DE102023117895A1 - Method for controlling a robot device - Google Patents

Abstract

The invention relates to a method for controlling a robot device, in particular an industrial robot, with two or more joints each actuated by a fluidic drive, each drive having two pressure chambers with valves or valve devices for filling and emptying the pressure chambers, the method comprising setting a target fluid mass flow on at least one of the valves, taking into account a non-linearity compensation vector and a joint coupling matrix, such that the joints are actuated by the drives, taking into account predetermined joint target values, such that the robot device follows a predetermined trajectory. A device for controlling a robot device and a computer program product are also provided.

Description

The invention relates to a method for controlling a robot device, as well as a device and a computer program product.

Robots with fluid drives are, for example, from the EN 10 2018 200 413 The fluidic drive provides that each joint of the robot is moved by a drive that has two pressure chambers into which the drive fluid can be introduced as a mass flow via controlled valves or can be drained from. The pressure change in one or both pressure chambers results in a pressure difference. This pressure difference then results in a drive force or torque that is applied to a drive element, causing the joint to move. The drive fluid can be a gas, a gas mixture, a liquid or a liquid mixture. For example, air or oil can be used.

The use of pneumatics to shift the position of elements of the arm of an industrial robot is known from JE Bobrow and BW McDonell, “Modelling, Identification, and Control of a Pneumatically Actuated, Force Controllable Robot”. A swivel drive suitable for shifting the position of elements of the arm of an industrial robot is described, for example, in the DE 195 10 11 488 C2 Its application in an industrial robot is shown, for example, in the EN 10 2017 202 369 B3 out.

Such robots can exhibit highly nonlinear behavior and are therefore currently controlled in a two-stage process.

This is described, for example, in Hoffmann, K., et. al. “On trajectory tracking control of fluid-driven actuators”. In a first step, a target torque is calculated from a target path on which a robot arm or the joint is to move. In a second step, the required target mass flow of the fluid for the drives is determined from this. This mass flow is then adjusted using valves.

This two-step calculation method has a limited bandwidth due to the cascaded controller structure, because the target torque is calculated first as an intermediate step, which is then set by an actuator, sometimes also called an actuator, using a subordinate torque controller. However, due to their physical properties, fluidic actuators require calculation methods with a higher bandwidth. Significant friction torques occur in robots with fluidic drives. The combination of friction and the limited bandwidth of the two-step calculation method impairs the path accuracy.

Disturbance torques, such as friction torques, act physically at the acceleration level. In the known solutions, no feedback of measured torques or accelerations is used. Accordingly, disturbance torques cannot be corrected directly, but only indirectly through the resulting position error.

The object of the invention is therefore to overcome the disadvantages of the prior art and to improve the path accuracy. The invention therefore represents a further development of the prior art.

A method for controlling a robotic device, as well as a device and a computer program product are provided.

The object of the invention is solved by the subject matter of the independent claims. Advantageous further developments can be found in the dependent claims.

This object is achieved in a first embodiment in particular by using a method for controlling a robot device, in particular an industrial robot, with two or more joints each actuated by a fluidic drive, wherein each drive has two pressure chambers with valves or valve devices for filling and emptying the pressure chambers, wherein the method comprises setting a target fluid mass flow at at least one of the valves, taking into account a non-linearity compensation vector and a joint coupling compensation matrix, such that the joints are actuated by the drives, taking into account predetermined joint target values, such that the robot device follows a predetermined trajectory.

The holistic calculation for mechanics and actuator dynamics is used to calculate one or more target fluid mass flows directly from a given trajectory, without calculating target torques in an intermediate step. The fact that the coupled mechanical and fluidic model of the robot are used together as the basis for the model-based control design results in structural advantages.

The target fluid mass flows to be set are also called control signals in this document, since the target fluid mass flows are transmitted as a signal to the valves so that they set the corresponding target fluid mass flow.

According to an advantageous development, the joint target values for one or more joints comprise at least one each of a target angle, a target angular velocity, a target angular acceleration, a target angular jerk, a target mean pressure and/or a first derivative of the target mean pressure with respect to time.

This provides increased flexibility, as a variety of possible specifications can be used to determine target fluid mass flows.

According to a further advantageous development, the setting of the target fluid mass flow for each valve comprises a pilot control and a control.

Because the adjustment includes feedforward control and control, the feedforward control can lead to better follow-up behavior, i.e. following the specified trajectory, and the feedback has better disturbance behavior, i.e. follow-up behavior in the event of model errors or other disturbances.

According to a further advantageous development, the feedforward control is carried out on the basis of target angles and the control is carried out on the basis of actual-target differences of angles, angular velocities and angular accelerations of the joints.

This allows the control to be designed linearly, i.e. as for a linear system, even though the system as a whole behaves non-linearly.

According to a further advantageous development, the method further comprises recording actual chamber pressures in the pressure chambers of the joint and determining an actual mean pressure by averaging the recorded actual chamber pressures of the joint. If the target mean pressure of the joint is not available, determining the target mean pressure from the first derivative of the target mean pressure of the joint is also included. The method further comprises determining a differential mean pressure as the difference between the actual mean pressure and the target mean pressure of the joint. The pilot control is carried out on the basis of the first derivatives of the target mean pressures of the joints, and the control is carried out on the basis of the differential mean pressures of the joints.

The additional consideration of mean pressures offers a further improvement.

According to a further advantageous development, the pre-control also includes friction compensation.

By taking friction into account in the form of friction compensation, the accuracy of following the given trajectory is further improved.

According to a further advantageous development, the non-linearity compensation vector is formed as a function of the angles and angular velocities as well as the chamber pressures of the joints.

According to a further advantageous development, the nonlinearity compensation vector takes into account Coriolis and centrifugal forces, gravitational moments, friction moments as well as drive moments and compression effects.

wobei

q

der Vektor der Winkel der Gelenke,

M(q)

die Massenmatrix,

C(q, q̇)q̇

der Vektor der Coriolis- und Zentrifugalkräfte,

g(q)

der Vektor der Gravitationsmomente,

τf-

der Vektor der Reibungsmomente,

τ̇p,V̇

der Vektor der Momentenänderungen durch Kompression und/oder Expansion,

und

ṗm,V̇

die Änderungen der Ist-Mitteldrücke der Gelenke aufgrund von Kompression bzw. Expansion darstellen.

According to a further advantageous development, the nonlinearity compensation vector a is described by $$a=\left[\underset{{\dot{p}}_{m,\dot{V}}}{\frac{\text{d}}{\text{d}t}\left(M^{-1}\right)M\ddot{q}+M^{-1}\left(-\dot{C}\dot{q}-C\ddot{q}-\dot{G}-{\dot{\tau }}_{\text{e}}+{\dot{\tau }}_{p,\dot{V}}\right)}\right]$$

where

q

the vector of the angles of the joints,

M(q)

the mass matrix,

C(q, q̇)q̇

the vector of Coriolis and centrifugal forces,

g(q)

the vector of gravitational moments,

τf-

the vector of friction moments,

τ̇p,V̇

the vector of moment changes due to compression and/or expansion,

and

ṗm,V̇

represent the changes in the actual mean pressures of the joints due to compression or expansion.

By forming the nonlinearity compensation vector according to one of the above ways, the best results in accuracy were achieved.

According to a further advantageous development, the joint coupling compensation matrix takes into account a polytropic exponent, a specific gas constant and the temperature of the drive fluid of the joints, and block diagonal matrices of the drive geometry.

wobei

M(q)

die Massenmatrix,

η

der Polytropenexponent des Antriebsfluids der Gelenke,

R

die spezifische Gaskonstante des Antriebsfluids der Gelenke,

T

die Temperatur des Antriebsfluids der Gelenke,

und

Λ1, Λ2

Blockdiagonalmatrizen der Antriebsgeometrie darstellen.

According to a further advantageous development, the joint coupling compensation matrix κ ^-1 is formed as the inverse of a joint coupling matrix, which is described by $$\kappa =\left[\begin{array}{c}\eta RT{M}^{-1}{\mathrm{\Lambda }}_1\\ \frac{\eta RT}{2}{\mathrm{\Lambda }}_2\end{array}\right],$$

where

M(q)

the mass matrix,

η

the polytropic exponent of the drive fluid of the joints,

R

the specific gas constant of the drive fluid of the joints,

T

the temperature of the drive fluid of the joints,

and

Λ1, Λ2

Represent block diagonal matrices of the drive geometry.

By forming the joint coupling compensation matrix according to one of the above manners, the best results in accuracy were achieved.

wobei

κ

die Gelenk-Kopplungsmatrix ist,

a

der Nichtlinearitäten-Ausgleichsvektor ist,

und

v

Soll-Werte und Ist-Werte der Robotervorrichtung umfasst.

According to a further advantageous development, the target fluid mass flows u at the valves are set by $$u\left(x\right)={\kappa }^{-1}\left(x\right)\left(\nu -a\left(x\right)\right),$$

where

κ

is the joint coupling matrix,

a

is the nonlinearity compensation vector,

and

v

Includes target values and actual values of the robot device.

By inverting the joint coupling matrix for the coupled mechanics and fluidics, the required control signals, i.e. the target fluid mass flows, for the actuators are determined from the desired target path.

According to a further embodiment, a device is provided for controlling a robot device, in particular an industrial robot, with two or more joints each actuated by a fluidic drive, wherein each drive has two pressure chambers with valves for filling and emptying the pressure chambers, wherein the device is designed to carry out the steps of a previously described method.

According to a further embodiment, a computer program product is provided with a program for a robot controller, the computer program product comprising software code sections for carrying out the steps according to a previously described method when the program is executed on the robot controller.

According to a further advantageous development, the computer program product comprises a computer-readable medium on which the software code sections are stored, wherein the program can be loaded directly into an internal memory of the robot controller.

The advantageous further developments are explained in the further course of the description.

The embodiments and developments show possible variants, whereby the invention is not limited to the specifically illustrated variants, but rather combinations of the individual variants with one another are also possible. In particular, the features of the device can also be implemented in a method and vice versa.

• 1 beispielhaft ein Gelenk einer Robotervorrichtung 10 gemäß dem Stand der Technik, und
• 2 ein Flussdiagramm des Verfahrens zum Steuern einer Robotervorrichtung 10.

For a better understanding of the invention, it is explained in more detail using the following figures. They each show in a highly simplified, schematic representation:

• 1 by way of example, a joint of a robot device 10 according to the prior art, and
• 2 a flow chart of the method for controlling a robot device 10.

To begin with, it should be noted that in the different embodiments described, identical parts are provided with identical reference symbols or identical component designations, whereby the disclosures contained in the entire description can be transferred analogously to identical parts with identical reference symbols or identical component designations. The position information chosen in the description, such as top, bottom, side, etc., also refers to the figure directly described and shown, and these position information must be transferred analogously to the new position if the position changes.

In 1 an exemplary joint of a robot device 10 can be seen, which is in particular an industrial robot. The robot device 10 comprises a first robot member 1, a second robot member 2 and a fluidic drive device for driving the second robot member 2 relative to the first robot member 1. A separate control device 9 can be provided for the robot device 10, but this can also be integrated. The control device 9 is designed to provide control of a position and/or movement of the robot device 10, in particular of the second robot member 2, under control of the fluidic drive device. The fluidic drive device comprises a drive body 4 and a drive member 5 that can be driven relative to the drive body 4 and that can rotate about an axis of rotation 23 relative to the drive body 4. The drive body 4 has a cavity that is divided by the drive member 5 into the two pressure chambers 6, 7. A circular cavity is shown, alternative designs also exist. The pressure chambers 6, 7 can be separated from each other by a partition wall 8. The fluidic drive device also has two valve devices, or valves, 16, 17, whereby the valves 16, 17 for the two pressure chambers 6, 7 can be actuated independently of one another. One valve device 16 or 17 is assigned to one pressure chamber 6 or 7.

In order to move the joint, the drive fluid is transported from a pressure fluid source 14 via the supply line 15 and the feed lines 11, 12 to the valve devices 16, 17 and then further into the pressure chambers 6, 7 or is discharged from them. In addition, a pressure sensor device 19 can be provided which can detect the pressure, in particular the absolute pressure or relative pressure, of each of the pressure chambers 6, 7. This can be implemented, for example, by two pressure sensor elements 21, 22, which can be designed as absolute pressure sensor elements or relative pressure sensor elements. One pressure sensor element 21 or 22 is assigned to each pressure chamber 6 or 7. The pressure sensor elements 21, 22 can, as shown, be provided outside the pressure chambers 6, 7 or, alternatively or in addition to the design shown, can be arranged inside the pressure chambers 6, 7.

A rotary encoder 24 can be provided to detect the angle of rotation and/or the rotary movement of the second robot link 2 relative to the first robot link 1. This allows the actual position of the robot links 1, 2 to be detected and a corresponding control signal to be fed back. The rotary encoder 24 is designed as an encoder, for example, and is expediently integrated in the joint.

2 shows a flow chart of the method 100, wherein the step of setting 110 a target fluid mass flow can include a feedforward control 120 and a control 130.

The setting 110 of a target fluid mass flow at at least one of the valve devices 16, 17 is then carried out taking into account a non-linearity compensation vector and a joint coupling compensation matrix, such that the joints are actuated by the drives taking into account predetermined joint target values such that the robot device 10 follows a predetermined trajectory.

Detailed explanations can be found below, whereby the individual options are discussed here only in their best form.

wobei

q ∈ ℝn mit n = 6

der Vektor der Gelenkwinkel,

M(q)

die Massenmatrix,

C(q, q)q

der Vektor der Coriolis- und Zentrifugalkräfte,

g(q)

der Vektor der Gravitationsmomente,

τf

der Vektor der Reibungsmomente,

τ(p)

der Vektor der Antriebsmomente ist.

The equations of motion of a robot are described by equation A. In this text, without loss of generality, a robot device with 6 joints or swivel drives is considered as an example. $$M\left(q\right)\ddot{q}+C\left(q.\dot{q}\right)\dot{q}+G\left(q\right)=\tau \left(p\right)-{\tau }_{\text{e}}.$$

where

q ∈ ℝn with n = 6

the vector of the joint angles,

M(q)

the mass matrix,

C(q, q)q

the vector of Coriolis and centrifugal forces,

g(q)

the vector of gravitational moments,

τf

the vector of friction moments,

τ(p)

is the vector of the driving torques.

The drive torques are caused by pressure differences between the pressure chambers 6, 7.

Each of the six rotary actuators with index i therefore comprises two pressure chambers 6,7, which are separated by the drive element 5 with an area A _i and an effective radius r _m,i . A pressure difference between the chambers of the drive i generates the drive torque according to formula B. $${\tau }_i\left(p_{i\mathrm{,1}},p_{i\mathrm{,2}}\right)=A_i{r}_{m,i}\left(p_{i\mathrm{,1}}-p_{i\mathrm{,2}}\right).$$

wobei

Vi,j(qi)

das Kammervolumen der Druckkammern 6, 7,

η

der Polytropenexponent des Antriebsfluids der Gelenke,

R

die spezifische Gaskonstante des Antriebsfluids der Gelenke, und

T

die Temperatur des Antriebsfluids der Gelenke ist.

For the pneumatic subsystem, the following pressure differential equation C is assumed for each chamber j ∈ {1, 2} of the actuator i. $${\dot{p}}_{i,j}=\frac{\eta }{V_{i,j}\left(q_i\right)}\left(RT{\dot{m}}_{i,j}-{\dot{V}}_{i,j}\left({\dot{q}}_i\right)p_{i,j}\right)$$

where

Vi,j(qi)

the chamber volume of the pressure chambers 6, 7,

η

the polytropic exponent of the drive fluid of the joints,

R

the specific gas constant of the drive fluid of the joints, and

T

the temperature of the drive fluid of the joints.

The fluid mass flows ṁ _i,j of the individual valve devices 16, 17 are referred to collectively as u in the following, see also formula D. $$u={\left[{\dot{m}}_{1.1}{\dot{m}}_{1.2}{\dot{m}}_{2.1}\cdots {\dot{m}}_{n\mathrm{,2}}\right]}^{\text{T}}{ℝ}^{2n}$$

A positive sign describes a mass flow into the pressure chamber 6, 7, a negative sign describes a mass flow out of the pressure chamber 6, 7.

The measured quantities for each drive can be the joint angle q _i and the chamber pressures p _i,1 and p _i,2 .

Then a state vector for the entire system consisting of mechanics and pneumatics can be described according to formula E: $$x={\left[q^{\text{T}}\cdot {\dot{q}}^{\text{T}}\cdot p^{\text{T}}\right]}^{\text{T}}\in {ℝ}^{4n}\text{with}p={\left[p_{1.1}{p}_{1.2}{p}_{2.1}\dots p_{n\mathrm{,2}}\right]}^{\text{T}}\in {ℝ}^{2n}$$

wobei e_k ∈ ℝ²ⁿ der k-te Einheitsvektor ist, und die Argumente der Übersichtlichkeit halber nicht dargestellt sind. Darin beschreibt der untere Teil die Druckdynamik je Druckkammer 6,7, in die die Fluid-Massenströme eingehen. Die Drücke der Druckkammern 6, 7 sind untereinander nicht verkoppelt, eine Verkopplung besteht jedoch über die Mechanik sowie zwischen Mechanik und Druckdynamik.The state variables q and p can be measured, the joint angular velocities q can be determined by filtering and differentiating q. The entire mechanics and pressure dynamics form a nonlinear input-affine multiple-input-multiple-output (MIMO) system according to the formula F. $$\dot{x}=\left[\begin{array}{c}\dot{q}\\ M^{-1}\left(-C\dot{q}-G-{\tau }_{\text{e}}+\tau \left(p\right)\right)\\ -\eta \text{diagnostics}\left(\frac{{\dot{V}}_{1.1}}{V_{1.1}},\frac{{\dot{V}}_{1.2}}{V_{1.2}},\frac{{\dot{V}}_{2.1}}{V_{2.1}},\cdots ,\frac{{\dot{V}}_{n\mathrm{,2}}}{V_{n\mathrm{,2}}}\right)p\end{array}\right]+{\displaystyle \sum _{k=1}^{2n}\left[\begin{array}{c}{0}_{n\times 1}\\ 0_{n\times 1}\\ \frac{\eta RT}{V_k}{\text{e}}_k\end{array}\right]}{u}_k,x\left(0\right)=x_0,$$

where e _k ∈ ℝ ²ⁿ is the k-th unit vector, and the arguments are not shown for the sake of clarity. The lower part describes the pressure dynamics for each pressure chamber 6,7 into which the fluid mass flows enter. The pressures of the pressure chambers 6, 7 are not coupled to each other, but there is a coupling via the mechanics and between mechanics and pressure dynamics.

gehen die Massenströme explizit ein. Durch zeitliche Differentiation der zweiten Zeile der Formel F ergibt sich die dazugehörige Formel G. $$\stackrel{⃛}{q}=\frac{\text{d}}{\text{d}t}\left(M^{-1}\right)M\ddot{q}+M^{-1}\left(-\dot{C}\dot{q}-C\ddot{q}-\dot{g}-{\dot{\tau }}_{\text{f}}+{\dot{\tau }}_{p,\dot{V}}\right){\dot{p}}_{m,\dot{V}}+\eta RT{M}^{-1}{\Lambda }_{1}u$$

It is clear that the mass flow u is not included in the two upper lines of the formula F, so it has no direct effect on the differential equations of the velocity q̇ and the acceleration q̈. Only when the position is differentiated three times with respect to time, i.e. in the jerk differential equation $\stackrel{⃛}{q},$

the mass flows are explicitly included. Differentiating the second line of formula F over time yields the corresponding formula G. $$\stackrel{⃛}{q}=\frac{\text{d}}{\text{d}t}\left(M^{-1}\right)M\ddot{q}+M^{-1}\left(-\dot{C}\dot{q}-C\ddot{q}-\dot{G}-{\dot{\tau }}_{\text{e}}+{\dot{\tau }}_{p,\dot{V}}\right){\dot{p}}_{m,\dot{V}}+\eta RT{M}^{-1}{\Lambda }_{1}u$$

Therein, the vector τ̇ _p,V̇ describes the change in the drive torque τ(p) due to compression or expansion in the pressure chamber 6, 7. The components of the vector τ̇ _p,V̇ ̇ are described by the formula H. $${\dot{\tau }}_{p,\dot{V},i}=-A_i{r}_{m,i}\eta \left(\frac{p_{i\mathrm{,1}}{\dot{V}}_{i\mathrm{,1}}}{V_{i\mathrm{,1}}}-\frac{p_{i\mathrm{,2}}{\dot{V}}_{i\mathrm{,2}}}{V_{i\mathrm{,2}}}\right)$$

The drive geometry is entered via a block diagonal matrix, as described in formula I. $${\Lambda }_1=\text{diagnostics}\left(A_1{r}_{m\mathrm{,1}}\left[\frac{1}{V_{1.1}}\frac{-1}{V_{1.2}}\right],\cdots ,A_n{r}_{m,n}\left[\frac{1}{V_{n\mathrm{,1}}}\frac{-1}{V_{n\mathrm{,2}}}\right]\right)$$

This means that a third-order differential equation has been found. The total dimension of the system of equations F is 4n. According to system theory, another input-output relationship can be found in the form of a first-order differential equation in order to clearly describe the dynamics of the overall system.

As an additional degree of freedom, the mean pressure p _m = (p ₁ + p ₂ )/2 can be introduced, which satisfies the differential equation J. $$\dot{p}={\dot{p}}_{m,\dot{V}}+\frac{\eta \mathrm{<figure-callout\; id="2"\; label="R\; T"\; filenames="DE102023117895A1\_0022.png"\; state="\{\{state\}\}">R</figure-callout>}T}{2}{\Lambda }_{2}u$$

Therein, the vector ṗ _m,V̇ describes the change in the mean pressure due to compression or expansion in the pressure chambers 6, 7. The components of the vector ṗ _m,V̇ are described by the formula K. $${\dot{p}}_{m,\dot{V},i}=-\frac{\eta }{2}\left(\frac{p_{i\mathrm{,1}}{\dot{V}}_{i\mathrm{,1}}}{V_{i\mathrm{,1}}}+\frac{p_{i\mathrm{,2}}{\dot{V}}_{i\mathrm{,2}}}{V_{i\mathrm{,2}}}\right)$$

The drive geometry is entered via a block diagonal matrix as described in formula L. $${\Lambda }_{}$$$${}_2=\text{diagnostics}\left(\left[\frac{1}{V_{1.1}}\frac{1}{V_{1.2}}\right],\cdots ,\left[\frac{1}{V_{n\mathrm{,1}}}\frac{1}{V_{n\mathrm{,2}}}\right]\right)$$

A system of joints in a robot device 10 can be described according to formula M, where the relationship between the joint angle q and the mean pressure p _m is apparent. $$\left[\begin{array}{c}\stackrel{⃛}{q}\\ {\dot{p}}_m\end{array}\right]=\left[\underset{\alpha }{\underbrace{\begin{array}{c}\frac{\text{d}}{\text{d}t}\left(M^{-1}\right)M\ddot{q}+M^{-1}\left(-\dot{C}\dot{q}-C\ddot{q}-\dot{G}-{\dot{\tau }}_{\text{e}}+{\dot{\tau }}_{p,\dot{V}}\right)\\ {\dot{p}}_{m,\dot{V}}\end{array}}}\right]+\underset{\kappa }{\underbrace{\left[\begin{array}{c}\eta RT{M}^{-1}{\Lambda }_1\\ \frac{\mathrm{<figure-callout\; id="2"\; label="\eta \; R\; T"\; filenames="DE102023117895A1\_0022.png"\; state="\{\{state\}\}">\eta </figure-callout>}RT}{2}{\Lambda }_2\end{array}\right]}}u$$

Here, a ∈ ℝ ^2n×1 represents the vector of nonlinearities and κ ∈ ℝ ^2n×2n represents the joint coupling matrix. u denotes the target fluid mass flows to be set.

The control, ie the transmission of a control signal, of a joint is then carried out starting from the formula M by inverting the joint coupling matrix into the joint coupling compensation matrix. Starting from the above expressions, the target fluid mass flows u to be set at the valve devices 16, 17 are described by the formula N. $$u\left(x\right)={\kappa }^{-1}\left(x\right)\left(\nu -a\left(x\right)\right)$$

This removes the nonlinearities a from the original system equations, resulting in a virtual linear system. In addition, the inverted joint coupling matrix κ ^-1 removes the couplings between the joints. A linear control can then be carried out for the resulting virtual linear, decoupled system with new input v.

Conversely, this means that state-of-the-art controllers designed for individual drives can be transferred to multi-axis kinematics using the concept presented here in such a way that the couplings between the individual axes are also taken into account. This is an added value compared to simply connecting controllers for individual axes in parallel, which does not take the coupling of the axes into account.

In summary, a method 100 for controlling a robot device 10 is provided, which takes the mechanical and fluidic behavior of the robot device 10 into account holistically, and in particular maps the coupling of these domains.

By inverting the previously mentioned model equations for the coupled mechanics and fluidics, the required control signals for the actuators are determined from the desired target path. In the control engineering sense, this can be viewed as a feedforward control. In this document, control signals mean, as previously described, the target mass flow through the valves.

Another component can be the feedback of the actual joint angles, speeds and accelerations to the controller. From this feedback, the control signal in the form of a target mass flow can be determined directly on a model-based basis. This is done in a one-stage process without an intermediate step, using the equations for the two subsystems of mechanics and fluidics and their coupling. This also results in a direct access to the actuator dynamics in the feedback part of the control signal, and the feedback for the path sequence goes directly into the fluidic domain.

As is known, from the existence of one of the actual values for a joint, i.e. angle, velocity or acceleration, any of the other quantities can be determined by differential or integral calculus.

By applying the control signal determined in this way, consisting of feedforward control and feedback, the robot follows a predetermined path. This compensates not only for the non-linearities from the movement of the mechanics, but also for the non-linear fluidic properties of the actuator. The holistic approach for mechanics and actuator dynamics means that a control signal for the drives is calculated directly from a target path, without calculating a target torque in an intermediate step.

The resulting structural advantages are that with the provided method 100, the coupled mechanical and fluidic model of the robot are jointly used as the basis for the model-based control design.

By taking both physical domains into account together, the control signal (target mass flow of the fluid) is generated directly in the process when a path following error occurs. This results in a higher bandwidth of the path following controller compared to the two-stage process from the state of the art. Path errors are compensated more quickly as a result, so that the position accuracy is higher and external disturbances are better compensated for.

Through the method 100, not only the joint angles and angular velocities are fed back into the controller, but also the acceleration signal. This contains information about disturbance torques, such as friction torques, which act physically at the acceleration level, and promotes the compensation of these disturbances.

This is an advantage for fluidic actuators with significant friction, as these disturbances do not have to be compensated for by additional calculations. The direct access to the actuator dynamics, compared to an implementation at the torque level, also increases the bandwidth. Therefore, the method provided not only determines information about disturbance torques, but also compensates them more quickly by directly accessing the actuator dynamics.

Optionally, before setting 110 the target fluid mass flow, if these are not already present, the actual chamber pressures in the pressure chambers (6, 7) of the joint can be recorded in a step 101. From the recorded actual chamber pressures of the joint, an actual mean pressure can then be determined in a step 102, for example by averaging. If the target mean pressure of the joint is not available, this can be determined in a step 103 from the first derivative of the target mean pressure of the joint. Finally, in a step 104, a differential mean pressure can be determined as the difference between the actual mean pressure and the target mean pressure of the joint. This means that the pilot control 120 can also be carried out on the basis of the first derivatives of the target mean pressures of the joints and the control 130 can also be carried out on the basis of the differential mean pressures of the joints.

Alternatively or additionally, one or more of the actual joint values, i.e. joint angles, speeds and accelerations, can also be recorded in step 101. The other values can then be determined from the recorded actual joint values in steps 102 or 103 by deriving or integrating. Finally, in step 104, differences between the target and actual values can be formed. This means that a target-actual joint angle difference, a target-actual joint speed difference and a target-actual joint acceleration difference can be determined for each joint. This means that the feedforward control 120 can also be carried out on the basis of a predetermined target angular jerk and the control 130 can then also be carried out on the basis of the differences between the target and actual values of joint angles, speeds and accelerations determined in step 104.

For the detection 101 of the actual joint angle, the rotary encoder 24 can be used, for example. For the detection 101 of the actual angular velocity and/or actual angular acceleration, other known methods and devices can be used, which, however, are described in the 1 are not shown.

A further embodiment is a device 9 for controlling a robot device 10, in particular an industrial robot, with two or more joints each actuated by a fluidic drive, wherein each drive has two pressure chambers 6, 7 with valve devices 16, 17 for filling and emptying the pressure chambers 6, 7, wherein the device 9 is designed to carry out one of the methods set out above.

Another embodiment is a computer program product comprising instructions which, when executed by a computer, cause the computer to perform one of the methods set out above.

Another embodiment is a computer-readable medium on which the computer program product is stored.

The embodiments show possible embodiment variants, whereby it should be noted at this point that the invention is not limited to the specifically illustrated embodiment variants thereof, but rather various combinations of the individual embodiment variants with one another are also possible.

The scope of protection is determined by the claims. However, the description and the drawings must be used to interpret the claims. Individual features or combinations of features from the different embodiments shown and described can represent independent inventive solutions in themselves. The task underlying the independent inventive solutions can be taken from the description.

All information on value ranges in the description in question is to be understood as including any range and all subranges thereof, e.g. the information 1 to 10 is to be understood as including all subranges starting from the lower limit 1 and the upper limit 10, i.e. all subranges begin with a lower limit of 1 or greater and end with an upper limit of 10 or less, e.g. 1 to 1.7, or 3.2 to 8.1, or 5.5 to 10.

Finally, for the sake of clarity, it should be noted that in order to better understand the structure, some elements have been shown not to scale and/or enlarged and/or reduced.

List of reference symbols

1

first robot limb

2

second robot link

4

Drive body

5

Drive link

6.7

Pressure chambers

8th

partition wall

9

Device for controlling a robot device 10

10

Robot device

11, 12

Supply lines

14

Pressure fluid source

15

supply line

16, 17

Valve devices

19

Pressure sensor device

21, 22

Pressure sensor elements

23

Rotation axis

24

Encoder

100

Method for controlling a robot device 10

101

Recording of actual chamber pressures or actual joint values

102

Determining an actual mean pressure or other actual joint values

103

Determining the target mean pressure or other actual joint values

104

Determining a differential mean pressure or target-actual joint value differences

110

Setting a target fluid mass flow

120

Input taxes

130

Regulate

QUOTES INCLUDED IN THE DESCRIPTION

This list of documents listed by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions.

Cited patent literature

• EN 102018200413 [0002]
• DE 1951011488 C2 [0003]
• DE 102017202369 B3 [0003]

Claims (15)

Method (100) for controlling a robot device (10), in particular an industrial robot, with two or more joints each actuated by a fluidic drive, each drive having two pressure chambers (6, 7) with valve devices (16, 17) for filling and emptying the pressure chambers (6, 7), the method (100) comprising: Setting (110) a target fluid mass flow at at least one of the valve devices (16, 17) taking into account a non-linearity compensation vector and a joint coupling compensation matrix, such that the joints are actuated by the drives taking into account predetermined joint target values such that the robot device (10) follows a predetermined trajectory. Procedure (100) according to Claim 1 , wherein the joint target values for one or more joints comprise at least one of a target angle, a target angular velocity, a target angular acceleration, a target angular jerk, a target mean pressure and/or a first derivative of the target mean pressure with respect to time. Method (100) according to one of the Claims 1 or 2 , wherein the setting (110) of the desired fluid mass flow for each valve device (16, 17) comprises a pilot control (120) and a control (130). Procedure (100) according to Claim 3 , wherein the feedforward control (120) is carried out on the basis of desired angles; and wherein the control (130) is carried out on the basis of actual-desired differences of angles, angular velocities and angular accelerations of the joints. Procedure (100) according to Claim 3 or 4 , wherein the method (100) for one or more joints further comprises: detecting (101) actual chamber pressures in the pressure chambers (6, 7) of the joint; determining (102) an actual mean pressure by averaging the detected actual chamber pressures of the joint; if the target mean pressure of the joint is not present, determining (103) the target mean pressure from the first derivative of the target mean pressure of the joint; and determining (104) a differential mean pressure as the difference between the actual mean pressure and the target mean pressure of the joint; wherein the pilot control (120) is carried out on the basis of the first derivatives of the target mean pressures of the joints; and wherein the control (130) is carried out on the basis of the differential mean pressures of the joints. Method (100) according to one of the Claims 3 until 5 , wherein the pilot control (120) further comprises friction compensation. Method (100) according to one of the Claims 1 until 6 , where the nonlinearity compensation vector is formed depending on the angles and angular velocities as well as the chamber pressures of the joints. Method (100) according to one of the Claims 1 until 7 , where the nonlinearity compensation vector takes into account Coriolis and centrifugal forces, gravitational moments, friction moments as well as drive moments and compression effects. Method (100) according to one of the Claims 1 until 8th , where the nonlinearity compensation vector a is described by $$a=\left[\begin{array}{c}\frac{\text{d}}{\text{d}t}\left(M^{-1}\right)M\ddot{q}+M^{-1}\left(-\dot{C}\dot{q}-C\ddot{q}-\dot{G}-{\dot{\tau }}_{\text{e}}+{\dot{\tau }}_{p,\dot{V}}\right)\\ {\dot{p}}_{m,\dot{V}}\end{array}\right]$$

where q is the vector of the angles of the joints, M(q) is the mass matrix, C(q, q̇)q̇ is the vector of the Coriolis and centrifugal forces, g(q) is the vector of the gravitational moments, τ _{f is} the vector of the friction moments, τ̇ _p , _V̇ the vector of moment changes due to compression and/or expansion, and ṗ _m,V̇ represent the changes in the actual mean pressures of the joints due to compression or expansion, respectively. Method (100) according to one of the Claims 1 until 9 , where the joint coupling compensation matrix takes into account a polytropic exponent, a specific gas constant and the temperature of the drive fluid of the joints, and block diagonal matrices of the drive geometry. Method (100) according to one of the claims Claim 1 until 10 , where the joint coupling balancing matrix κ ^-1 is formed as the inverse of a joint coupling matrix described by $$\kappa =\left[\begin{array}{c}\eta RT{M}^{-1}{\Lambda }_1\\ \frac{\eta RT}{2}{\Lambda }_2\end{array}\right],$$

where M(q) is the mass matrix, η is the polytropic exponent of the drive fluid of the joints, R is the specific gas constant of the drive fluid of the joints, T is the temperature of the drive fluid of the joints, and Λ ₁ , Λ ₂ are block diagonal matrices of the drive geometry. Method (100) according to one of the Claims 1 until 11 , wherein the target fluid mass flows u at the valve devices (16, 17) are set by $$u\left(x\right)={\kappa }^{-1}\left(x\right)\left(\nu -a\left(x\right)\right),$$

where κ is the joint coupling matrix, a is the nonlinearity compensation vector, and v includes target values and actual values of the robot device. Device (9) for controlling a robot device (10), in particular an industrial robot, with two or more joints each actuated by a fluidic drive, each drive having two pressure chambers (6, 7) with valve devices (16, 17) for filling and emptying the pressure chambers (6, 7), the device (9) being designed to carry out the steps of a method according to one of the Claims 1 until 12 to carry out. Computer program product with a program for a robot controller, the computer program product comprising software code sections for carrying out the steps according to one of the Claims 1 until 12 when the program is executed on the robot controller. Computer program product according to Claim 14 , wherein the computer program product comprises a computer-readable medium on which the software code sections are stored, the program being directly loadable into an internal memory of the robot controller.

Images