JP3322892B2 - Multi-axis robot controller - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a control device for a multi-axis robot, and more particularly to a method for linearizing a non-linear operation characteristic between an operation amount and a control amount of a multi-axis robot to remove an interference force acting between each axis. The present invention relates to a multi-axis robot control device that controls the multi-axis robot to follow a target trajectory.

[0002]

2. Description of the Related Art FIG. 2 shows a two-axis robot as an example of a multi-axis robot. In FIG. 2, one end of a first arm 3 is attached to an upper portion of a base 1. The first arm 3 is driven by a motor 4 via a speed reducer 5 and is rotated around the first shaft 2. An angle sensor 6 for measuring a motor rotation angle θ _M1 around the first shaft 2 is attached to the top of the motor 4. A second arm 8 is rotatably attached to the other end of the first arm 3. Second arm 8
Is driven by a motor 9 via a speed reducer 8 and a second shaft 7
Rotate around. An angle sensor 10 for measuring a motor rotation angle θ _M2 around the second shaft 7 is attached to the top of the motor 9.

Conventionally, when trying to make a multi-axis robot follow a target position trajectory, each axis of the multi-axis robot is independently controlled by PID (proportional, differential, integral) as shown in FIG. 3, to measure the motor rotation angle theta _M of each axis of the multi-axis robot 21, PID controller 20 of the measurement results
It is fed back to, and controls so that the motor rotation angle theta _M gives the motor torque command input τ in multi-axis robot 21 follows the target rotation angle theta _MR.

However, in the case shown in FIG. 3, when a multi-axis robot is operated at a high speed, a non-linear force such as an inertia force, a centrifugal force or a Coriolis force acts between the axes, and the axes operate independently. Disappears. For this reason, the accuracy of following the target position trajectory is insufficient by the PID control as shown in FIG. 3 which does not consider the influence of these nonlinear forces.

Accordingly, a control method based on the equation of motion of a two-axis robot as shown in equation (1) has been proposed. Equation (1) also takes into account non-linear forces acting between the axes of the two-axis robot, such as centrifugal force and Coriolis force.

[0006]

(Equation 1) In order to obtain the solution of equation (1), the values of physical parameters such as the moment of inertia and the friction coefficient appearing in the elements of M and c are identified (measured), and the estimated values of M and c, the hat M , Hat c. Then, as shown in Expression (2), the nonlinear forces acting between the axes are calculated, and these nonlinear forces are compensated.

[0007]

(Equation 2) FIG. 4 shows an example of a control method based on Expressions (1) and (2). Control system shown in FIG. 4, the new input u calculated online as shown in equation (3), and performs PID control so that the motor rotation angle theta _M follows the target angle theta _MR.

[0008]

(Equation 3) Nonlinear interference force compensation calculation unit 22 in FIG. 4, the influence of the nonlinear force to linearize the non-linear operating characteristic between the motor rotation angle theta _M is a motor torque command input τ and the controlled variable is an operation amount of the multi-axis robot 21 Is to be removed. PID
The control unit 20 calculates a new input u given by Expression (3).

However, in the control method shown in FIG. 4, when the load characteristics and dynamic characteristics of the multi-axis robot change over time, the values of the physical parameters that determine the mechanical characteristics of each axis change. The effects could not be fully eliminated.

Therefore, as shown in FIG. 5, a physical parameter identification unit 23 for obtaining physical parameters in real time is provided. Identify physical parameters online,
Based on the specified physical parameters, the influence of the nonlinear force is removed by the nonlinear interference force compensation calculating section 22 (Japanese Patent Laid-Open No. 61-226804).

[0011]

However, the control of the multi-axis robot shown in FIG. 5 has the following problems. That is, since only the solution that slowly changes over time can be obtained from the solution of equation (1), the nonlinear operation characteristics of the multi-axis robot are linearized by the nonlinear interference force compensation calculation unit 22 using the solution that slowly changes over time. I was Therefore, only the non-linear operation characteristics in the low frequency band are mainly compensated, and the dynamic characteristics of the drive system of each joint existing in the high frequency band, for example, the mechanical resonance characteristics due to the spring characteristics included in the reduction mechanisms 5 and 8. Could not be considered.

In order to improve the target value followability, the PI
When the D control constant is set large, there is a problem that mechanical resonance occurs in a high frequency band. If the PID control constant is set small in order to avoid mechanical resonance in this high frequency band, there is a problem that the target value followability is reduced. Due to such an opposing problem, the setting of the PID constant has been set by trial and error.

Further, if the load fluctuation and the dynamic characteristic of the multi-axis robot change over time, the mechanical resonance characteristic fluctuates. Therefore, the PID constant set by trial and error cannot be kept fixed, and the trial and error again occurs. It was necessary to set it and it was complicated.

An object of the present invention is to solve the above-mentioned problems of the prior art, and to reduce the interference force acting between the shafts, the fluctuation of the hand load, and the mechanical resonance generated in a high frequency band, while suppressing the problem. An object of the present invention is to provide a control device of a multi-axis robot that can make the axis robot follow a target trajectory with high accuracy.

[0015]

In order to achieve the above object, a control apparatus for a multi-axis robot according to the present invention uses the operation amount and the control amount during the operation of the multi-axis robot to change the physical parameters of the multi-axis robot. A physical parameter identification unit for identifying, a nonlinear interference force compensation calculation unit for linearizing a nonlinear operation characteristic in a low frequency band between an operation amount and a control amount of the multi-axis robot using the identified physical parameter, and the physical parameter For a linearized multi-input / output system obtained by linearization using the identification unit and the nonlinear interference force compensation calculation unit, the linearized multi-input / output system is operated by using the operation amount and control amount of the linearization multi-input / output system. A transfer function model identification unit for identifying a transfer function model representing a mechanical resonance characteristic in a high frequency band existing in the system; and the transfer function model obtained by the identification by the transfer function model identification unit. A control constant calculation unit that calculates a control constant of the linearized multi-input / output system so that the multi-axis robot follows a given target trajectory; and a target of the multi-axis robot based on the calculated control constant. A linear multi-input control operation unit for calculating an operation amount of the linearization multi-input / output system from a target value indicating a trajectory and a feedback control amount of the linearization multi-input / output system.

[0016]

[Function] The physical parameter identification unit identifies the physical parameters of the multi-axis robot, and uses the identified physical parameters to linearize the nonlinear characteristics of the multi-axis robot by the nonlinear interference force compensation calculation unit. Form. Next, a transfer function model in a high frequency band of the linearized multi-input / output system is identified by a transfer function model identification unit using the operation amount and the control amount of the linearized multi-input / output system. Next, a control constant of the linearized multi-input / output system is calculated by a control constant calculation unit so that the multi-axis robot follows a given target trajectory, and based on the calculated control constant, a linear multi-input control calculation unit An operation amount of the linearized multi-input / output system is calculated from a target value indicating a target trajectory and a feedback control amount of the linearized multi-input / output system.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a control device for a multi-axis robot according to the present invention will be described below with reference to FIG. The control device according to the present embodiment is a control device based on Expressions (1) and (2).

In FIG. 1, a linearized multi-input / output system 30 comprises:
Multi-axis robot 3 that is a two-axis robot that turns in a horizontal plane
1, the physical parameter identification unit 32 for identifying the physical parameters of the multi-axis robot 31 by using the motor rotation angle theta _M is a motor torque command input τ and the controlled variable is an operation amount of the multi-axis robot 31, the identification and a nonlinear interference force compensation computing section 33 for linearizing the non-linear operating characteristics between been the motor torque command input τ and the motor rotation angle theta _M of multi-axis robot 31 using the physical parameters.

The control device of the multi-axis robot includes a transfer function model identification unit 34. The transfer function model identification unit 34 includes an operation amount u, which is input data to the linearized multi-input / output system 30, and a control amount. identifying a transfer function model representing the mechanical resonance characteristics in a high frequency band of the linearized multi-input multi-output system 30 on-line using a motor rotation angle theta _M is.

Further, the control device of the multi-axis robot includes a control constant calculation unit 35 and a linear multi-input control calculation unit 36. The control constant calculation unit 35 calculates and updates the control constant of the linearized multi-input / output system 30 online so that the multi-axis robot 31 follows the given target trajectory based on the identified transfer function model. In addition, the linear multi-input control operation unit 36 calculates a target angle θ _{M R} for instructing a target trajectory of the multi-axis robot 31 based on the calculated control constants and a motor rotation angle θ _M fed back from the linearization multi-input / output system 30. Next, the operation of the present embodiment having such a configuration will be described.

First, the procedure for identifying a physical parameter by the physical parameter identification unit 32 will be described in detail. In the case of the two-axis horizontal turning type robot shown in FIG. 2, each matrix or vector of the equation of motion represented by Expression (1) is represented as Expression (4).

[0022]

(Equation 4) In equation (4), α, β, γ, and δ are quantities determined by the length, mass, center of gravity position, and moment of inertia of each axis of the two-axis robot, and d ₁ and d ₂ are friction coefficients of each axis of the two-axis robot. It is the amount determined by Further, n _G1 and n _G2 represent the reduction ratio (≦ 1) of each axis.

Next, the physical parameter vector φ is given by φ =
[Α, β, γ, δ, d ₁ , d ₂ ] ^T , Equation (1)
Is rewritten as in equation (5).

[0024]

(Equation 5) , 2 × 6 matrix.

By constructing Y and τ from the input / output data of the two-axis robot using equation (5), the unknown physical parameter φ can be identified using the least squares method. The identification of the actual physical parameter φ is performed by taking in the input / output data during the operation of the two-axis robot and performing the sequential least-squares method (for reference, Kawasaki and Nishimura, Manipulator parameter identification, Transactions of the Society of Instrument and Control Engineers) , Vo
l. 22-1, pp76-83 (1986)).

Next, the operation of the transfer function model identification unit 34 will be described. The input of the linearization multi-input multi-output system 30, which is linearized by the nonlinear interference force compensation calculation unit 33 u, output and y = (d / dt) θ M. A two-input two-output linear continuous system as the linearized multi-input / output system 30 is discretized at a sampling period ΔT, and input / output data {u (k), y
(K), k = 1, 2,... N}. Here, it is assumed that the identification target can be described by a discrete-time model shown in Expression (6).

[0027]

(Equation 6) The parameters A _i and B _ij in the equation (6) are determined by a sequential least squares method for a multi-input / output system (see Takayoshi Nakamizo, Signal Analysis and System Identification, Corona Corp. (198
8)). This A _i , B
_From the estimated values of _ij , hat A (z ⁻¹ ) and hat B (z
^-1 ).

Thus, the estimated value of the pulse transfer function matrix, hat G (z ^-1 ), is given by equation (7).

[0029]

(Equation 7) Also, the estimated value of the frequency response matrix, hat G (jω), is
It can be calculated by equation (8).

[0030]

(Equation 8) In equation (8), the hat G (j
ω) can be obtained. Therefore, by using Expression (8), it is possible to obtain a component in a high frequency band that cannot be included in the solution of Expression (1).

Next, the estimated value of the frequency response matrix, hat G (jω), obtained by equation (8) is converted online to a transfer function model, hat G (s) (for reference, see Yamashita et al., Frequency Method for Obtaining Transfer Function from Response Measurement, Control Engineering, Vol. 14-11, pp. 15-22
(1970)). As a result, this transfer function model, hat G (s), can represent not only the low frequency band but also the mechanical resonance characteristics in the high frequency band due to the spring characteristics of the speed reduction mechanism that cannot be included in equation (1). .

Although the identified transfer function model, hat G (s), uses (d / dt) θ _M as an output, the transfer function model of output θ _M with respect to input u is (hat G (s) )) / S.

Next, the operation of the control constant calculator 35 will be described. The control constant calculation unit 35 uses the pole designation method based on the transfer function model hat G (s) or (hat G (s)) / s to obtain a parallel compensator F that can obtain a good target tracking characteristic. (S) and the coefficient of the series compensator K (s) are calculated online (for reference, see Nakano and Mita,
Basic control theory, Shokodo (1982)).

Next, the operation of the linearizing multi-input / output control calculation unit 36 will be described. Here, using the already calculated parallel compensator F (s) and series compensator K (s), a servo operation as shown in Expression (9) is performed, and the two-axis robot follows the target trajectory.

[0035]

(Equation 9) According to the configuration of the present embodiment, since the transfer function model identification unit 34 for identifying the transfer function model representing the mechanical resonance characteristic in the high frequency band of the linearized multi-input / output system 30 is provided, each axis of the multi-axis robot 31 is provided. Provided is a control device for a multi-axis robot capable of causing the multi-axis robot 31 to follow a target trajectory with high accuracy while suppressing interference force acting between the robot, hand load fluctuation, and mechanical resonance generated in a high frequency band. can do.

Further, even if the load fluctuation and the dynamic characteristics of the multi-axis robot change over time, it is not necessary to reset the PID constant by trial and error.

Also, since the physical parameters 31, the nonlinear interference compensation calculation unit 33, the transfer function model identification unit 34, the control constant calculation unit 35, and the linearization multi-input / output control calculation unit 36 function online, the multi-axis robot is operated in real time. 31
Can follow the target trajectory with high accuracy.

[0038]

As is apparent from the above description, according to the present invention, a transfer function model identification unit for identifying a transfer function model representing a mechanical resonance characteristic in a high frequency band of a linearized multi-input / output system is provided. , A multi-axis robot control device that can make a multi-axis robot follow a target trajectory with high accuracy while eliminating interference force acting between each axis of the multi-axis robot and suppressing mechanical resonance generated in a high frequency band Can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a control device of a multi-axis robot according to an embodiment of the present invention.

FIG. 2 is a plan view showing a two-axis robot as an example of a multi-axis robot to be controlled by the present invention.

FIG. 3 is a block diagram showing a conventional control device for a multi-axis robot.

FIG. 4 is a block diagram showing another conventional control device for a multi-axis robot.

FIG. 5 is a block diagram showing a control device of still another conventional multi-axis robot.

[Explanation of symbols]

 Reference Signs List 30 linearized multi-input / output system 31 multi-axis robot 32 physical parameter identification unit 33 nonlinear interference force compensation calculation unit 34 transfer function model identification unit 35 control constant calculation unit 36 linearization multi-input / output control calculation unit

──────────────────────────────────────────────────続 き Continued on front page (58) Field surveyed (Int. Cl. ⁷ , DB name) B25J 9/16 G05D 3/12 305

Claims (1)

(57) [Claims] A physical parameter identification unit for identifying a physical parameter of the multi-axis robot by using an operation amount and a control amount during operation of the multi-axis robot; and an operation amount and a multi-axis robot using the identified physical parameter. A non-linear interference compensation calculation unit for linearizing a non-linear operation characteristic in a low frequency band between control amounts, the physical parameter identification unit and the non-linear interference compensation operation
To the linearized multi-input / output system obtained by linearization using the
And, a transfer function model identification section for identifying a transfer function model representing the mechanical resonance characteristics in a high frequency band to be present with the operation amount and the control amount of the linearized multi-input multi-output system to the linearized multi-input multi-output system, Before obtained by identification by the transfer function model identification unit
A control constant calculation unit that calculates a control constant of the linearized multi-input / output system so that the multi-axis robot follows a given target trajectory using the transfer function model; A linear multi-input control operation unit that calculates an operation amount of the linearization multi-input / output system from a target value indicating a target trajectory of the axis robot and a feedback control amount of the linearization multi-input / output system. Characteristic multi-axis robot controller.