JPH11134012A - Robot with track error correcting function - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a robot capable of correcting a track in reproducing operation by an operation program generated without considering the effect of gravitation. SOLUTION: Error corrections of respective axial position command values as outputs based upon a normal function as the CNC of a robot controller are made. Error correction quantities are found by interpolation cycles from the attitude of the robot and the relational expression of the axial moments and error quantities. The addition of the correction quantities is done after filtering. The transfer function of the filter is represented as 1/(K*(A/K*S+1)) (A: coefficient of dynamic friction, K: spring constant). Filtered commands are inputted for servocontrol to perform the servocontrol based upon the corrected position commands as inputs, thereby compensating track errors caused under the effect of the deflection of an arm, the strain of a rotary mechanism part, and play of a backlash, clearance, etc., due to the moment of gravitation and a force of inertia.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an industrial robot (hereinafter simply referred to as "robot"), and more particularly, to a robot having a function of correcting a trajectory error caused by the influence of gravity or the like.

[0002]

2. Description of the Related Art For example, when an off-line teaching is performed to a robot and a reproducing operation is performed based on the off-line teaching, a bending of an arm due to a gravitational moment or an inertial force, an elastic distortion of a rotating mechanism including a speed reducer, and a backlash play. A trajectory error that reflects the difference between the robot model on the computer and the actual robot occurs due to the effects of play, play, and the like. That is,
A deviation occurs between the position of the teaching point specified in the off-line created operation program and the actual robot tip position (usually the position of the tool tip point).

Conventionally, in order to compensate for this, a playback operation of offline teaching is performed to move the robot to or near the teaching point, and further, the tip position of the robot is adjusted (fine jog feed) to obtain the actual desired position. It is customary to correct the position data of the teaching point so that the robot is taught by the robot. However, such a work of correcting the teaching point position requires a great deal of time and skill in robot operation, and places a heavy burden on the operator.

Further, in this method, the effect of reducing the above-described error in the trajectory between the teaching points cannot be reliably expected. This is because factors such as the influence of gravity change variously depending on the robot posture at various points on the trajectory between the teaching points. This is because it is considered to be insufficient (coarse) to correct the generated trajectory error.

[0005]

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a robot having a function of positively correcting a trajectory error generated due to the above-mentioned influence of gravity or the like even with respect to a trajectory between teaching points. Is to do. Another object of the present invention is to reduce the burden on the operator who operates the robot and to improve the work efficiency.

[0006]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the present invention corrects a servo control by correcting a command created based on an operation program for at least one robot axis during a reproduction operation of the operation program. It is intended to be passed to the system. The correction at that time compensates for at least an error generated by the gravitational moment applied to the robot axis.

In a preferred embodiment of the present invention, a moment that changes according to the posture of the robot is calculated for at least one robot axis, and further, the at least one robot axis is calculated using a relationship that describes the moment and the amount of error. An error amount is obtained for.

If there is hysteresis in the relationship describing the moment and the error amount, a relationship related to the hysteresis is selected at the time of calculating the error amount according to the sign directionality when the torque 0 passes. When a command is passed to the servo control system,
It is preferable to perform filtering in consideration of dynamic friction and elasticity.

The error to be corrected can include, besides the bending of the arm due to the gravitational moment, the error caused by the bending of the arm due to inertial force, the backlash of the rotating mechanism, the play, and the like. is there.

According to the present invention, regarding the reproduction operation of the operation program in which the error caused by the influence of gravity is not taken into consideration, the distortion of the rotating mechanism, the bending of the arm,
Alternatively, in addition to the above, the trajectory error caused by the influence of the bending of the arm due to the inertial force, the backlash of the rotating mechanism, the play, and the like is automatically corrected in real time.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a main block diagram illustrating the hardware configuration of a system including a robot to which the present invention is applied. As shown in FIG.
The robot controller designated by “0” controls the robot RB in which the load heavy object 10 is attached to the arm tip 1. Tool tip point TC representing the position of the robot tip
P is set near the load heavy object 10.

The robot controller 30 has a host CPU 3
1. Shared RAM 32, servo CPU 33, servo amplifier 34, memory 35, teaching operation panel interface 36
Also, an input / output device 38 for a general external device is provided.
The memory 35 stores the RO in which the system program is stored.
M, a RAM for temporarily storing data, and a non-volatile memory for storing various program data defining the operation of the robot RB.

The teaching operation panel 37 connected to the teaching operation panel interface 36 is used for inputting, correcting, and registering program data, and for manual input of a manual feed (jog feed) command, a regeneration operation command, and the like. The input / output device 38 for the external device has an offline programming device 40.
In addition, various external devices (for example, a welding power supply device in the case of a welding robot) are connected according to the application.

Regeneration operation or manual feed (jog feed)
At the time of execution, the host CPU 31 creates a movement command for each axis of the robot RB, corrects the error, and outputs it to the shared RAM 32. The servo CPU 33 reads this out in a short cycle, executes servo processing based on a position signal (feedback) signal sent from a position detector (pulse coder) of each axis of the robot, and sends a current command to the servo amplifier 34 of each axis. Output and drive the servo motor of each axis of the robot.

The servo CPU 33 periodically writes the current position of each axis of the robot to the shared RAM 32 based on a position signal (feedback) signal sent from a position detector (pulse coder) of each axis of the robot.

What is important here is that an error correction according to the features of the present invention is performed on a movement command for each axis, and other configurations and functions are not particularly different from those of a normal robot system. . It should be noted that the program data defining the processing for performing the error correction and the related set values are stored in the memory 35 in advance.

As an operation program in which the influence of gravity or the like is not considered, an off-line program created on the off-line programming device 40 and transferred to the robot control device 30 is considered. Hereinafter, error correction performed during the reproduction operation of the offline program transferred to the robot control device 30 will be described step by step.

The errors to be eliminated on the trajectory by the present invention are derived from bending of the arm driven by each axis, distortion of the rotating mechanism, backlash play, play, and the like. It is considered that the factor substantially depends on the magnitude of the moment applied to each axis. Therefore, as a premise for evaluating the magnitude of the error generated in each axis, the relationship between the magnitude of the moment and the amount of error generated thereby is given to the robot controller 30 in some form in advance.

There are generally known various methods for storing the relationship between two variables in a robot. Therefore,
The two variables may be used as a moment and an error, and these methods may be appropriately used. Here, as an example, (1) storage of a moment-error relationship based on actual measurement and (2) storage of a moment-error relationship based on a model combining a strong spring and a weak spring will be briefly described.

(1) Storage of Moment-Error Relationship Based on Actual Measurement FIG. 2 is an example of a graph showing the moment-error relationship for one axis (for example, the third axis) of the robot. The abscissa represents the moment in kgm unit, and the ordinate represents the amount of error generated in the arm (body portion) driven by the axis in the unit of minute. In general, it is known that there is hysteresis in the relationship between the moment and the error, and this graph is also created based on this.

As shown, the entire graph has a portion h0 representing an initial transition from the moment = 0 state.
And a part h + representing the characteristic when the moment changes from negative to positive, and a part h- representing the characteristic when the moment changes from positive to negative.

The crosses distributed along the respective portions h0, h- and h + represent data obtained by actual measurement.
, H−, h + are created from the data according to a known estimation method (for example, least square approximation). The measured data indicated by the crosses is obtained by applying a known force to the tip of each axis arm and measuring the resulting displacement of the tip of each axis arm at an angle.

Of the parts h0, h-, and h +, those actually used are usually h- and h +, so that the data of h- and h + may be input to the robot controller 30. h-,
There are various data formats for h +. For example, parameters of a function expression (for example, a coefficient and a constant term in polynomial approximation or various parameters defining a line function) approximated to h− and h +, and numerical tables of h− and h + are considered. Can be

In any case, the moment-error data for each axis is input to the robot controller 30 in advance. Note that error correction is performed on some axes (for example,
In the case of applying only to the second axis and the third axis in which the error is likely to be large, it is needless to say that the moment-error relationship data need only be prepared for only those axes (formally, the error = 0). You may set an equal function).

(2) Storage of Moment-Error Relationship Based on Combination Model of Strong and Weak Springs FIG. 3 shows the moment-error relationship for one axis (for example, the third axis) of the robot in the same manner as in FIG. It is another example of a graph. The abscissa represents the moment in kgm unit, and the ordinate represents the amount of error generated in the arm (body portion) driven by the axis in the unit of minute.

The data corresponding to this graph is often prepared as a part of the specifications representing the mechanical characteristics of the rotating mechanism including the speed reducer of the robot. This example is based on a combination model of strong and weak springs. In the range where the absolute value of the moment is relatively small, the moment-error relationship is expressed by the strong spring, and in the range where the absolute value of the moment is relatively large, the weak spring is used. The moment-error relationship is expressed.

Further, in a certain range (-μ1 to μ1) around the moment = 0, the existence of hysteresis is expressed, and the function representing the error amount with respect to the moment is represented by f +
And f-. The function f + represents the error amount when the moment changes from negative to positive, while the function f- represents the error amount when the moment changes from positive to negative. In the graph of the present example, in the case of the present example, the data of the moment-error relationship for each axis (all axes or some axes) is stored in the robot controller 30 in the same data format as the graph of FIG.
Is input in advance.

By storing the moment-error relationship in the robot controller 30 as described above, if the moment applied to each axis is known, the amount of error generated in each axis can be estimated. The moment applied to each axis can be calculated from the robot posture (position of each axis) if the structural parameters of the robot (arm length, arm mass, etc.) and the expression parameters of the load weight (load weight, center of mass, etc.) are known. ) Can be calculated.

FIG. 4 shows the axes J1 to J6 of the robot having a first axis J1, a second axis J2,.
It is a figure for explaining the technique of calculating the moment applied to the motor of J6 as a function of the robot posture. In the method described here, the moment torque required for each axis is obtained in order from the hand side from the balancing condition for each link.

Referring to this figure, the calculation procedure will be described by a calculation formula as follows. The illustration of the axis configuration and related quantities in this drawing is simplified as appropriate, and is illustratively performed around the J3 axis.

1. 1. Moment of gravity applied to the motor of the J6 axis = load weight value * distance from the rotation center of the J6 axis to the center of mass of the load weight. 2. Gravity moment applied to J5 axis motor = gravity moment applied to J6 axis motor + (15 * mass of J5 axis arm * SIN θ5) 3. Gravity moment applied to J4-axis motor = gravity moment applied to J5-axis motor + (14 * mass of J4-axis arm * SIN θ4) Gravitational moment applied to motor of J3 axis = J4 axis ~
4. Moment due to J6 axis + (l3 * mass of J3 axis arm * SIN θ3) The gravitational moment applied to the J2-axis motor = the gravitational moment applied to the J3-axis motor + (l2 * mass of the J2-axis arm * SIN θ2) In the above equations, li is the body driven by the Ji-axis from the rotation center of the Ji-axis. The distance to the position of the center of gravity of the portion is set in advance in the robot controller 30. The load weight value, the distance from the rotation center of the J6 axis to the center of mass of the load weight, the arm mass, and the like are also set in advance in the robot controller 30 in advance. .theta.2 to .theta.5 are angular positions of the respective axes expressed with reference to the horizontal direction, and are obtained from the current position at each time point (each calculation processing cycle ITP) during the movement of the robot.

Next, on the premise of the above, control of each axis in this embodiment will be described. FIG. 5 is a block diagram showing the main points of control for each axis. As shown in the figure, a movement command (each axis position command value), which is an output based on a normal function as a CNC of the robot controller 30, is input to a servo control system after error correction. The error correction is performed by adding the correction amount. Instead of adding the error amount calculated for each axis as it is, a value obtained by performing appropriate filtering is added.

This filtering is considered as follows. If the dynamic friction coefficient of a certain axis is represented by A, the spring constant is represented by K, and the error amount is represented by Δθ, the following equation (1) holds.

[0034]

(Equation 1) The relational expression of the transfer function is: Δθ = gravity / {K * (A / K * S + 1)} (2), and the transfer function of the filter is 1 / ｛K * (A / K *
(S + 1)} (S is a Laplace operator).

As described above, the purpose of the filtering is to take into account the gravitational moment and the delay of the occurrence of an error corresponding to the gravitational moment due to the operating friction and elasticity of each axis. Calculate using the coefficient A and the spring constant K, or determine an appropriate value by tuning.

The command filtered as described above is input to the servo control, and the servo control using the corrected position command as an input is executed. Since the details of the servo control are not particularly related to the features of the present invention, the description is omitted.

FIG. 6 shows an off-line programming device 4
During the reproduction operation of the operation program created in Step 0, following the normal processing (reading and interpretation of the operation command statement of the path movement, creation of the trajectory plan), when writing the output as the CNC to the shared RAM 32, every ITP ( 9 is a flowchart (Process 1) schematically illustrating a process executed at each interpolation cycle) and reflecting the features of the present invention.

As shown in the figure, first, in step S1, each axis position is obtained from a position command value (a value held by a position register on the CNC side), and each axis (all axes or one axis) is obtained by the above-described calculation. Calculate the gravitational moment of the part axis. Further, an interference torque is calculated as required. The calculation of the interference torque is not described here in detail because various calculation methods are known, but it is calculated from the dynamic model software provided to the robot controller 30.

In step S2, an error amount is calculated from the gravitational moment calculated for each axis or the sum of the gravitational moment and the interference torque. This calculation is executed using the data of the moment-error relation given to the robot controller 30 in the above-described manner.

In the following step S3, a calculation (digital filtering) corresponding to the above-described filtering processing is performed on the error amount obtained for each axis. Then, in step S4, a process of adding the output to a movement command output for each interpolation process is performed. Output after processing is shared RAM
32 is input to the normal servo control by the servo CPU 33.

In step S2, when the data of the moment-error relationship having the hysteresis is used (for example, when the data corresponding to the graphs of FIGS. 2 and 3 is used), the data related to the hysteresis is used. Therefore, it is necessary to select one of the functional relationships. When using the relationship of FIG. 2, it is necessary to select one of h + and h− for the entire assumed value range of the moment. When using the relationship shown in FIG. 3, when the absolute value of the moment value is equal to or less than μ1, it is necessary to select one of f + and f−.

These selections are made by judging from which direction the torque code has become 0 at the latest opportunity of torque = 0. For this purpose, a flag F (F = 0 or 1) indicating the direction of the zero passage of the torque code is set in the memory 35 of the robot control device 30, and the flag F is set to a required value by the processing 2 shown in the flowchart of FIG. Update sometimes.

That is, for each axis, the torque value (calculated in step S1 of the process 1) is checked for each ITP, and it is determined whether or not the value is 0 (step Q1). If it is 0, the sign of the immediately preceding (previous ITP) torque value is checked (step Q2). If it is positive, the torque value is set to F = 0 (step Q3), and if negative, the torque value is set to F = 1 (step Q4).

When the relationship shown in FIG. 2 or FIG. 3 is used, F = 0 represents the latest history of zero passage from positive to negative.
-Or f-. Conversely, since F = 1 represents the latest history of zero passage from negative to positive, h * or f + is selected. Note that the initial value of the flag F is set to F = 0 here, but may be set to F = 1 in some cases (determined by design in consideration of the operation content and the like).

[0045]

As described above, according to the present invention,
For example, during the regenerating operation of an off-line created operation program, the arm bending caused by gravity, or in addition to the arm bending due to inertial force, backlash of the rotating mechanism, play error, etc. But,
Corrected in real time. In addition, since the work of correcting the position data of the operation program is reduced, the work efficiency is improved.

[Brief description of the drawings]

FIG. 1 is a main block diagram illustrating a hardware configuration of a system including a robot to which the present invention is applied;

FIG. 2 is an example of a graph showing a moment-error relationship with respect to one axis of a robot.

FIG. 3 is another example of a graph showing a moment-error relationship with respect to one axis (for example, a third axis) of the robot in the same manner as in FIG. 2;

FIG. 4 is a diagram for explaining a method of calculating a moment applied to a motor of each axis J1 to J6 of a six-axis robot as a function of a robot posture.

FIG. 5 is a block diagram showing a main point of control for each axis.

FIG. 6 is a flowchart (Process 1) showing an outline of a process that reflects the features of the present invention and is executed for each ITP (for each interpolation cycle) together with normal processing during the regeneration operation of the operation program.

FIG. 7 is a flowchart (process 1) showing an outline of a process for determining the latest direction of zero torque passage in relation to hysteresis regarding the relationship between the error amount and the moment.

[Explanation of symbols]

 Reference Signs List 1 arm tip 10 load heavy object 30 robot controller 31 host CPU 32 shared RAM 33 servo CPU 34 servo amplifier 35 memory 36 teaching operation panel interface 38 input / output device 37 teaching operation panel 40 offline programming device RB robot

Claims (4)

[Claims] 1. A robot having a trajectory error correction function, which corrects a command created based on an operation program and passes it to a servo control system for at least one robot axis during a reproduction operation of the operation program. Wherein the correction is to compensate for an error generated by at least a gravitational moment applied to the robot axis,
The robot. 2. A robot having a trajectory error correction function for correcting a command created based on an operation program and passing it to a servo control system for at least one robot axis during a reproduction operation of the operation program. The correction is for compensating at least an error generated by a gravitational moment applied to the robot axis. For the correction, a moment that changes according to the posture of the robot is calculated for the at least one robot axis. The robot further comprising determining an error amount for the at least one robot axis using a relationship describing a moment and an error generation amount. 3. A hysteresis exists in the relationship describing the moment and the error amount, and a relationship selection related to the hysteresis is made in accordance with a sign direction at the time of passing the torque 0 at the time of calculating the error amount. Item 2. The robot according to Item 2. 4. The method according to claim 1, wherein when the command generated based on the operation program is corrected and passed to the servo control system, filtering is performed in consideration of dynamic friction and elasticity of the at least one robot axis. A robot according to claim 2 or claim 3.