermined height and for rotating and positioning the work gripper (1) to the predetermined angle in order to excite the elastic vibration. And a configuration including:

As is clear from the above configuration, the present invention is a flexibility control device for positioning the tip end of a work W held by a work holding portion 1 provided at the tip end of a robot arm at a desired height position. The robot controller has a function of rotating and positioning the work gripper 1 to a predetermined angle to excite the work W with elastic vibration.

Then, when elastic vibration is excited in the work W, the vibration is detected by the displacement sensor 2 provided in the work holding unit 1, and the vibration of the work W is detected based on the detection signal of the displacement sensor 2.
The next natural frequency is calculated by the natural frequency calculating means,
Based on the primary natural frequency, the distance Rh from the base point O of the work W gripped by the work gripper 1 to the grip point, the distance L from the grip point to the tip of the work W, and the gravitational acceleration g. In addition, the deflection of the tip of the work W is determined, and the compensation angle for rotating the workpiece gripper 1 in the vertical direction in order to correct the deflection of the tip is calculated by the compensation angle calculation means.

Then, the workpiece gripper 1 is rotated in the vertical direction by the calculated compensation angle by the robot control device, and the tip end of the workpiece W is positioned and controlled at a predetermined height position.

Therefore, according to the present invention, even when the work W held by the work holding portion provided at the end of the robot arm vibrates, the end of the work can be accurately positioned at a predetermined position. It is. At this time, since the work holding portion is rotated by a predetermined angle in advance to excite elastic vibration in the work W to obtain the primary natural frequency, for example, in a case where works of various dimensions are mixed. It is easy to obtain the primary natural frequency of the work W in advance, and for example, it is possible to position the work of various dimensions between the punch and the die of the bending machine to continuously perform the bending. In such a case, efficient processing can be performed.

[Brief description of the drawings]

FIG. 1 is an explanatory view showing an example of a five-axis robot embodying the present invention, FIG. 2 is an explanatory view of a control system of the robot, FIG. 3 is an explanatory view of a work shape, and FIG. FIG. 5, FIG. 5 is an explanatory diagram showing a model, FIGS. 6 and 7 are explanatory diagrams showing identification results of coefficient parameters, and FIGS. 8, 9, and 10 are compensations using vibration control together. FIGS. 11 and 12 are explanatory diagrams showing identification results of coefficient parameters by experiments, and FIGS. 13 and 14 are explanatory diagrams showing a comparison of compensation control results, FIGS. 15 and 16. FIG. 4 is an explanatory diagram showing a change in the amount of static deflection. 1 ... Grip part 2 ... Non-contact displacement sensor 9 ... Robot B ... Rotating axis of end / effector unit W ... Flexible structure (work)

Continuation of the front page (56) References JP-A-1-173117 (JP, A) JP-A-62-181889 (JP, A) CHRISTIAN AD SEE RING WP, Design cosidification for an art based flexible robotics m, Ploc IEEE Int Conf Rob Autom, (United States of America), 1989, Vol. 3, page e. 1396-1401 Fumito Arai (2 others), Research on Flexible Material Handling Robot, Proceedings of the 8th Annual Meeting of the Robotics Society of Japan, November 1, 1990, 3, pp. 945-946 Fumito Arai (2 other students), Research on Flexible Material Handling Robot, Proceedings of the 29th Annual Conference of the Society of Instrument and Control Engineers, July 24, 1990, Vol. 1, p271-272 (58) Fields surveyed (Int. Cl. ⁷ , DB name) B25J 13/00 B25J 9/10 JICST file (JOIS)

Claims (1)

(57) [Claims] A flexible structure for positioning a tip of a workpiece (W) gripped by a workpiece gripper (1) provided at a tip end of a robot arm so as to be rotatable in a vertical direction to a desired height position. In the flexibility control device, a predetermined distance (ΧS) from the gripping point of the workpiece gripper (1)
In the above, a displacement sensor (2) for detecting a displacement of the work (W) is provided in the work gripper (1), and the work gripper (1) is rotated from a stationary state to a predetermined angle to perform elastic vibration. Natural frequency calculating means for inputting a detection signal of the displacement sensor (2) when exciting the work (W) to calculate a primary natural frequency of the work (W); Work (W) calculated by
, The distance (Rh) from the base point (O) of the workpiece (W) gripped by the workpiece gripper (1) to the grip point, and the tip of the workpiece (W) from the grip point. The deflection of the tip of the work (W) is determined based on the distance (L) to the object and the gravitational acceleration (g), and the workpiece gripper (1) is corrected to correct the determined deflection of the tip. A compensation angle calculating means for calculating a compensation angle to be rotated in the vertical direction; and a tip end of the work (W) by rotating the work gripper (1) up and down by the compensation angle calculated by the compensation angle calculation means. The work gripper (1) for positioning control to a predetermined height and exciting the elastic vibration.
And a robot controller for rotationally positioning the robot to the predetermined angle. A flexibility controller for a flexible structure, comprising: