METHOD AND DEVICE FOR DEFINING A MOVEMENT SEQUENCE FOR A ROBOT

Abstract

The present invention relates to a method and to a device for defining a movement sequence for a multi-axis manipulator of a robot system, which manipulator has a plurality of elements which form different rotational axes, and an end element for interaction with an effector, wherein the effector is intended to carry out at least one arbitrary operation in a working space, and wherein in order to carry out the at least one arbitrary operation the end element of the manipulator is to be transferred into an arbitrary target pose with respect to the working space, wherein the manipulator moves in a plurality of steps to the target pose while approaching the end element, and for each step at least one defined impedance pattern and/or admittance pattern is defined with respect to at least one axis which forms the axis of a coordinate system which is linked to the manipulator.

Claims

1. Method for determining a motion sequence for a multi-axis manipulator of a robot system which has a plurality of links forming a plurality of different rotational axes and an end link for interaction with an effector, the effector being intended to carry out at least one arbitrary operation in a workspace, and the end link of the manipulator being intended to be transferred into any arbitrary target pose with respect to the workspace in order to carry out the at least one arbitrary operation, comprising: moving the manipulator in several steps (Si; Sj) with the end link approaching the target pose, wherein for each step (Si; Sj) at least one defined impedance pattern and/or admittance pattern is determined with respect to at least one axis forming the axis of a coordinate system associated with the manipulator.

2. Method according to claim 1, in which the at least one axis refers to a translational orientation and/or to a rotational orientation.

3. Method according to claim 2, in which for a step (Si), a defined impedance pattern and/or admittance pattern is determined with respect to an axis in a translational orientation, and for a further step (Sj), a defined impedance pattern and/or admittance pattern is determined with respect to an axis in a rotational orientation.

4. Method according to claim 3, in which at least one of the steps (Si; Sj) or all of the steps (Si; Sj) is/are repeated n times until the target pose is reached.

5. Method according to claim 4 in which, for each step (Si; Sj) of the n-th repetition, the respectively defined impedance pattern and/or admittance pattern is maintained or varied.

6. Method according to claim 3, in which the impedance patterns and/or admittance patterns are designed to be constant, time-varying and/or sate-dependent during a step (Si; Sj).

7. Method according to claim 1, further comprising the step: determining at least one arbitrary coordinate system with respect to said manipulator.

8. Method according to claim 7, in which an arbitrary coordinate system is determined with respect to an axis member of the manipulator.

9. Method according to claim 7 in which an arbitrary coordinate system is determined with respect to a joint between two axis members of the manipulator.

10. Method according to claim 7, in which an arbitrary coordinate system is determined with respect to the effector.

11. Method according to claim 7, in which an arbitrary coordinate system is determined with respect to the workspace.

12. Method according to claim 8, in which the arbitrary coordinate system is determined as a function of the target pose.

13. Method according to claim 8, in which the arbitrary coordinate system is designed to be time-variant.

14. Method according to claim 8, in which the arbitrary coordinate system is determined as a function of the operation to be performed.

15. Method according to claim 1, further comprising the step: converting the manipulator into a gravitation-compensated state and/or centrifugal force-compensated state and/or coriolis force-compensated state and/or inertia-compensated state.

16. Method according to claim 1, in which a total impedance pattern and/or total admittance pattern for the motion sequence to be determined with respect to the target pose, which pattern(s) is/are generated after carrying out the steps (Si; Sj), is/are applied to at least one further target pose while maintaining a common orientation within the framework of the impedance behavior and/or admittance behavior, the position of the further target pose being offset relative to the position of the target pose within a common plane and/or at angularly relative thereto.

17. Computer program comprising program instructions which cause a processor to execute and/or control the steps of the method according to claim 1 when the computer program is running on the processor.

18. Data carrier device on which a computer program is stored in accordance with claim 17.

19. Computer system comprising a data processing apparatus, the data processing apparatus being arranged such that a method according to claim 1 is performed on the data processing apparatus.

20. Robot system comprising a multi-axis manipulator and an end member of the manipulator for performing an operation, comprising means for performing the method according to claim 1.

21. Device for determining a motion sequence for a multi-axis manipulator of a robot system which has a plurality of links forming a plurality of different rotational axes and an end link for interaction with an effector, the effector being intended to carry out at least one arbitrary operation in a workspace, and the end link of the manipulator being intended to be transferred into any arbitrary target pose with respect to the workspace in order to carry out the at least one arbitrary operation, wherein the device is designed such that the following steps can be carried out: moving the manipulator in several steps (Si; Sj) with the end link approaching the target pose, wherein for each step (Si; Sj) at least one defined impedance pattern and/or admittance pattern is determined with respect to at least one axis forming the axis of a coordinate system associated with the manipulator.

22. Robot equipped with a manipulator and a device as described in claim 21.

Description

[0046] Further advantages and features of the invention do become apparent from the description of the embodiments as shown in the enclosed drawings.

[0047] FIG. 1 is an exemplary illustration of a multi-axis manipulator of a robot system, in which possible coordinate systems for the method according to the invention are schematically indicated;

[0048] FIG. 2 is a flowchart illustrating the essential steps of an embodiment of the method according to the invention;

[0049] FIG. 3a is a diagram illustrating a step-by-step method according to the invention as compared to known methods; and

[0050] FIG. 3b is a scheme of the possible mutual interrelationships of individual compliance patterns, whereby only stiffnesses are provided.

[0051] FIG. 1 shows an example of a robot system with a manipulator M consisting of several axis links or members A, which are connected to each other via joints G. At the end of the manipulator M, an effector E is provided, which is to carry out a certain operation in a workspace or task space R.

[0052] Several coordinate systems can be assigned to the manipulator M, which is to take a pose x.sub.i, which are schematically shown in FIG. 1, in this case as Cartesian coordinate systems. However, other coordinate systems are also conceivable, e.g. coordinate systems associated with manifolds.

[0053] A first coordinate system C.sub.A can, for example, refer to one of the axis elements A and have corresponding axes A.sub.A within this coordinate system C.sub.A, which define this coordinate system C.sub.A.

[0054] A second coordinate system C.sub.E is directly related to the effector E and has axes A.sub.E defining this coordinate system C.sub.E accordingly.

[0055] A third coordinate system C.sub.G can refer directly to a single joint G and is defined accordingly by the axes A.sub.G.

[0056] The fourth coordinate system C.sub.R can be a coordinate system that refers to the workspace R and is defined via the corresponding axes A.sub.R.

[0057] In the Teach-in method according to the invention, the manipulator M is transferred into a pose xi in several steps S.sub.i, S.sub.j (see FIGS. 2 and 3a) by approximating the effector E or the end member of the manipulator M carrying it to the final pose x.sub.i, whereby the pose x.sub.i results from the workspace R itself, which corresponds to the location of an operation to be performed in this, e.g. an assembly workstation, and from the type of operation to be performed, e.g. screw motion. However, it can also be a matter of simply positioning the effector E in space, which is not directly derived from a task.

[0058] For each step S.sub.i, S.sub.j, at least one defined impedance pattern and/or admittance pattern is defined, in the present case a compliance pattern, resulting from an impedance or stiffness matrix K.sub.x.

[0059] According to the invention, the impedance patterns and/or admittance patterns should be designed in such a way that they relate to at least one axis of a selected coordinate system, e.g. to at least one axis A.sub.A of the coordinate system C.sub.A of one or more axis members A, to at least one axis A.sub.G of the coordinate system C.sub.G of one or more joints G, to at least one axis A.sub.E of the coordinate system C.sub.E of the effector and/or to at least one axis A.sub.R of the coordinate system of one or more work or task spaces R.

[0060] FIG. 2 shows schematically a flow chart of an example of the execution of the method according to the invention, which can be carried out manually by an operator on a robot system for the programming thereof.

[0061] In a first step 10, the manipulator M is set to a compensated mode. For this purpose, corresponding counterforces and counter-torques are generated by corresponding control of the drive units in the joints G in order to counteract the force of gravity, possibly a centrifugal force and/or a Coriolis force and/or initiated inertial forces, whereby the dead weight and thus the inertia of the manipulator M and any self-locking of the gears or joints is cancelled, so that the manipulator M can in the first place exhibit a repellable behaviour.

[0062] The operator is now able to bring the robotic arm or effector E approximately into the desired pose and/or move it to the desired position.

[0063] If, for example, a Cartesian coordinate system is taken into account as the relevant coordinate system, possible Cartesian task-related stiffness elements, which define the translational and rotational Cartesian stiffness in the decoupled case, result as

k.sub.x,i custom-character .sup.+, i {1 . . . 6}

[0064] If the manipulator M can be moved freely in certain directions, the n.sub.t-th stiffness elements assigned to it are defined as

k.sub.x,i=0

[0065] For this purpose, a specific Cartesian task-related damping d.sub.x,i can be selected for the directions restricted to 6n.sub.t by specifying or selecting a damping profile or compliance pattern.

[0066] It should be mentioned that in practice an attenuation or impedance with respect to the null space should not be specified in detail in order to exclude an interaction with the task or operation under real conditions (e.g. due to sensor noise). Nevertheless, it can also be provided that an independent, e.g. possibly time-variant compliance pattern is assigned to the null space within the framework of the method according to the invention.

[0067] In simple terms, this results in a Cartesian stiffness matrix as

[00001] $K_{x} = (\begin{matrix} K_{x, t} & 0 \\ 0 & K_{x, r} \end{matrix})$

in which

K.sub.x,t, K.sub.x,r custom-character .sup.33

reflect the translational and rotational diagonal, positively defined stiffness matrices.

[0068] In a first step S.sub.i of the above-mentioned approximation, a compliance pattern is then defined with respect to one axis of a Cartesian coordinate system, e.g. the axis A.sub.A of the coordinate system C.sub.A, in a translational alignment. The corresponding stiffness matrix results then as

[00002] $K_{x}^{t - teach} = (\begin{matrix} k_{x, tx} & 0 & 0 & 0 & 0 & 0 \\ 0 & k_{x, ty} & 0 & 0 & 0 & 0 \\ 0 & 0 & k_{x, tz} & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \end{matrix})$

[0069] A single execution of this step S.sub.i could already be sufficient to reach the target pose x.sub.i (step 20 in FIG. 2). However, step S.sub.i could also be repeated one or more times (step 30 in FIG. 2).

[0070] In a further step S.sub.j of the approximation, a defined compliance pattern in a rotational orientation is then defined for this axis A.sub.A of the coordinate system. The corresponding stiffness matrix results as follows

[00003] $K_{x}^{r - teach} = (\begin{matrix} 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & k_{x, rx} & 0 & 0 \\ 0 & 0 & 0 & 0 & k_{x, ry} & 0 \\ 0 & 0 & 0 & 0 & 0 & k_{x, rz} \end{matrix})$

[0071] These steps of the teach-in procedure according to the invention can be repeated as often as necessary, but not necessarily alternately (steps 30, 30 in FIG. 2), until finally pose x.sub.i has been reached (step 40 in FIG. 2).

[0072] The steps, in which the focus is only once to the translational alignment and only once to the rotational alignment, are therefore essentially simplified teach-in steps in themselves.

[0073] The method according to the invention can preferably be used to transfer a once determined and set pose x.sub.i to another target pose x.sub.j, which differs from pose x.sub.i only by a different position, but has a common target orientation or alignment (step 50 in FIG. 2).

[0074] For example, in pose x.sub.i there is a screw which is to be screwed into a component by the effector E of the manipulator M. At further positions of the component there are s-1 screws, which are also to be screwed in.

[0075] After the transition to the e.g. gravitation-compensated mode, a change is made between the translational and rotational states, as mentioned above. The stiffness matrix is user-defined by the selection of the compliance patterns, i.e.

K.sub.x=User(K.sub.x.sup.tteach, K.sub.x.sup.rteach)

If the pose x.sub.i is reached, it is stored as a reference.

[0076] After that, the system switches to a mode in which the manipulator M is guided to the further s-1 positions of the further screws, and there only the position is saved, since the orientation is the same. The corresponding matrix results then as

[00004] $K_{x}^{t - top} = (\begin{matrix} 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & k_{x, rx} & 0 & 0 \\ 0 & 0 & 0 & 0 & k_{x, ry} & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \end{matrix})$

[0077] For each further pose x.sub.j therefore then

K.sub.x=K.sub.x.sup.ttop

[0078] It becomes clear that by using the previous compliance patterns, which are defined by the stiffness matrices, any number of further poses x.sub.i, x.sub.j . . . x.sub.s can be programmed in a systematic way.

[0079] FIG. 3a illustrates in an exemplary manner the advantage of the method according to the invention in comparison to known teach-in methods. This FIG. 3a shows the course of the individual steps in which the effector E of a manipulator M is to assume a pose at a specific target point B in a workspace R.

[0080] It should be emphasized that the target point B, at which an operation by the effector E of the manipulator M is ultimately to be performed, is not known as an input variable or parameter for programming the motion sequence that the manipulator M is to follow through for this purpose.

[0081] A movement of a manipulator M is to be started from an initial state at position A, which can be anywhere in the space, which is completely separated and decoupled from the workspace R, in which position B is located.

[0082] With a pure motion programming (dotted line) the manipulator M with its effector E would end at any point B, which inevitably cannot match with the desired position, because it is either not known or only insufficiently known. Since B is not known beforehand or only insufficiently known, but only implicitly by the result to be achieved (pose, operation at point B), no environmental model can be generated that could be used for pure motion programming.

[0083] In Teach-in programming, in which the manipulator M is only guided in a gravitation-compensated state (dashed line), the effector E always ends in a position B that is too inaccurate and therefore deviates from the desired position B, even if only minimally. However, a minimal deviation is already sufficient to ensure that the desired operations, such as screwing in a screw, cannot be carried out error-free and in a reliably replicable manner. In addition, the teach-in procedure, i.e. ultimately the guiding of the manipulator, proves to be much more difficult to perform in the present case.

[0084] According to the invention, the guidance of the manipulator M is therefore divided into several steps S1 to S4, which can have different durations, and each step is then assigned a stiffness matrix K1 to K4 each defining a compliance pattern. In this way, the effector E can approach position B exactly (solid line) in order to assume the pose x.sub.B necessary for the intended operation.

[0085] When the different stiffness matrices K1 to K4 are used, possibly with simultaneous compensations with respect to gravity, inertia, centrifugal forces and/or Coriolis forces, these can in turn be coordinated with one another via steps S1 to S4, i.e. the resulting individual compliance patterns are all interrelated with one another, as FIG. 3b illustrates.

[0086] It becomes clear that through the targeted selection of the number of steps on the one hand and through the targeted selection of the impedance patterns and/or admittance patterns, in the simplest case of compliance patterns, on the other hand a step-by-step realization of the desired pose(s) becomes possible, which is not included into the programming of the motion sequence in advance due to lack of knowledge.

METHOD AND DEVICE FOR DEFINING A MOVEMENT SEQUENCE FOR A ROBOT

Assignee

Inventors

Cpc classification

Classification Explorer

G05B2219/40387

PHYSICS

Classification Explorer

G05B2219/35478

PHYSICS

Classification Explorer

B25J9/1664

PERFORMING OPERATIONS; TRANSPORTING

Classification Explorer

G05B19/423

PHYSICS

Classification Explorer

G05B2219/36432

PHYSICS

Classification Explorer

G05B2219/39001

PHYSICS

Classification Explorer

B25J9/163

PERFORMING OPERATIONS; TRANSPORTING

International classification

Classification Explorer

B25J9/16

PERFORMING OPERATIONS; TRANSPORTING

Classification Explorer

G05B19/423

PHYSICS

Abstract

Claims

Description