JPS63238602A - Method and apparatus for adaptive control of control parameter of industrial robot - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention includes a plurality of motion axes, a drive motor for each axis,
The present invention relates to a method for optimal parameter control of an axis controller of an industrial robot, comprising an axis controller that controls the axis movement according to a desired value supplied to the controller, and a computer device for controlling the robot.

The invention also relates to a device for carrying out the method.

The present invention can be applied, for example, to ASEA pamphlet: CK09-1
101 E%CKO9-1102E%CK09-110
3g%CKO9-1109E%AO9-10303%A
09-1111g% A09-1113g relates to an industrial robot of the type described.

Such robots are equipped with several axes of motion, six as standard. In one axis, the mass moment of inertia, the influence of gravity and the off-axis influence of motion in other axes can vary within wide limits depending on the current position of all axes and the mass of the loads produced by the robot. When designing and adjusting the position control system for each axis, the control parameters must be limited to obtain fast and stable operation even in the case of the most undesirable operation from a control point of view. This means that the acceleration and deceleration of the axis movement is generally far below the optimum value. This in turn means that the time required for movement between two points is unnecessarily long, and this applies especially to movements over short distances. In many applications, such as spot welding and assembly, the robot's work program consists of a large number of relatively short movements. The so-called cycle time, ie the time for executing a program, is particularly long in these applications, which is a serious obstacle from an economic and production technology point of view.

It is an object of the present invention to provide a method and apparatus for controlling an industrial robot in which cycle times are significantly reduced compared to prior art robots.

It is a further object of the invention to provide a method and a device by which these advantages can be achieved by means of a computer of moderate computational capacity included in the control equipment of a robot.

The features of the method and apparatus according to the invention will be apparent from the appended claims.

Embodiments of the invention will be described in detail below with reference to the accompanying drawings.

FIG. 1 shows a schematic diagram of an industrial robot to which the present invention is applied. The mechanical part 1 of the robot includes a base plate 11 fixedly arranged on the floor or other foundation.

Column 12 can rotate about a vertical axis.

The lower arm 13 can rotate about a horizontal axis passing through the upper part of the column 12. The upper arm has two parts 1
4a and 141), the part 141) being the arm 1
4 can be rotated about part 14a about an axis coinciding with the longitudinal axis of 4. Arm 14 supports a robot hand 15 that can rotate about an axis passing through its outer end.

The robot's hand consists of two parts 15a and 151), the part 15N) being able to rotate with respect to the part 15a about an axis that coincides with the longitudinal axis of the hand 15. The outer part 151) of the hand 15 is a gripper 16 that supports the object L head-wise.
is supported. Depending on the field of application, the robot hand 15 can alternatively or additionally carry a work tool, such as a screwdriver, a spot welding instrument, or a spray gun.

The control unit 2 of the industrial robot comprises the equipment for supplying the necessary voltages, the drives for the drive motors on the different robot axes, and the calculations and logic decisions necessary to control the robot during programming and operation. a computer device which stores the coordinates of a number of positions and which determines the movement of the robot and the orientation of the robot during automated operation, and instructions which the robot is programmed to execute at various positions; and a program storage device for storing the program.

Furthermore, the control unit includes the necessary communication units, digital-to-analog converters, etc.

Furthermore, the industrial cIM kit is equipped with an operating unit 3 adapted to operate the robot primarily during the zologram phase. The unit 3 is equipped with a control frame 31 that can be moved with three degrees of freedom, by means of which the position and orientation of the robot's hand 15 is controlled during programming. Further, the operating unit 3, together with the display unit 32, commands,
It is equipped with a large number of operation push buttons 33 for inputting commands, numerical data, etc.

The control unit 2 is connected to the mechanical part 1 and to the operating unit 3 by electrical cables.

FIG. 2 schematically shows some of the units included in the control unit 2 and their connections to the mechanical part 1 and to the operating unit 3; Control unit 2 includes a data bus 26 on which units 21-24 communicate with each other. Operating unit 3 includes interface unit 2.
1 to the data bus 26. moreover,
A computer device 22 is connected to data bus 26.

The computer device 22 controls the 1:l #t during the programming phase and the automatic operation phase and performs the calculations necessary for the control, e.g. in a manner known per se, a number of intermediate positions between each pair of programmed positions. Insert position.

The congiator device 22 also performs the calculations performed for the control of the parameters of the axis controller according to the invention.

Furthermore, a storage device 23 is connected to the data bus 26. The storage device 23 stores general programs necessary for controlling the robot.

Furthermore, in the storage device 23, during programming of the robot, the coordinates of points defining the desired movement and orientation of the robot during automatic operation are stored. Furthermore, the storage device 23 stores data necessary for parameter control according to the present invention, as described below.

Units 22 and 23 are only functional and there is no need to structurally separate the two binits.

Additionally, combinator device 22 may serve as one or more computers and one or more storage devices.

FIG. 2 shows the drive of one of the axes of the robot and the specific control device belonging to that axis. The robot axis has a drive motor MO, which drives the movement in the axis, to which is connected the lever ψ and the speedometer generator TG. The motor MO can be an AC motor powered by a drive 25, which can be a pulse width modulated inverter, for example. The drive device 25 then applies an alternating voltage with a frequency, amplitude and phase position controlled by the computer device 22 of the control unit 2 via the interface 24 in such a way that the desired motor torque is obtained at each instant. supply

Alternatively, the drive motor MOd may consist of a direct current motor, whereby the direct current supplied to the power windings of the motor MO is controlled by a drive 25, which may then e.g. It can be a converter.

The resolver section is mechanically connected to the motor shaft and provides a signal θ representing the instantaneous rotation of the robot shaft with respect to the zero position (synchronized position). Furthermore, a speedometer generator TG is mechanically connected to the robot axis and provides a signal δ representative of the current rotational speed of the axis in question. In other embodiments, the speed of rotation can be calculated from the resolver signal. The interface unit 24 includes a supply device necessary for the resolver section, and also a generator TG for the resolver section and the speedometer.
Contains the necessary converters to convert the signals from to digital form.

For simplicity, FIG. 2 only shows the arrangement for a single axis of the robot. Drive and control devices for other axes 24.2
5.1 is formed in the same manner as shown in FIG.
Digital and analog multiplexing of the signals is performed.

FIG. 6 shows an embodiment of a robot axis position control system. The functions shown inside the rectangle 22+23+24 are
Although performed by a computer, for simplicity the functions are described below as if performed by hardware circuitry.

The interpolator IP interpolates between the six pairs of positions and orientations stored in the program towards a large number of positions and wrist directions. The interpolated positions and orientations are converted into coordinates in the axis system of the robot and fed to the position controllers of the different axes in the form of position reference values (angle reference values). The position reference value obtained for one axis is represented by θr. In the adder circuit S1, the reference value is compared with the actual value θ obtained from the ψ sulfur section. The difference is fed to a position controller PR. This is a purely proportional controller, in which case the position reference value is such that the axis position θ follows a parabolic path depending on time during acceleration and deceleration, and the acceleration and deceleration are constant. method is sent to the position controller appropriately. Alternatively,
The position controller may have a parabolic gain function such that the controller's output signal is a parabolic function of its input signal. This method also provides constant acceleration (current limiting) and deceleration (controlled linear velocity ramp) in the robot motion, respectively. The gain Kp of the position controller is variable and is controlled as follows.

The output signal θr of the position controller constitutes the speed reference value of the system and is compared in an adder S2 with the speed reference value θr obtained from the speedometer generator TG (alternatively the speed value δ calculated from the resolver signal). be done. The error signal is fed to the speed controller 1, the gain Kv of which may be constant or variable (see below). The output signal of the speed controller VR constitutes the current reference Ir for the drive 25. The current reference is compared in an adder 86 with the actual value of the motor current obtained from the measuring device M. The deviation between the currents is
Drive device 2 in such a way that the motor current follows the desired value Ir.
Control the voltage of 5.

FIG. 4 schematically shows the calculation and determination procedure for determining the gain of the position controller PR. In the block denoted by MODEL, the mass moment of inertia of the axis J1, the combined mass moment of inertia between the axes Jij (Jjj is the mass moment of inertia between axis i and axis j), and the power due to The moments Mgi are calculated for different axes of the robot (1 is an index of 1 for axis 1, 2 for axis 2, etc.). Calculate the position θ1 of different axes
, the rotational speed δ1, and the mass mL of the load carried by the robot at a particular time in a manner detailed below. The calculations are carried out with the help of pre-set-up mathematical models which are more or less approximate depending on the robot mechanics and the available processor capacity.

In the block represented by PARCALC, equations are created that describe the relationship between each robot axis' acceleration, motor torque, friction torque, gravitational torque, mass moment of inertia, combined mass moment of inertia, and direction of motion.

Information about the direction of movement is obtained from the block MON, whose output signal 8ON(θj) indicates the direction of movement in the axis. The equations are solved assuming that the maximum available motor torque prevails on each axis, resulting in the maximum available acceleration 1θ11 for each axis based on the above assumptions. For each axis, the latter value is compared with the maximum acceleration/deceleration allowed for the axis with good control characteristics maintained. 2
The one with the lowest absolute value of the two compared values is used to determine the gain Kpi of the position controller.

Since dynamic calculations must be performed in real time, it is important to choose an appropriate mathematical model for the robot's mechanical properties; if an inappropriate mathematical model is selected, the calculation volume will be reduced. large enough to make it difficult or impossible to carry out with a fair contribution of computer capacity.Furthermore, the necessarily compromised approximation is compensated for in a simple way by increasing its accuracy. It should also be of such a type that it can be used to describe new mouthpiece structures without the need for expensive modifications.

A mathematical model is described below, in which the above-mentioned advantageous properties are achieved by appropriate selection of the approximations. The first approximation involves omitting physical phenomena that are less influential in this case, such as centripetal forces. The second approximation involves describing the mechanical properties of the robot with the aid of virtually point masses as described with respect to FIG. The sixth approximation involves selecting an appropriate geometric approximation that describes the position of the point mass as a function of the robot structure in question.

FIG. 5 schematically shows the various parts of the robot with the same symbols as in FIG. The motion of all robot axes consists of rotation, and θ1 (1-1, 2--6) represents the position of each axis (the angle of rotation with respect to the synchronous or reference position). The selected point mass is expressed as ml-m9. Mass m1,
m4 and ml are used to calculate the moment created by the effect of gravity on axes 2-6 from arms 13 and 14, hand 16, and the weight. Mass 2 is used to calculate the mass inertia of arm 13, masses m6 and m5 are used to calculate the mass inertia of arms 14a and 141), and masses m8 and m9 are used to calculate the mass inertia of the load. used for. This relatively simple model provides a good approximation of the 11b characteristics of current robots if the size and location of the point masses are chosen appropriately.

The combined mass-inertia is calculated as either an inertial mass or a gravitational mass, depending on the design of the shaft transmission. For example, the gravitational mass is used for the combined mass inertia between shafts 2 and 3 due to the tie rod transmission of shaft 3, while for example the combined inertia between shafts 5 and 6 is Calculated by inertial mass.

Mass m6 is the additional mass used to calculate the mass inertia due to the gravitational torque and extra loads carried by arm 14 (eg, a reel of welding wire or a current supply unit for a switch welding gun). Once the mathematical model has been selected in this way, equations describing the robot's dynamics are created from known mechanical relationships - for example, the equation for the mass inertia J6 of axis 3 and the effect of gravity on axis 3. The formula for the torque Mg3 is shown below: Js = Jso + m3 (lss)” + m
5 (lsa)” + 1!16 (lsg)” +
m-1(Lffs + '15? Cog
(θ6−θs) ” (1) Mgs =
m4g ms4 gin(θ3+θ:s4) + m6
g 1s6sin (θ30θsg) 10my g
(Li2 5in(03) + 1stcos(θ6
-03) cosθ3- 157 8in(θ6
- θ5) coxθ4)
(2) where Jl: Total mass inertia of axis 1 Jjo: Moment of inertia of that and axis 1 m1: Mass of point mass 1 1jj: Distance between axis 1 and point mass j bg: Axis 1 and axis j distance between θi: position (angle) of axis 1 Mgi: torque of axis 1 due to gravity g: gravitational constant θ1j: position of point mass j expressed as angular position with respect to axis 1 The following geometrical estimation is performed here. The distance from the X-axis of the wrist coordinate system to the point mass representing the knee load applied is omitted.

The mass inertia of a load is calculated using the gravitational mass rather than the equivalent mass of inertia.

The effect on the axis 4 when the position of one axis 5 deviates from the reference position is omitted.

Since the axes influence each other, relationships must be defined that indicate the coupling between different axes.

- as an example, the combined mass inertia J between axes 5 and 6
The formula representing 65 is as follows: J6s = -ms
16s,, CG65 16n + 15s
Co5(θ5B) sin(θ6+θ6a)) −m91
ao (G65 16o + 159cos (θs
9) 5in(θ6+θe9))
(3) where trij: combined mass inertia between axis 1 and axis j.

Gjj: Transmission ratio between axis 1 and axis j.

Once the mass inertia, moment of gravity, and combined mass inertia are calculated, these parameters are used to update the dynamic equations defined for each axis: where θJ and )°j: axis 1 and axis 1, respectively. Acceleration of J.

Mi: Axis 1 motor torque.

Fl: Friction torque of axis 1.

When defining the above-mentioned equations describing the axis dynamics, the different parts of the robot are assumed to be rigid and play-free, which provides a good model for determining the control parameters that respectively give the optimum acceleration and deceleration. Of course, if considered justified, the model can be complicated, for example by taking into account elasticity and any pre-existing play.

As mentioned above, the desired optimal acceleration/deceleration in each axis can be obtained in two different ways. By the first method,
The position controller PR is arranged with a parabolic transfer function, and in other ways the position reference θr is calculated in such a way that the position during acceleration and deceleration follows a parabolic course as a function of time.

The present invention shows methods used for both of these alternative embodiments of position control. The next part of the method consists of two steps.

In the first stage, these maximum acceleration and deceleration values are calculated,
These are obtained assuming the maximum available motor torque is used for each axis. For each axis, the dynamic relationship equation (4th equation) defined above is then used for each axis. The equations for acceleration are defined and used as follows, where Equation 4 is of the form: where 1θ11: maximum possible acceleration value Mjmax: maximum available motor torque υir):
In the case of speed-based deceleration, the fourth equation can be written as: Jl 1θ11- l IMlmaxl + FBI
-8GN(θj)
(6J where θ1: Actual value of velocity The two equations regarding acceleration and deceleration determined as above are respectively solved, and thereby the maximum achievable absolute values of acceleration and deceleration are obtained for the various axes. - For example: , the equation thus obtained for the deceleration of axis 2 is shown below, provided that only the inertia coupled to axis 2 is derived from axis 3: J3. (IIMsmaxl + F31
-8ON(θ3) -Mg5) ・5()N(0m)
/ J3J / (Ji - J2s・J2g:s)
(7) With the help of the method just described, 1i11
is constantly updated by various axis equations corresponding to the seventh equation above. Since Ji, Mg1 and Jlj all depend on the position of various axes, 1θ11 depends on the current robot shape at each instant.

For high calculated values of the absolute value 1θj1, it appears that in some cases the control characteristics of the position controller are not good, and therefore in the second step of the procedure the following selection is made: l#ilc-Mjn (ILII , l″jitma
x) (s) However, 1Δ11c: Acceleration/deceleration value used for the next regulation of control parameters.

IMi+: Acceleration/deceleration value calculated by the seventh equation above and illustrated by the equation.

1'ii1max: Maximum expected acceleration/deceleration value while maintaining good control characteristics.

The value 1aii1ma depends on the mass inertia and also on the adjustment of the position and speed controllers (see FIG. 3). Mlmax is suitably determined by experiment or by simulation as a function of mass inertia.

When the optimum acceleration/deceleration value is determined by the above-mentioned formula (8), the control parameters are determined so as to provide four controller operations in accordance with the value determined by the formula (8). How this is done depends on which of the two above-mentioned control procedures is used.

If the position controller PR has a parabolic gain function, the speed reference θir can be written as follows: where eθ is the slip (error signal) of the position controller.

n is a constant 2.

Kp1 is a proportionality constant (gain of the position controller).

C is a parabolic constant.

ffa e, is an offset signal.

Optimization of the parameters can now be done as an adjustment of Kpl according to the following formula: Kpi −F(lθji
c, al, jir) α [However, F
is a continuous function derived from the characteristics of the nonlinear position controller.

In the normal case, the function F, together with the above equation 9, is as follows: where n and C are constants that define the parabolic transfer function of the position controller.

eδ1■θjr-01 =ai: Time rate of change in eθ1.

In the second position control method using motion path zoning, the desired position value of the position controller is created as a parabolic function of time, so the adjustment of the control parameters of the position controller is performed as follows, and the robot computer It seems to be calculated by periodically extracting the value of .

'ir (t+Tl3) -0ir (t) + 19
1r (t) ・TB + 1/2 ・l'1lc-T
B” ir (t+T5) =σir (t)+1σilc
-T5 Q Sae However, TB is the sampling interval.

It has been explained above how the slope of the ramp function formed by the gain of the position controller alone and the position reference value supplied to the position controller is influenced to obtain optimal control behavior. The desired change in the gain of the position control loop is obtained alternately by the simultaneous influence of the gains xp and Kv of the position and velocity controllers, respectively.

FIG. 6 shows how data and calculation functions are organized in the method according to the invention. The work pit model consists of the block DYNALC. selection of acceleration values for different axes,
The determination of the optimal values of the control parameters and their prescription are carried out in block PARCALC. In order to minimize the required computational effort by avoiding unnecessary table iterations, intermediate results of the calculations used in subsequent calculations are constantly stored in the database C0MM0N. As soon as the device is started, the largest possible part of the required calculations is carried out in block I.
NITCALC, i.e. the results of such calculations are unaffected by shape changes that the line robot undergoes during operation. These calculations are controlled by the block IN IT. Functions requiring extensive calculations, such as trigonometric functions, for example, can be suitably implemented in the form of tables placed in module 'rAB, so as to reduce the required computational capacity. Parameter-defined functional blocks are data areas (
) This is also similar to LOBAL, which includes data in which the parameter-defined functions have common functions of other robot control systems.

Modules k LOAD and C0NTR contain the data and functions necessary to initiate and control the parameter defined functions. These functions also lead to a data area pos containing position and velocity values of the different robot axes, etc. The optimum values of the control parameters calculated by the parameter calculation function PARCALC are supplied to the data area PARDATA, from which they are continuously retrieved during operation of the robot and used to influence the control parameters of the position controller. .

Optimization of the parameters according to the invention leads to a significant reduction in cycle times, especially in the case of short movements where deceleration immediately precedes acceleration. This covers six different cases l
-1 is illustrated in FIG. The top diagram for each case shows velocity, the middle diagram shows acceleration, and the bottom diagram shows motion as a function of time. All three cases relate to axis 2. In FIG. 7, θ represents speed, θ represents acceleration, M represents motor torque, Mg represents torque due to gravity, and J represents moment of inertia.

Case ■ shown on the left relates to the case where the moment of gravity has its maximum value. During acceleration, the parabolic gain function of the position controller forces the drive motor to provide its maximum force, in this case to which gravity torque is applied. Similarly, during the deceleration phase, the parabolic gain function of the position controller provides near maximum motor torque, but since the gravity torque is now in the opposite direction, deceleration is significantly slower than acceleration. In the worst-case adjustment, Kp is defined to a value corresponding to the deceleration in case (2).

In case ①, the parameters of the position controller are expected to be adjusted for the worst case according to case I.

In case (2), the robot shape appears to be such that the moment of gravity is zero.

The solid curve for case H relates to this case, while the dashed line by comparison shows the sequence according to case I. As can be seen, the maximum motor torque is obtained in the acceleration phase, while the deceleration in the deceleration phase is at the adjusted value for case ■.

Deceleration therefore takes an unnecessarily long time in relation to the available motor torque. The time of movement is tl, which exceeds the corresponding time in case I to'i6%.

Case I shows the improvement obtained by the invention under the same conditions as Case II. During both acceleration and deceleration, the drive motor operates at maximum torque. The total exercise time t2 is 25% shorter than in case ■.

The operating case shown in FIG. 7 is based on significantly simplified assumptions, for example only axial gravity is taken into account. More complete calculations that also take into account changes in the combined moment of inertia and mass inertia have shown that the time required for a given motion can be reduced by up to 40%. This magnitude reduction is extremely economical and of practical importance in robotic applications such as spot welding, assembly, gluing, etc. In addition to this significantly increased speed of action in the robot's work, a reduction in the average positional defects during movement is also obtained with the help of the parameter optimization according to the invention.

FIG. 8 schematically shows how parameter optimization according to the invention is integrated with a robot controller. Blocks DYNCALC and PARCALC have the functions described above, namely to perform a dynamic calculation and calculate optimal parameter values, respectively. Calculated parameter value PA
R is supplied to the axis controller W. The dynamic calculation provides the maximum achievable acceleration values θtim for the different axes. In the function block PATHPLAN, the acceleration values that are always limited by these value lines and axes are changed into acceleration values in other axes that allow the robot to move along the desired path.
Supplied. Function block PATHPLAN provides a position reference value θr that is supplied to the axis controller. These supply current to the drive motors of the robot and receive information from the robot regarding the position θ and velocity θ of the various axes.

As mentioned above, dynamic calculations are approximate, and as a result the optimized parameter values may be too low. To compensate for this, adaptive calculations may be performed to modify the parameter calculations made based on the actual robot motion. This function is the function block AI) CAL
, C! fulfilled by In this block, measurements of the dynamic deflections of the axes are carried out, for example during deceleration steps, and are used to compensate for approximate defects in the robot model. For example, the deviation of the motor current from the desired value is measured and compensation is performed by adjusting the gain created by the parameter control.

9 to 13 show another embodiment of the invention.

9 and 10 show an industrial robot, known per se, in which this embodiment of the invention is used. The robot and its control device can, for example, be of the type described in the following A8EA publications: CK 09-1101gr Industrial Robot Systems IRB 672 and IRB L6/2'l'J;
cx 09-11033 r industrial robot system IRB 90/2 J; CK 09-1109g r industrial robot system RB1000.

The robot comprises a substrate 41 that rests on a fixed foundation (for example the floor). The column 42 has an axis A shown in FIG.
The rotating wheels are arranged to rotate with respect to the substrate 41. Rotation is effected 9 with the aid of a drive motor (not shown), the axis of rotation being designated 9. At the top of the column 42, a lower arm 43 is journaled to be rotatable about an axis B perpendicular to the plane of the paper shown in FIG.

The rotation angle of the lower arm 43 with respect to the vertical direction is represented by α. The lower arm 43 is rotated by a drive motor 48 connected to the lower arm 43 by a ball screw and an arm. At the top of the lower arm 43, the upper arm 4
4 is rotatably journaled on the shaft Co. Upper arm 440 rotation is controlled by a second drive motor (not shown) arranged adjacent to drive motor 4B and connected to upper arm 44 via linkage arm 49. Upper arm 4 in vertical direction
The direction of 4 is denoted β and is defined by the drive of the upper arm 44 regardless of the rotation of the lower arm 430.

At the outer end of the upper arm 44, a wrist 45 is journaled for rotation about an axis D(7) perpendicular to the plane of the paper. The orientation of the wrist 45 with respect to the vertical direction is designated r and is defined by a drive motor (not shown). The gripper 46 is attached to the wrist 45 and fully supports the object. The robot acts as an operating bit, i.e. it grabs the object 41 in the collection position and carries it to the delivery position;
And it seems that object 477 is dropped there.

The length of the lower arm 43, ie the distance between the axes B and C, is designated Ll, and the length of the upper arm 44, ie the distance between the axes C and D, is designated L2.

The robot's control system is arranged in a control cabinet 410. The package includes a computer (possibly several computers) with program storage. During robot programming, sequences of instructions are stored in a program storage device that define the robot's work cycle and are executed sequentially during automatic operation. The instructions include the coordinates of a number of points that define the path of the robot, as well as the required number of coordinates that define the orientation of the robot's hand at each point. Additionally, at least some of the instructions include information that defines the tasks that the robot is to perform during automatic operation. Programming unit 411 includes a control frame 412 for manually positioning the robot during programming procedures. Unit 411 further controls the robot during programming and automatic operation;
It includes a number of operating members for storing commands as well as a number of numeric keys for inputting all numerical data. The programming unit 411 is connected to the control cabinet 410 via a preferably digital communication channel 413 consisting of a serial or parallel transmission circuit of a digital demuter. The control cabinet 410 further includes feedback control systems for the different axes of the robot, as well as devices for supplying current for the controller and drive motors. The control cabinet 410 is connected to the actual robot via a channel 414, which is likely to be a cable that carries analog signals and perhaps even digital signals in both directions between the control cabinet 410 and the robot. It's sad.

"Degree of freedom" here means the motion of the robot that is independent of the motion of other degrees of freedom. Rotation about axis A thus constitutes the first degree of freedom, and rotation about axis B of the lower arm 43
The rotation of the upper arm 44 of the axis C constitutes the third degree of freedom, and the rotation of the wrist 45 of the axis C constitutes the fourth degree of freedom. . Thus, although the robot shown in FIG. 9 has four degrees of freedom, the wrist 45 actually has an additional degree of freedom.
Therefore, robots that are actually used regularly have five or six degrees of freedom.

A servo system is arranged for each degree of freedom, and FIG. 10 shows an example of such a known servo system.

From a computer included in the control system, the servo system is supplied with position reference values Yr t''. These position reference values consist of values entered into the program storage device fc5. In the adder circuit 421, the reference value is compared with the position response y obtained from the resolver section S.The output signal from the adder 421 constitutes the position error ep, and is supplied to the position controller 422. .

This results in a non-linear, so-called parabolic gain, whose gain is denoted by Kp. The output signal of the controller 422 constitutes a fast-wave reference value Vr and is fed to an adder 423, where it is added to the speed response V obtained from the speedometer generator TC).
compared to The output signal from the adder 423 constitutes the speed error ev, and the proportional differential and integral coefficient effect and the gain Kv k
A suitable speed controller 424 is provided. The output signal U of the speed controller 424 is fed to a drive device 425, for example a controlled converter, which powers a drive motor M that carries out the movement in the degree of freedom in question. A speedometer generator TG and a resolver RES are mechanically connected to this motor M.

FIG. 10 thus shows a servo system that controls movement in one of the degrees of freedom or axes of the robot. Each one of the other degrees of freedom or axes is equipped with the same or similar type of servo.

In prior art robots of this type, the gains of the position and velocity controllers are selected and defined in such a way that stability of the feedback control system is achieved in all actual actuations.

Thus, the gain must be selected taking into account the exact operating case, and in any other operating case the servo system will operate with unnecessarily low feedback control performance.

According to this embodiment of the invention, at least in a first degree of freedom a calculation element is arranged for calculating the mechanical moment of inertia of the robot moving in that degree of freedom. The computing element is supplied with information regarding the position of the robot in at least one of its other degrees of freedom and information regarding the loading of the robot.

Depending on this information, the calculation element forms a quantity representing the mechanical moment of inertia of the robot, taking into account the aforementioned first degree of freedom. This quantity is fed to a gain control member adapted to control the gain of the position controller so that it is reduced with increasing moment of inertia. According to one embodiment, the gain of the position controller is controlled such that the position feedback control system operates with a constant overswing regardless of the current moment of inertia. Preferably, the gain of the position controller is thereby controlled to vary inversely with the moment of inertia.

Since the moment of inertia of a given motion is significantly influenced by the mass of the load supported by the robot's hands, the preferred embodiment provides that the calculation member is supplied with information about the magnitude of the mass of the load and takes this mass into account. Calculate the moment of inertia. Information regarding the size of the mass can then be appropriately input manually with the aid of a suitable input member.

In industrial robots, where it is necessary to take into account the rigidity of the mechanical part of the robot and whose servo system includes a dependent velocity loop tube according to the invention, the gain control member is dependent on the calculated moment of inertia and this Preferably the speed feedback control system is arranged to control the gain of the speed controller so that it operates with a constant damping regardless of the current moment of inertia so as to increase with increasing moment of inertia.

Next, the gain of the speed controller is preferably controlled to vary in proportion to the square root of the moment of inertia, while at the same time the gain of the position controller is controlled to vary inversely to the square root of the moment of inertia. FIG. 11 shows a method of designing a servo system for one of the robot axes according to the present invention. The servo systems of all other degrees of freedom, or preferably only some of them, are designed in a corresponding manner. Calculation member 426 is Kuzat's moment of inertia)
It is designed to calculate J. Computing element 426 can be of analog or digital type, in the latter case for example a microprocessor. Alternatively, the calculation of J is performed with the help of a computer sub-program that is permanently included in the control system of the robot. The calculation element 426 supplies information regarding the current position of the robot in the degree of freedom that affects the moment of inertia of the axis to which the servo system belongs, as shown in FIG. In the form of the robot shown in FIG. 9, the movements of the different degrees of freedom consist of rotations, and the information ψ1 about the positions of the different axes is then obtained from angular position transducers, for example resolvers of the different axes.

The calculation element 426 is also provided with information regarding the mass of the load supported by the robot's hand. For certain operations, the magnitude of the load is generally known and manually defined as input into calculation member 426.

FIG. 11 schematically shows how this definition is carried out with the help of a potentiometer 427. Digital calculation member 426
In this case, the magnitude of the load is instead suitably entered by means of a digital input member, for example numeric keys arranged on the programming unit 411 shown in FIG. Furthermore, the control system is adapted to change values with respect to the size of the load l when the object is grabbed or dropped and when objects with different masses are handled during the work cycle.

Calculation of the moment of inertia is done more or less approximately as required. As an example of the calculation of J, it is considered that the servo system shown in FIG. 11 controls the rotation angle ψ of the rotary shaft A in FIG. The following assumptions are made: the moment of inertia of the column 42 with 122) placed on it is constant and is denoted by J2, - the load 47 is approximated as a point mass, - the arms 43 and 44 and robot hand 45-4
As far as 6 is concerned, their centers of gravity are placed at 1/2 of their lengths, and their moment of inertia of the axis of rotation lying through the single center of gravity in the plane of the paper and perpendicular to the longitudinal axis of each arm is J3G%. The other moments of inertia of one arm and hand are omitted, and the masses of one arm and hand are respectively expressed as MS%M4 and M5.

Given these assumptions and symbols shown above and in FIG.
Isjn"β+"4(Llaj, nα10-5inβ)2
J5 Dew J, 5gain"r + M5 (Llainα
+L2sinβ+-5inr)"JL -ML(L15
1nα10L2sinβ+L3sinj)”J-J2+J
3+J4+J5+Jj (Current Moment of Inertia) The above-mentioned calculations regarding J are merely examples, and the calculations may be more or less approximated if necessary.

The output signal from calculation member 426 is calculated by the quantity K according to the following formula:
It is fed to a function generator 42B which calculates a'.

J' Ka -- where J' is the maximum moment of inertia of the robot axis in question. The quantity Ka is fed to a multiplier 429 whose other input is fed with the velocity reference value Vr from the position controller 422 . The output signal vr' from multiplier 429 thus follows the following relationship: J'``r''' Ka ' vr -'' vr multiplier 429
The output signal vr' of
The remainder of the servo system corresponds to that shown in FIG.

The embodiment of the position feedback control system according to the invention shown in FIG. 11 is particularly suitable for robots where the different mechanical parts are considered to be fully rigid binits.

In such a robot, the optimal control characteristics are determined by the feedback control system according to the invention within the entire work envelope and regardless of the current moment of inertia, i.e. optimal start-up time, constant overswing, and constant damping. obtained with the help of

This involves a significant improvement over prior art robots in which such optimal properties are obtained only at a single point.

As mentioned at the outset, mechanical parts, especially in large robots, can no longer be considered completely rigid units, but exhibit a certain tenacity or elasticity. FIG. 12 shows a position feedback control system according to the invention made for such a robot. 11th
In a manner similar to that described above with respect to the figures, the calculation member 426 calculates the moment of inertia of the robot axis controlled by the servo system)
will be calculated. The function generator 428 is adapted to calculate the quantity Kl, with the exception that the quantity Xa is calculated by a multiplier 429 arranged after the position controller 422.
, where the speed reference value vr is multiplied by Ka so that the output signal vr' of the multiplier 429 becomes . The output from function generator 430 is read. The output signal U' from the multiplier 431 is as follows.

In the case of a slender robot, the available resources of the feedback control system shall be utilized to the maximum within the total work envelope and regardless of the current moment of inertia of the feedback control system according to Figure 12. It is clear that Thus, optimum rise time, constant overswing and constant damping are obtained over the entire working envelope.

Furthermore, with respect to FIG. 11, oscillations of the spiky system are avoided despite the fact that the available moments are utilized in a better way than in prior art robots.

FIG. 16 shows a preferred embodiment of the feedback control system according to the invention shown in FIG. Thanks to the help of the two multipliers 429 and 431 shown in FIG. 12, the effective gain of the position and speed controller is changed depending on the moment of inertia.
These multipliers are one multiplier 429 in FIG.
has been replaced by The speed response y and the following quantities are supplied to this input.Thus, the output signal d of the multiplier 429 is as follows.The feedback control system according to FIG. 16 operates in exactly the same way as the system according to FIG. It can be shown in a simple way that feedback provides the same significant improvement in control characteristics. However, in the system according to FIG. 13, this improvement is obtained very simply, ie with the aid of one multiplier 429.

In the embodiments of the invention described with respect to FIGS. 9-13, the effective gains of the position and velocity controllers are the outputs of the respective controllers for the embodiments shown in FIGS. The same result can be obtained by multiplying the signal by a factor depending on the moment of inertia, and in the case of the embodiment shown in FIG. 13, by multiplying the velocity response by a factor depending on the moment of inertia. changed by a multiplier. However, it is a self-evident fact that the change in gain depending on the moment of inertia can be obtained by methods other than the above.

Although in the above example the invention was described with respect to an industrial robot, the total number of degrees of freedom consists of rotation of the rotary shaft around different axes. The "position" concept then relates to the angle of rotation with respect to a certain reference position, and the "velocity" concept then relates to the speed of rotation. However, the invention can also be applied to such robot types known per se, in which certain degrees of freedom consist of translational rather than rotational movements. The concepts "position, position" and "velocity" are then given their conventional meanings, namely linear translation of motion and linear velocity.

Although the invention has been described above with respect to one form of industrial robot, it is of course applicable to other forms of robots, for example having a different mechanical structure and number of degrees of freedom than the robots described above.

In a manner known per se, the feedback control system for the different axes of the robot is based either on analogue or digital methods or as a combination of these two methods.
can be designed. However, regardless of the design of the feedback control system, the present invention can be applied and provides significant advantages.

[Brief Description of the Drawings] Figure 1 is a schematic diagram of an industrial robot to which the present invention is applied;
Figure 2 shows the structure of the robot's control system, the robot's operating unit, and its connections to the robot's drive top and bottom; Figure 3 shows the operation mode of the robot axis controller;
FIG. 5 shows the sequence of calculations and data flow in the method and apparatus according to the invention; FIG. FIG. 6 is a structural diagram of the parts of the robot engine equipment according to the present invention;
FIG. 7 is a diagram showing an example of how the time of movement between two points is reduced according to the invention; FIG. 8 is a general diagram of the functions included in the method or robot according to the invention; FIGS. FIG. 11 is a diagram showing a robot and a control system according to another embodiment of the present invention, FIG. 9 is a schematic diagram of the structure of the robot and its control system, and FIG. FIG. 11 is a diagram showing a shaft servo design method according to the present invention, and FIGS. 12 and 13 are diagrams showing two modifications of this embodiment. Explanation of main symbols: 1- Mechanical part of robot: 2- Control unit; 3- Operating unit: MO- Pivoting motor: PR- Position controller; ■
-Speed controller; 22.23-Control fan computer

Claims (19)

[Claims] (1) A plurality of motion axes (1-6), a drive motor (MO) for each axis, and a desired value (θr) supplied to the controller
Axis controllers (PR, VR) that control the axis movement according to
and computer equipment for robot control (22, 23)
A method for optimal parameter control of an axis controller of an industrial robot (1-3) comprising Thus, a maximum acceleration/deceleration value (■i) for at least one axis is constantly calculated during operation of the robot, and at least one control parameter (Kp) or one path planning parameter (|■i| _c) or both are selected depending on the calculation. (2) For each axis, based on a mathematical model, a mathematical relationship is defined that describes the dynamics of the axis in terms of the expected maximum motor torque, and from these relationships the maximum achievable acceleration/deceleration ( 2. A method according to claim 1, characterized in that i) is calculated. (3) For at least one axis, each calculated maximum acceleration/deceleration value represents the maximum possible acceleration/deceleration that can be used in the current robot geometry while maintaining good control characteristics in the control system of the axis. is compared with a predetermined value (|■i|_m_a_x) selected as The method according to claim 2, characterized in that the selection is made according to the lower of the values. (4) The mathematical model is created taking into account at least one of the following quantities: mass inertia, gravity, combined mass inertia, centrifugal force, and Coriolis force. A method in which any one of Items 1 to 3 is described. (5) The robot model is determined by at least one point mass (mi) per robot arm and at least one point mass of the load carried by the robot, and the positions of these masses are determined by a number of different robot geometries. Claim 4, characterized in that the mass inertia is calculated on the basis of experimentally determined values of the acceleration and motor torque of the robot in the case of the robot model. method. (6) The robot model is determined by at least one point mass per robot-arm and at least one point mass of the load carried by the robot, and the positions of these masses are determined by the motors of the robot in case of different geometries.・
It is calculated based on the experimentally measured values of the torque, which allows the influence of gravity and the mutual influence between the axes to be
6. A method according to any of claims 4 and 5, characterized in that the method is calculated from a model. (7) In industrial robots where the axis controller of an axis includes a position controller (PR), the acceleration/deceleration values calculated from the mathematical model are each multiplied by a multiplier and then immediately constitute the gain of the position controller. A method according to any one of claims 1 to 6, characterized in that: (8) The gain of the position controller is varied by the difference between the motor current in question and the maximum permissible motor current so as to optimally use the available motor torque. The method described in Section 7. (9) In an industrial robot that includes a device configured to supply a desired position value (θr) to an axis controller in such a manner as to obtain a constant acceleration/deceleration during each acceleration/deceleration, the acceleration/deceleration value is 9. A method according to any one of claims 1 to 8, characterized in that each acceleration/deceleration value is determined to be equal to an acceleration/deceleration value calculated from a mathematical model. (10) The acceleration value is varied by the difference between the current in question and the maximum permissible motor current, in order to optimally utilize the available motor torque. Method by description. (11) When the axis controller includes a speed controller (VR),
Claim 1, characterized in that the control parameter (Kv) of the speed controller is varied by a mathematical model that compensates for damping changes due to changes in mass inertia and effective spring constant, as well as by the robot geometry. A method according to any one of Items 1 to 10. (12) For at least one first degree of freedom (y),
Depending on information about the current position of the robot in at least one further degree of freedom, a quantity (Ka) is made representative of the mechanical moment of inertia (J) of the robot in motion in said first degree of freedom and the axis control 2. A method according to claim 1, characterized in that the gain of the device is controlled by said quantity in such a way that it decreases with increasing moment of inertia. 13. A method according to claim 12, characterized in that the gain of the axis controller is controlled such that the feedback control system operates with a constant overswing with a varying moment of inertia. (14) In a robot where the position control system includes a subordinate velocity loop with a velocity transducer (TG) and a velocity controller (24), the moment of inertia is adjusted such that the gain of the velocity controller increases with the increase in the moment of inertia. 13. The method according to claim 12, characterized in that it is controlled by the expressed quantity (Ka). 15. A method according to claim 14, characterized in that the gain of the speed controller is controlled such that the control system operates with a constant damping at varying moments of inertia. (16) A method according to claim 14, characterized in that the gain of the speed controller is controlled to vary in proportion to the square root of the moment of inertia. (17) A method according to claim 16, characterized in that the gain of the position controller is controlled to vary in inverse proportion to the square root of the moment of inertia. (18) The velocity response (v) from the velocity converter is multiplied by an amount proportional to the square root of the moment of inertia before being compared with the output signal (Vr) of the position controller. A method according to scope item 17. (19) Multiple motion axes and a drive motor for each axis (MO
) and an axis controller (PR, VR) that controls axis motion according to a position reference value (θr) supplied to the controller, and computer equipment (22, 23) for robot control. (1-3) A device that performs optimal parameter control of the axis controller, which controls at least one axis based on a mathematical model representing the static and dynamic characteristics of the robot and based on the expected maximum torque of the drive motor. and at least one control parameter (Kp) or one path planning parameter (|■i|_c) in at least one axis controller; ) or both of them depending on the calculation; and a control device comprising: