Method for operating an industrial robot, and control device

Abstract

The invention relates to a method for operating an industrial robot, wherein a plurality of movements are carried out by the industrial robot during a work cycle to be completed within a specified target cycle time, the speed of the movements being described by a speed parameter. According to the invention, the following steps are carried out in order to approximate the target cycle time and / or reduce the load on the industrial robot: a. selecting a plurality of movements in order to form a movement corpus and ascertaining movement data belonging to the movements in the movement corpus, the movement data comprising a load resulting from the movement, b. temporarily storing the movement data, including a unique assignment to the corresponding movement, c. evaluating the temporarily stored movement data and selecting at least one movement to be controlled by means of a control intervention, d. carrying out a control attempt involving at least one control intervention, a new speed parameter for a movement to be controlled being determined as part of a control intervention, whereby the target cycle time is approximated and / or the load on the industrial robot is reduced, and e. storing the new speed parameter generated during the control intervention as the new speed parameter for the movement to be controlled. The invention also relates to a control device.

Description

[0001] R.414449

[0002] - 1 -

[0003] Description

[0004] Title:

[0005] Method for operating an industrial robot, control unit

[0006] The present invention relates to a method for operating an industrial robot with the features of the preamble of claim 1. Advantageous embodiments of the invention are described in the dependent claims. The invention also relates to a control unit.

[0007] State of the art

[0008] Programming industrial robots is a central component of modern production processes and plays a crucial role in automation. The primary goal is to control the robot so that it moves precisely, quickly, and safely to perform complex tasks. Determining the optimal movement speeds is crucial. While excessively high speeds can reduce cycle time, they often lead to increased wear and tear. Conversely, excessively low speeds cause unnecessary delays and negatively impact cycle time, thus reducing the efficiency of the production line. Cycle time describes the time the robot requires for a complete work cycle. It is essential for synchronization with other machines in the production line and significantly determines production capacity. Therefore, adhering to the cycle time is of paramount importance from an economic perspective.Therefore, it is important to program the robot's movements so that they are executed within the required cycle time without compromising quality.

[0009] As a rule, the software developer of the machine manufacturer defines a selection of fixed standard functions in the application program of an industrial robot. R.414449

[0010] - 2 - speeds, for example, as global variables. The selection often includes three to five different standard speeds. These standard speeds are available to him when he programs the industrial robot for its intended tasks. In the program code, he assigns a selected standard speed to individual movement commands. These are chosen so that the cycle time is as close as possible to a predefined value, the so-called target cycle time.

[0011] Due to the technically standard program structure, one and the same movement command triggers different robot movements in practice, for example, due to different starting positions, changing payloads, etc. This results in significantly more different movements occurring in space than the number of actually programmed movement commands. Consequently, in the operation of an industrial robot, a large number of different movements are executed at a small number of (fixed) speeds. These speeds, in conjunction with other parameters (e.g., payload, leverage ratios due to different joint positions, etc.), lead to varying stresses on the robot's mechanics. As a result, over a movement cycle, some movement commands cause higher robot stresses than others.

[0012] The invention therefore addresses the problem of adjusting the target cycle time of an industrial robot, taking the load into account. To solve this problem, a method with the features of claim 1 is proposed. Preferred embodiments are described in the dependent claims. A control unit is also proposed.

[0013] Disclosure of the invention

[0014] A method for operating an industrial robot is proposed, wherein the industrial robot performs a multitude of movements during a work cycle to be completed within a predetermined target cycle time, the speed of which is described by a speed parameter. According to the invention, the following steps are carried out to approximate the target cycle time and / or reduce the load on the industrial robot: R.414449

[0015] - 3 - a. Selecting several movements for a motion set and determining motion data belonging to the movements in the motion set, wherein the motion data includes a load resulting from the movement, b. Temporarily storing the motion data including a unique assignment to the associated movement, c. Evaluating the temporarily stored motion data and selecting at least one movement to be controlled by control intervention, d. Performing a control attempt comprising at least one control intervention, wherein, within the scope of a control intervention, a new speed parameter for a movement to be controlled is determined, thereby approximating the target cycle time and / or reducing the load on the industrial robot, and e. Storing the new speed parameter generated in the control intervention as the new speed parameter for the movement to be controlled.

[0016] The proposed invention enables the required target cycle time to be achieved with the lowest possible load. This is made possible by the automatic optimization of speed parameters, which can also be performed regularly during operation. This allows the industrial robot to be flexibly adjusted to a changed target cycle time. The load reduction significantly improves the service life of the industrial robot. Furthermore, adjusting the target cycle time leads to optimized utilization and better availability of the industrial robot's technical capabilities. Preferably, the control system operates continuously, reacting to its previous control result—that is, a new speed parameter determined and set in a previous control test. This also significantly reduces the effort required to achieve the target cycle time, especially compared to the prior art.

[0017] A work cycle is understood to be a defined work process for which a target cycle time is specified, whereby the work process is defined not only by the movements to be performed but also by the desired work result (for example: "Place object X in shelf Y") R.414449

[0018] - 4 - can. A movement refers to a trajectory in space, typically performed by the working tool of the industrial robot, including start and end points. The latter means that two spatially identical trajectories with different starting points in space must be considered two different movements. This differentiation is important because two movements with the same form but spatially offset can lead to different loads if, for example, one is performed very close to the robot base and the other is performed at a greater distance, over which the industrial robot must extend, resulting in a load with a longer lever arm.

[0019] The robot typically controls which axes are moved and how they are moved to perform a movement. The speed of a movement is therefore related to the time it takes to complete the movement and not necessarily to the angular velocities of individual robot axes.

[0020] Process steps a. and b., which can be executed not only sequentially but also in parallel, are preferably performed during the ongoing work cycle. Motion data is collected, which must include, in particular, the resulting load of a movement. Load refers to the load values ​​acting on the individual axes of the robot. These load values ​​can be physical quantities such as torques, forces, required drive power, and the like. They are expressed, for example, as a percentage of the design or nominal value. Continuing the example, the highest value occurring during the movement could then correspond to the resulting load on the robot. Other, statistically more complex methods for determining the resulting load are also conceivable, such as using an average of the various axis loads.

[0021] An exemplary determination of a resulting load is illustrated by a case study: an industrial robot has three axes. During a movement, at time X, a load vector of (20% 55% 28%) is applied. T measured. It means that the three axes are each at 20%, 55% and 28% respectively. R.414449

[0022] - 5 -

[0023] % of their respective design limits are loaded. The resulting load on the industrial robot corresponds to the maximum value, i.e., 55%.

[0024] After a collection of individual movements from a work cycle (basic movement set) and their resulting loads, as well as preferably other related data, has been temporarily stored in step b, these are first evaluated in step c. This evaluation preferably determines the direction in which a control attempt should be made. For example, if the cycle time of the work cycle lags significantly behind the target cycle time, an initial interest in accelerating the movements might outweigh a load reduction. Conversely, a control attempt could be primarily aimed at reducing the load if the cycle time of a work cycle deviates only slightly from the target cycle time. Furthermore, the evaluation preferably includes an assessment of which movements should be subjected to speed adjustment.

[0025] In the subsequent control attempt, a new velocity parameter is determined during a control intervention and assigned to a movement for future executions. Consequently, this movement will henceforth be executed at the velocity regulated by the new velocity parameter. If this movement is later added to a set of motions again, it is evaluated based on this new velocity parameter. If it is subsequently used for a control intervention, this new velocity parameter is in turn superseded by a new one.

[0026] The following control scenarios are possible, among others: For an industrial robot that completes its work cycle within the target cycle time, movements with high stress are slowed down while movements with low stress are accelerated. This reduces the overall stress on the industrial robot while maintaining the same cycle time. For a robot that operates faster than the required target cycle time, particularly stressful movements are slowed down. For a robot that does not reach the required target cycle time, those movements that previously resulted in low stress are accelerated. Simultaneously, movements with high stress can be slowed down if necessary.

[0027] - 6 - can be slowed down as required. Should the required cycle time change in the long term, for example due to slower part supply by another machine part, the robot can react by readjusting the speed parameters of essential movements.

[0028] In a further development of the invention, it is proposed that steps a. and b. be repeated until a sufficient number of motion data points have been temporarily stored. This preferred embodiment ensures that not only is sufficient data available for conducting a control test, but that this data is also sufficiently reliable. It is typical for industrial robots that the actual cycle time of a recurring work cycle is never exactly uniform. Depending on various factors, a cycle time can vary, particularly in the tenths of a second range. The preferred embodiment ensures that a subsequent evaluation of the baseline motion data can yield more accurate results. The precise number of data points that constitutes a sufficient number must be assessed on a case-by-case basis.

[0029] It is also proposed that steps a. to e. be repeated at predetermined control intervals, whereby the target cycle time is approximated by iteratively determining new speed parameters. Possible control intervals include a time interval, a fixed number of movements, a fixed number of executed work cycles, a fixed number of movements added to the baseline movement set, and others. A single control attempt will realistically not achieve a target cycle time with sufficient certainty. This preferred embodiment ensures that cycle time deviations are reliably compensated for by iterative control attempts.

[0030] It is also proposed that in step a. the load resulting from the movement be determined based on torques and / or forces acting on the axes of the industrial robot. This preferred embodiment is advantageous because these quantities, due to their importance for robot design, are typically measurable on the individual axes anyway. Furthermore, R.414449

[0031] - 7 - these are essential design parameters, compliance with or exceedance of which is directly causally related to wear and tear and service life of an industrial robot.

[0032] Furthermore, it is proposed that in step a. the motion data be supplemented with a following distance and / or an axis velocity of at least one axis. These two parameters are also directly related to the load on the industrial robot. Including them allows for a more precise quantification of the robot load. Following distance is defined as the difference between the current target position and the actual position of a controlled axis.

[0033] It is further proposed that in step a. or b., the motion data be supplemented with the execution duration of the motion, a preceding motion, a subsequent motion, and / or a timestamp of the occurrence of the highest load resulting from the motion. This data provides important information about the influence of the motion on the cycle time, as well as on transition loads between two motions. For example, accelerating a motion during the transition from a previous or to a subsequent, significantly slower motion can introduce new loads simply due to the large change in velocity required during the transition. Knowledge of the preceding and subsequent motions is therefore advantageous. The timestamp of the highest load also provides important information about when and, if applicable, why the highest load occurs.In extreme cases, building on the example above, the highest stress can occur during the transition to another movement. In this case, it may be useful to consider two movements together as a single movement. The execution time of this fictitious single movement then corresponds to the sum of the two individual execution times.

[0034] Furthermore, it is proposed that in step a., motion data measured as time series be expressed in the form of statistical individual values, preferably maximum values. With this preferred embodiment, tensorial results are reduced to individual values ​​and made accessible to simple one-dimensional statistical calculations. This allows results from unordered R.414449

[0035] - 8 -

[0036] Tensor spaces are mapped to well-ordered spaces, thus making them accessible to comparative analysis.

[0037] Furthermore, it is proposed that in step b., in order to assign the motion data to the motion, the latter be provided with an identifier. Such an identifier makes it possible to identify a single motion independently of the motion command that triggered it. Preferably, the identifier is designed as a two-part key, the first part of which contains a reference to a task and a subroutine to be executed for it. The second part preferably contains a sequential number of the motion command called in the subroutine.

[0038] Furthermore, it is proposed that in step c. a predetermined number of movements to be controlled are selected from the pool of movements, with the selection based on their load and preferably an equal number of movements with high and low loads being selected. The most and least load-bearing movements are those whose speed can or should be varied the most. By preferentially selecting equal numbers of movements with high and low loads, it is ensured that the effect on the cycle time for each acceleration of one movement tracked by a control intervention can be compensated by deceleration of another.

[0039] It is also proposed that in step d. the new speed parameter is determined from a product of the previous speed parameter and a control factor calculated in the current control attempt. This preferred embodiment is advantageous because a reference to the movement speed originally programmed by the machine manufacturer is always maintained. Thus, the movement speed is always expressed as a multiple of the original movement speed. Furthermore, the movement speed can be limited relatively to a multiple of the original speed by restricting the control factor; such a restriction could, for example, be set at a maximum of four times the original speed. Alternatively, the speed can also be limited absolutely. In combination R.414449

[0040] - 9 - could be a limit of a maximum of four times the original speed, but in no case faster than 10 m / s.

[0041] In a further development of the invention, it is proposed that in step c. the evaluation of the temporarily stored motion data comprises at least one of the following substeps: a comparison of an average cycle time with the target cycle time and / or a comparison of an actual load with a maximum possible load and / or

[0042] Defining a control objective to be pursued in the rule experiment.

[0043] For a successful control attempt, clarity regarding the existence of a control deviation should be obtained during the evaluation. In particular, in a method according to the invention, two possible control variables, cycle time and load, are available, which are initially opposed to each other. This preferred embodiment clarifies whether a problem exists, what form it takes, and consequently, what the goal of the control should be. The mean cycle time is defined as the arithmetic mean of the cycle times of several, preferably identical, work cycles.

[0044] Building on this, it is proposed that, if a predefined minimum deviation of the mean cycle time from the target cycle time occurs, approximating the target cycle time should be chosen as the uniform control objective. Since adhering to the target cycle time is typically of paramount and immediate interest in robotics, even from an economic perspective, it seems sensible to define a limit value for the cycle time deviation, beyond which the focus on cycle time control renders consideration of the workload unnecessary. This is achieved with this preferred embodiment. In this case, the control interventions in the control test are limited exclusively to accelerating (if the target cycle time is undershot) or decelerating (if it is exceeded) individual or all movements of the work cycle.Preferably, for accelerating control interventions, movements that exert a low load on the industrial robot are selected, whereas for decelerating control interventions, movements with a high resulting load are selected. Furthermore, the minimum deviation should be chosen as specified in R.414449.

[0045] - 10 - that a deviation in cycle time below this threshold can be considered at least acceptable in operation. In the case of such a deviation, the interests in cycle time optimization and load reduction are equally important.

[0046] Therefore, it is proposed that, if the minimum deviation is not met, a reduction in load, preferably while simultaneously approaching the target cycle time, be chosen as a mixed control objective. This preferred embodiment is advantageous because, with a mixed control objective, the remaining control deviation between cycle time and target cycle time is already sufficiently small to be considered acceptable. Thus, control interventions in the cycle time may only result in relatively small changes. These are to be expected anyway when controlling the load, so that, with appropriately chosen control interventions, cycle time optimization can be achieved as a byproduct of load optimization.

[0047] It is proposed that the following sub-steps be carried out in the control experiment with a mixed control objective:

[0048] Determining the relevance for each movement to be controlled by multiplying an occurrence frequency by an execution duration,

[0049] Defining a reference relevance based on a statistical function of the specified relevance, where preferably a minimum value or mean value is chosen as the statistical function,

[0050] Determining a preliminary velocity change for all movements to be controlled, depending on the control objective.

[0051] Determining a relevance ratio as the quotient of one's own relevance to the reference relevance for all movements to be controlled, and determining a new velocity change by scaling the preliminary velocity change with the relevance ratio for all movements to be controlled.

[0052] Movements within a work cycle occur with varying frequencies; their durations thus contribute differently to the cycle time. With this preferred embodiment, changes in speed are relatively R.414449

[0053] - 11 - are scaled to each other so that they have the same magnitude of influence on the cycle time. This advantage will be illustrated by an example:

[0054] A work cycle consisting of two movements A and B lasts 8 seconds and only slightly misses a target cycle time of 8.1 seconds. Movement A lasts 2 seconds and occurs three times, while movement B lasts 1 second and occurs twice. The load on B is to be reduced, and therefore B must be slowed down. To optimize only the load initially, the control test should be performed in such a way that a change in the movement speeds does not result in a significant change in the cycle time. According to the proposed preferred embodiment, the relevance RA and RB for A and B are determined from the product of the frequency of occurrence and the duration of execution:

[0055] R A = 2 sx 3 = 6 s

[0056] R B = 1 sx 2 = 2 s

[0057] Subsequently, a reference relevance is defined, in this example based on the minimum value. The minimum value corresponds to the relevance of B: RB = 2 s. In the present example, B is to be adjusted by a preliminary change in the velocity of A. v The speed must be slowed down by 10%. Therefore, the original speed parameter must be multiplied by the control factor 1 - 0.1 = 0.9.

[0058] Would this provisional speed change A vApplying a corresponding acceleration of 10% to A to compensate for the deceleration in B (multiplying the original speed parameter by the control factor 1 + 0.1 = 1.1) would, however, unintentionally change the cycle time because A, due to its different execution duration and frequency, has a different proportion of the cycle time than B. Therefore, according to the proposed preferred embodiment, the relevance ratio Q of A and B is determined. Due to the choice of the reference relevance, QB = 1. The relevance ratio of A is then given by R.414449

[0059] - 12 -

[0060] The preliminary speed change A v For an application to A, it is then scaled by the relevance ratio of A: 0.033

[0061] Therefore, a 10% deceleration of B must be offset by a 0.033 (3.3%) acceleration of A to maintain the cycle time. The control factor for A is accordingly 1 + 0.033 = 1.033. If, at the same time, the cycle time is to be slightly adjusted closer to the target cycle time, these control factors must be adjusted individually and to a small extent after calculation, without making any compensating changes to other control factors.

[0062] Furthermore, it is proposed that steps d. and e. are only performed if at least one rule condition is met. This preferred embodiment reduces the computational effort by preventing any control attempt from being made unless the user-defined rule condition is met. At the same time, a baseline set of motions and associated motion data are still collected, so that a control attempt can be made immediately upon the occurrence of the rule condition. The rule condition could be, for example: the expiration of a fixed time interval, a change in the target cycle time, the presence of a sufficient number of motion data points and / or movements in the baseline set, a manual change in the robot speed by the operator, manufacturer-specific special cases, or similar.

[0063] Furthermore, a control unit for an industrial robot is proposed, which is configured to carry out steps of a method according to the invention. Such a control unit has the aforementioned features and advantages.

[0064] The invention is explained in more detail below with reference to a figure. This figure shows a schematic representation of a process sequence according to the invention.

[0065] Character description

[0066] The figure shows a schematic representation of a process sequence according to the invention. For the execution of a work cycle of an industrial robot, an R.414449 is required.

[0067] - 13 -

[0068] A target cycle time is specified. The industrial robot performs a multitude of movements during the work cycle. The execution time t and the resulting load tp of these movements are determined, the latter preferably by measurement. In a first decision step E1, it is decided whether, based on the measurement data, the movement should be added to a pool of potentially controllable movements. If the decision is positive, in a first process step S1, an identifier referencing the movement is entered into the pool of movements. This entry is assigned the load resulting from the movement as well as other movement data such as execution time, following error, a preceding movement, a subsequent movement, and the like. Decision step E1 and process step S1 are repeated until the pool of movements contains a sufficient number of movements.In a further process step S2, the baseline of movements is then evaluated. The average cycle time is compared with the target cycle time, and the actual load of each movement is compared with the permissible maximum load. A control objective is then defined. This control objective is preferably a uniform or a mixed control objective. Subsequently, the movements in the baseline are evaluated to determine which should be subjected to control intervention in order to achieve the control objective. In a subsequent decision step E2, it is checked whether a condition for triggering a control attempt exists. Such a condition could be, for example, a sufficient number of movements in the baseline or a change in the target cycle time compared to the last control. If the condition is met, a control attempt is carried out in the next process step S3.New speed parameters are determined for the previously selected movements based on their load and their influence on the cycle time. In a subsequent process step S4, the new speed parameters are assigned to the respective movement for future use. If necessary, the decision and process steps are repeated until both the cycle time corresponds sufficiently closely to the target cycle time and the movement loads have been adequately optimized.

Claims

R.414449 - 14 - Claims 1. A method for operating an industrial robot, wherein the industrial robot performs a plurality of movements during a work cycle to be completed within a predetermined target cycle time, the speed of which is described by a speed parameter, characterized in that the following steps are carried out to approximate the target cycle time and / or reduce the load on the industrial robot: a. Selecting several movements for a set of movements and determining motion data belonging to the movements in the set of movements, wherein the motion data includes a load resulting from the movement, b. Temporarily storing the motion data including a unique assignment to the associated movement, c. Evaluating the temporarily stored motion data and selecting at least one movement to be controlled by a control intervention, d.Performing a control attempt comprising at least one control intervention, wherein, within the framework of a control intervention, a new speed parameter for a movement to be controlled is determined, thereby approximating the target cycle time and / or reducing the load on the industrial robot, and e. Saving the new speed parameter generated in the control intervention as a new speed parameter for the movement to be controlled.

2. Method according to one of the preceding claims, characterized in that steps a. and b. are repeated until a sufficient number of motion data has been temporarily stored. R.414449 - 15 - 3. Method according to claim 1, characterized in that steps a. to e. are repeated in predetermined control intervals, whereby the target cycle time is approximated by iterative determination of new speed parameters.

4. Method according to one of the preceding claims, characterized in that in step a. the load resulting from the movement is determined on the basis of torques and / or forces acting on axes of the industrial robot.

5. Method according to one of the preceding claims, characterized in that in step a. the motion data are additionally supplemented by a following distance and / or an axle speed of at least one axle.

6. Method according to one of the preceding claims, characterized in that in step a. or b. the motion data are additionally supplemented by an execution duration of the motion, a preceding motion, a subsequent motion and / or a timestamp of an occurrence of the highest load resulting from the motion.

7. Method according to one of the preceding ones, characterized in that in step a., movement data measured as time series are expressed in the form of statistical individual values, preferably maximum values.

8. Method according to one of the preceding claims, characterized in that in step b. the movement is provided with an identifier for the purpose of assigning the movement data to the movement.

9. Method according to one of the preceding claims, characterized in that in step c. a predetermined number of movements to be controlled is selected from the set of movements, wherein the selection is based on their load and preferably movements with high load and movements with low load are selected in equal numbers. R.414449 - 16 - 10. Method according to one of the preceding claims, characterized in that in step d. the new velocity parameter is determined from a product of previous velocity parameter and a control factor calculated in the current control attempt.

11. Method according to one of the preceding claims, characterized in that in step c. the evaluation of the temporarily stored motion data comprises at least one of the following substeps: a comparison of an average cycle time with the target cycle time and / or a comparison of an actual load with a maximum possible load and / or Defining a control objective to be pursued in the rule experiment.

12. Method according to claim 11, characterized in that, in the case of a predetermined minimum deviation of the mean cycle time from the target cycle time, an approximation of the target cycle time is chosen as a uniform control target.

13. Method according to claim 12, characterized in that, if the minimum deviation is not exceeded, a reduction of the load, preferably with simultaneous approximation of the target cycle time, is selected as a mixed control objective.

14. Method according to claim 13, characterized in that, in the control attempt with a mixed control objective, accelerating control interventions are combined with decelerating control interventions.

15. Method according to claim 13 or 14, characterized in that the following partial steps are carried out in the control test with mixed control objective: Determining a relevance for each movement to be controlled by multiplying an occurrence frequency by an execution duration, defining a reference relevance based on a statistical function of the determined relevance, wherein a minimum value or mean value is preferably chosen as the statistical function, Determining a preliminary velocity change for all movements to be controlled, depending on the control objective. R.414449 - 17 - Determining a relevance ratio as the quotient of one's own relevance to the reference relevance for all movements to be controlled, and determining a new velocity change by scaling the preliminary velocity change with the relevance ratio for all movements to be controlled.

16. Method according to one of the preceding claims, characterized in that steps d. and e. are only carried out if at least one rule condition is met.

17. Control unit for an industrial robot, which is configured to perform steps of a method according to claims 1 to 16.

Images