Adjusting grasp simulation through parameter identification in real environment

Abstract

A method for automatically optimizing parameters of a robot-assisted gripping process, comprising: determining gripping actual data which describe success of at least one gripping at a gripping position on an object gripped by a gripping robot; and / or determining holding actual data, wherein the holding actual data describes at least one parameter related to the gripping process, in particular the gripping, in particular a force on the object gripped by the gripping robot; simulating, in particular replicating, at least one gripping process, in particular gripping, based on the determined gripping actual data and / or the determined holding actual data in a simulation environment; the parameter is optimized on the basis of the data, in particular on the basis of the determined captured actual data and / or on the basis of the determined holding actual data, in order to determine the optimized parameter.

Description

Technical Field

[0001] The present invention relates to a method for automatically optimizing parameters, in particular for a robot-assisted grasping process, a method for controlling a grasping robot, a system for automatically optimizing parameters, and a computer program or computer program product. Background Art

[0002] In the prior art, technical systems are partly trained in a simulation environment. For this purpose, any number of simulation data are usually generated. However, these simulation data are only qualitative responses to reality. The difference between simulation and reality is usually also referred to as the simulation-to-reality gap (English: "Sim2Real-Gap"). Summary of the Invention

[0003] The object of the present invention is in particular to narrow the simulation-to-reality gap, in particular to further optimize the parameterization of the simulation.

[0004] The object of the present invention is achieved according to the teachings of the independent claims. Various embodiments and extensions of the present invention are given by the dependent claims.

[0005] In one embodiment of the present invention, a method for automatically optimizing parameters is provided, in particular for a robot-assisted grasping process. In one embodiment, the method includes determining grasping actual data (Greif-Realdaten), where the grasping actual data describes at least one grasping success (Greiferfolg) at the grasping position, in particular on an object to be grasped by the grasping robot or an object that has been grasped by the grasping robot. Alternatively or additionally, the method includes determining holding actual data (Halte-Realdaten), where the holding actual data describes at least one force, in particular the force on an object grasped by the grasping robot and / or the force on the gripper of the grasping robot, where the gripper has in particular (successfully) grasped the object to be grasped. In one embodiment, the holding actual data describes at least one parameter or characteristic important for the grasping status (Bestand) on the grasped object, in particular a parameter or characteristic related to grasping. In one embodiment, the important parameter or characteristic is a force, in particular a normal force, a frictional force, especially friction or a coefficient of friction, for example especially (for) static friction, rolling-sliding friction Rolling friction (English: “rolling friction”), spinning friction (English: “spinning friction”) and / or lateral friction (English: “lateral friction”), which acts on the grasped object and / or at least one gripper finger, in particular the closing force of at least one gripper finger, and / or the mass, inertia and / or orientation (Lage) of the center of mass of the grasped object and / or at least one gripper finger. In one embodiment, the method further comprises, in particular in a certain step, performing a simulation, in particular a replication, more particularly in a simulation environment, wherein at least one grasping process, in particular a grasp, is simulated in the simulation environment based on the actual grasping data determined according to the grasping process, in particular the grasp, and / or the actual holding data determined. In one embodiment, the method further comprises, in particular in a certain step: performing a data-based optimization of the parameters of the simulated grasping process, in particular the simulated grasp. Alternatively or additionally, the method comprises: optimizing the parameters of the simulated grasping process, in particular the simulated grasp, based on the determined actual grasping data and / or based on the determined actual holding data.

[0006] In one embodiment, optimized parameters can thus be advantageously determined, in particular an improved simulation environment can be advantageously achieved, more particularly a more realistic or near-realistic / closer-to-real simulation or simulation environment, in particular compared to a simulation or simulation environment that is not optimized or has been optimized not based on the actual grasping data and / or the actual holding data.

[0007] The term “actual grasping data” used herein should be particularly understood as information, more particularly as digital information, which describes the data of the actual grasping process, in particular the actually performed grasp. In one embodiment, the term should be particularly understood as the counterpart of “simulated grasping data”, where “simulated grasping data” refers to information that describes the data of the simulated grasping process, in particular the simulated grasp, in a simulation environment.

[0008] The term “actual holding data” used herein should be particularly understood as information, more particularly as digital information, which describes the data held during or in the actual grasping process, in particular during or when the actual grasp is performed. In one embodiment, the term should be particularly understood as the counterpart of “simulated holding data”.

[0009] In one embodiment, “actual data” and “simulated data” respectively describe the same facts or information for reality and the simulation or simulation environment.

[0010] In one embodiment, the data-based optimization includes an automatic adjustment of parameters, in particular an automatic adjustment of the simulation quality of a simulation, an automatic adjustment of (simulated) dynamic parameters and / or simulated forces, in particular frictional forces or friction. In one embodiment, the automatic adjustment of parameters is based on at least one evaluation of a cost function (Kostenfunktion), which serves as the basis for the optimization or is optimized based on this cost function.

[0011] In one embodiment, the data-based optimization, in particular the optimization of parameters, is based on a cost function of grasping actual data and grasping simulation data and / or holding actual data and holding simulation data. In one embodiment, the grasping simulation data describes at least one simulated grasping parameter, in particular the grasping success at the grasping position in the simulation. In one embodiment, the holding simulation data describes at least one simulated holding parameter, in particular the force of the simulated grasp. In one embodiment, the cost function can be based on a comparison of consistent entries in the form of a success vector for grasping in reality and in simulation (übereinstimmender )), in particular the cost function c = Summe(Betrag(Greiferfolg_real–Greiferfolg_sim)), where Greiferfolg_real corresponds to the success vector for grasping in reality, in particular during an actually performed grasp, and Greiferfolg_sim corresponds to the success vector for grasping during the corresponding grasp in the simulation environment. In one embodiment, the entries in the vector can be 1 or 0, corresponding to a successful or unsuccessful grasp. In one embodiment, the data-based optimization includes an adjustment of parameters, in particular abstract parameters, such as in particular the time steps (Zeitschritte) of the simulation, solver iterations (English: “solver-iterations”), etc., and more particularly the “lateral friction”, “rolling friction”, “spinning friction” of the grasped object and / or the “lateral friction”, “rolling friction”, “spinning friction” of the gripper, in particular of at least one gripper finger, the mass, inertia and / or center of mass of the grasped object, etc.

[0012] Thereby, in one embodiment, it is advantageously possible to allow an estimation based on the evaluation of the cost function Decisions regarding parameters, especially dynamic parameters. Thus, in one embodiment, especially when the grasping is close to the edge of the object and / or at the fingertips of the gripper fingers and the object cannot be held or falls out of the gripper or slips out of the fingertips, the weight of the object can be increased in the simulation or the friction parameters can be adjusted, especially preferably automatically and / or data - based.

[0013] In one embodiment, optimization, especially data - based optimization, is performed with a gradient - free optimizer. In one embodiment, the optimization algorithm is based on the Monte - Carlo algorithm, or the optimization algorithm is the Monte - Carlo algorithm, especially the Covariance Matrix Adaptation Evolution Strategy (English: "Covariance Matrix Adaptation Evolution Strategy"; CMA - ES), etc.

[0014] Thereby, in one embodiment, it is possible to allow for (more) rapid optimization of the parameters. In one embodiment, the optimization can be advantageously performed automatically based on this.

[0015] In one embodiment, determining the actual grasping data includes: determining a plurality of grasping positions on the object to be grasped (actual), especially collision - free and / or graspable grasping positions. In one embodiment, determining the actual grasping data further includes: determining a plurality of grasping positions, especially collision - free and / or graspable grasping positions. In one embodiment, determining the actual grasping data further includes: selecting one of the determined grasping positions. In one embodiment, determining the actual grasping data further includes: grasping the object based on the selected grasping position, especially using a grasping robot. In one embodiment, determining the actual grasping data further includes: storing the grasping success of the selected grasping position, especially for the selected grasping position on the object to be grasped and the pose of the object to be grasped, especially on the working surface of the grasping robot.

[0016] In one embodiment, determining the actual grasping data includes a pre - step: randomly placing the object to be grasped, especially using a grasping robot.

[0017] In one embodiment, determining the actual grasping data further includes: repeating the above - mentioned method steps for determining the actual grasping data.

[0018] In one embodiment, especially after determining the actual grasping data, the object to be grasped is moved by the robot to a random position, especially to a random position in the working area of the grasping robot, and then is put down.

[0019] Thus, in some embodiments, it is possible to allow the object to be grasped to land in a random pose, especially on the working surface of the robot. Accordingly, in some embodiments, a grasping position corresponding to the random pose of the object to be grasped is determined.

[0020] Thus, in some embodiments, it is also possible to advantageously obtain a pre-sorted set (vorsortierte Menge) of object grasps, especially a set (Menge) pre-sorted according to grasping success.

[0021] In one embodiment of the present invention, a method for optimizing a grasping sampler is provided. In one embodiment, the method includes steps that are at least substantially the same as those for determining grasping actual data, especially including: based on the selection frequency of the grasping position Select a determined grasping position as described above. Preferably, in one embodiment, the grasping sampler can be optimized by determining the grasping actual data, especially to more quickly determine grasps with high or higher grasping success, especially determined by the grasping sampler, and more particularly based on the determined, especially stored, grasping actual data.

[0022] In one embodiment, especially when repeating the method steps for determining the grasping actual data described herein, selecting a determined grasping position is based on the selection frequency of the determined grasping position, especially selecting the determined grasping position with the lowest selection frequency from the determined grasping positions.

[0023] Thereby, it is possible to advantageously obtain, especially more quickly obtain, a pre-sorted set of grasping positions of the object to be grasped.

[0024] In one embodiment, determining the holding actual data includes determining a holding - grasping position, particularly on the grasped object, wherein the holding - grasping position is arranged on the object to be grasped and / or the grasped object such that a force can be applied through the holding - grasping position, which force can counteract or counteracts at least one force on the object to be grasped and / or the grasped object, particularly in a direction opposite to the grasping direction of the grasping robot. In one embodiment, determining the holding actual data further includes: grasping, particularly holding the object, at the determined holding - grasping position, particularly by a holding robot different from the grasping robot. In one embodiment, this can include compensating for the gravitational force acting on the grasped object, particularly when the object is a relatively large object compared to the gripper or the grasping robot, and / or particularly when the object is a relatively heavy object compared to the maximum force of the grasping robot. In one embodiment, determining the holding actual data further includes: applying a force, particularly a force acting in a direction opposite to the force on the grasped object, more particularly a force acting in a direction opposite to the force applied on the object by the grasping robot (not by the holding robot), particularly in a direction opposite to the grasping direction of the grasping robot, particularly through the holding - grasping position and more particularly by means of the holding robot.

[0025] Thereby, in one embodiment, it is advantageously possible to allow (more) accurate, particularly (more) rapid determination or calculation of parameter values for simulation. Thereby, in one embodiment, compared to a situation where, in particular, no holding robot is used or only a simulated environment is relied on, it is advantageously possible to allow better determination of, in particular, rotational forces, and more particularly rotational friction values.

[0026] In one embodiment, applying a force by means of the holding robot, particularly, includes: gradually increasing the force in predetermined steps, particularly until the grasping robot loses the grasped object and / or until a particularly predetermined maximum force is about to be exceeded or is exceeded. In one embodiment, applying the force includes (numerically) measuring the applied force.

[0027] Thereby, in one embodiment, it is advantageously possible to allow (more) accurate, particularly step - by - step determination of parameters for simulation. In one embodiment, it is also advantageously possible to allow (more) accurate determination of at what force the grasp will be lost.

[0028] In one embodiment, determining the grasping actual data and / or the holding actual data includes: photographing the object by means of a photographing device. In one embodiment, photographing the object includes determining the pose of the object, particularly over time, and more particularly includes localizing the grasped object (English: “In - Hand - Localization”), particularly in the gripper, and / or tracking the object (English: “tracking”), particularly as long as the object is held.

[0029] In one embodiment, particularly by using a capturing device to capture an object, and more particularly by tracking, the cost function can be modified. In one embodiment, it can be modified to c = Summe(Betrag(traj-sim–traj_real)), where traj_sim is the trajectory of the object captured in the simulated environment, particularly describing the trajectory, and where traj_real is the trajectory of the object captured in reality, particularly describing the trajectory; or the cost function is applied to optimization. In one embodiment, particularly by using the capturing device to capture the object, at least one starting position of the object, particularly the starting frame, and the ending position of the object, particularly the ending frame, can be used for the cost function.

[0030] Thereby, in one embodiment, it is advantageously possible to allow for improved parameter estimation for optimization, particularly since the trajectory of the object contains more information, especially compared to the grasping success variable that indicates or describes the success or failure of grasping.

[0031] Alternatively or additionally, in one embodiment, the movement of the object can be approximated by maintaining the ending position of the robot, and particularly incorporated into the cost function.

[0032] Thereby, in one embodiment, information about the grasping process can be advantageously obtained that goes beyond the information content of grasping success, and particularly can provide or provide an improved cost function.

[0033] In one embodiment of the present invention, a method for controlling a grasping robot is provided. In one embodiment, the method includes determining control data based on optimization according to any of the above embodiments, particularly for robot-assisted grasping, and more particularly for a robot-assisted grasping process. In one embodiment, the method includes controlling and / or moving the grasping robot based on the optimized control data.

[0034] Thereby, in one embodiment, it is possible to allow: particularly for better and / or (more) rapid grasping of the object to be grasped, the movement and / or control of the robot can be first optimized in the simulated environment and then transferred to the robot. Thereby, in one embodiment, it is also allowed that the grasping force, particularly the closing force of the gripper fingers, can be (more) optimally utilized or (more) optimally utilized.

[0035] In one embodiment, as long as it is technically feasible and / or applicable, the embodiments of the above method can be applied to or transferred to applications with multiple objects, especially applications where multiple objects to be grasped are in a container. In particular, in one embodiment, the parameters determined for free objects can be transferred to, especially applied to, the grasping process where multiple objects are in a container. More particularly, the control data of the grasping robot is determined according to the optimized parameters, and the grasping robot is to grasp at least one object from multiple objects, especially objects arranged in a container.

[0036] In one embodiment of the present invention, a system for operating and / or monitoring at least one robot is provided. In one embodiment, the system includes a grasping robot and means for determining a plurality of grasping positions, especially collision-free and / or graspable grasping positions, especially a grasp sampler (English: “graspsampler”). In one embodiment, the system and / or its means includes means for selecting one of the determined grasping positions, especially a processing unit, which is designed to select one of the determined grasping positions. In one embodiment, the system and / or its means includes means for grasping an object. In one embodiment, the system and / or its means includes means for storing grasping success, especially a processing unit and / or especially a memory, especially data-connected to the processing unit.

[0037] In one embodiment, the system includes means for determining grasping actual data and / or means for determining holding actual data, especially at least one sensor, which is designed to collect at least one parameter related to grasping actual data and / or collect parameters related to holding actual data. In one embodiment, the system includes means for simulating, especially replicating, at least one grasping process, especially grasping, of the grasping robot in a simulated environment. In one embodiment, the system includes means for optimizing parameters based on data, especially parameters of the simulated environment, especially a processing unit.

[0038] In one embodiment, the system includes a grasping robot and a holding robot. In one embodiment, the system includes means for determining holding-grasping positions, especially a processing unit. In one embodiment, the system, especially the holding robot, has means for grasping, especially holding, the holding-grasping positions. In one embodiment, the system, especially the holding robot, has means for applying a force to the object.

[0039] In one embodiment, the system includes an imaging device. The imaging device described herein particularly preferably includes an imaging device for capturing digital and / or two-dimensional, especially three-dimensional images, and may particularly have at least one 2D camera, 3D camera, and / or at least two spatially separated cameras and / or at least one scanner, preferably for three-dimensional scanning. In one embodiment, the imaging device is designed to capture point clouds and / or color information, preferably three-dimensional point clouds, and more particularly point clouds with color information, especially color information corresponding to the points of the point cloud. Accordingly, the three-dimensional point cloud and / or color information captured particularly by means of the imaging device is referred to as a frame or image (captured by means of the imaging device), and in one embodiment is typically a three-dimensional image and / or an image with color information.

[0040] In one embodiment, the imaging device is arranged on at least one robot. In one embodiment, alternatively or additionally, the imaging device is arranged away from the robot, especially arranged to be able to capture, especially track, the object to be grasped and / or the grasped object.

[0041] The system and / or device in the sense of the present invention can be designed in hardware technology and / or software technology, especially having: at least one processing unit, preferably data-connected or signal-connected to a storage system and / or a bus system, especially a digital processing unit, especially a microprocessor unit (CPU), a graphics card (GPU), etc., and / or one or more programs or program modules. The processing unit can be designed for this purpose to: process the instructions of a program implemented and stored in the storage system; acquire input signals from the data bus; and send output signals to the data bus. The storage system can have one or more, especially different, storage media, especially optical, magnetic, solid-state, and / or other non-volatile media. The program can be provided to be able to embody or execute the method described herein, such that the processing unit can execute the steps of such method and thereby especially can operate or monitor the robot.

[0042] In one embodiment, a computer program product can have, especially can be, a computer-readable, non-volatile storage medium for storing a program or instructions, or a storage medium having a program or instructions stored thereon. In one embodiment, the execution of the program or instructions causes the system or controller, especially a computer or an array of multiple computers, to execute the program or instructions, such that the system or controller, especially one or more computers, executes the method described herein or one or more of its steps, or the program or instructions are designed for this purpose.

[0043] In one embodiment, one or more, especially all, steps of the method are performed fully or partially automatically, especially by a controller or its device. Brief Description of the Drawings

[0044] More advantages and features are given by the dependent claims and the embodiments. To this end, some are schematically shown as follows:

[0045] Figure 1 A system according to an embodiment of the present invention;

[0046] Figure 2 A system according to an alternative or additional embodiment; and

[0047] Figure 3 A schematic block diagram of a method according to an embodiment. Detailed Description of the Embodiments

[0048] Figure 1 Schematically shown is a system 1, which has a grasping robot 2, and the grasping robot has a gripper 3 for grasping an object 5 to be grasped. The object 5 to be grasped is shown on a schematic work plate 6 of the system 1, and one object 5 to be grasped is shown in solid lines. In one embodiment, the system 1 is designed to grasp a plurality of objects 5, 5' (shown in dotted lines) in a container (not shown), especially for performing the method described herein. In one embodiment, the system 1 may have a photographing device 8, which is shown in dotted lines in Figure 1 and is not arranged on the robot 2. In one embodiment, the photographing device may be arranged on the robot 2, especially on the flange of the robot 2. In addition, Figure 1 the system in

[0049] Figure 2 Schematically shown is a system 1 having a holding robot 10. In Figure 2Also shown is a robot 2 that has grasped the object 5 to be grasped. Similar to Figure 1 that, Figure 2 the imaging device 8 and the processing unit 7 are shown in dashed lines, and the processing unit is in a data connection (also shown in dashed lines) with the robot 2 and in particular with the holding robot 10. The holding robot 10 is depicted in Figure 2 such that it has grasped the determined holding-grasping position and holds the object that has been grasped by the grasping robot in a direction opposite to the grasping direction of the grasping robot 2. In some embodiments, the holding robot 10 can compensate for gravity here, especially in cases where the object is heavy and / or large. Further, in Figure 2 it is also shown by an arrow that the holding robot 10 can exert a force on the object 5, especially step by step, and this force counteracts the force exerted by the robot 2. In some embodiments, this force can be rotational (opposite to the rotational force of the robot 2), especially it can be increased step by step until the robot 2 loses the object or drops the object. In some embodiments, based on the maximum force used, parameterized values for simulating the environment or grasping simulation can be derived, especially these values can be transferred to the simulation. This is represented in Figure 2 by the data connection to the processing unit 7 shown in dashed lines. Further, the grasped object 5 can be tracked by the imaging device 8 during the above process or method, especially the trajectory of the object is recorded, and in some embodiments especially continuously and / or in time steps. The imaging device 8 can be mounted externally as shown by the dashed lines in Figure 2 or is not arranged on at least one of the robots 2, 10, or in some embodiments is arranged on the flange of at least one of the robots 2, 10. Figure 2 The system 1 shown in

[0050] Figure 3 can then be constructed at least substantially identically in the simulation environment such that the simulated (held) grasp on the object can match or be matched to the grasp on the object in reality, especially in some embodiments the simulation can be repeated until the simulated grasp is at least substantially (or within a preset reproduction accuracy boundary) the same as the real grasp. Figure 3 Also schematically shown in Figure 3S16 in particular relates to grasping the object 5 based on the grasping position selected in S14. S18 schematically represents storing the grasping success of the selected grasping position, in particular in a database or memory. These steps can be repeated, in particular as indicated by the dashed arrows. In some embodiments, the grasping actual data can also be determined separately, and optimization can be used, in particular only based on the stored grasping success data, to improve or optimize the grasping sampler.

[0051] In addition, Figure 3 The step S20 of determining the actual data to be held is also schematically shown in FIG. 1 . In the illustrated embodiment, determining the actual data to be held S20 comprises determining a holding-grasping position S22 on the object 5 grasped by the grasping robot. In one embodiment, as previously described in Figure 2 As shown in , the gripping position of the holding robot 10 is arranged to be able to exert a force on the object opposite to the force exerted by the gripping robot 2 on the object 5. As shown by way of example, step S24 follows: gripping is performed by, in particular, the holding robot 10 at the determined holding-gripping position. Then as Figure 3 As exemplarily shown in FIG. 5 , step S26 is: applying a force in a direction opposite to the force of the gripping robot 2 on the object 5 , in particular, applying a force on the object 5 .

[0052] In some embodiments, determining to retain the actual data S20 may follow determining to grab an object S16 or to store S18 of the actual data S10 , as particularly indicated by a dashed arrow between S18 and S20 .

[0053] In addition, Figure 3 It is also shown that at least one of the aforementioned or above-mentioned (actual) grasping processes is simulated S30. In one embodiment, in the simulation S30, an attempt is made to reflect the actual grasping process in the simulation environment, in particular as closely as possible. Figure 3 Optimization S32 of parameters is shown by way of example. Optimization S32 is based on previously determined actual data for grabbing and / or determined actual data for holding. Furthermore, as indicated by the dashed line, the method may include a step S40 of controlling, moving and / or monitoring the robot 2, 10, in particular during the grabbing process, based on the determined optimization parameters, which in one embodiment are transferred to the control data for the robot 2, 10 for this purpose.

[0054] Although exemplary embodiments have been described in the foregoing description, it should be noted that there may be many variations. It should also be noted that the exemplary embodiments are merely examples and should not form any limitation on the scope of protection, application, and construction. On the contrary, the foregoing description can impart to those skilled in the art the teaching of implementing conversions of at least one exemplary embodiment, wherein various changes, especially regarding the functions and arrangements of the said components, can be implemented without departing from the scope of protection of the present invention, for example, can be obtained according to the claims and their equivalent feature combinations.

[0055] List of Reference Numerals

[0056] 1 System

[0057] 2 Gripping Robot

[0058] 3 Grasper

[0059] 5, 5’ Object to be Gripped

[0060] 6 Working Surface

[0061] 7 Processing Unit

[0062] 8 Imaging Device

[0063] 10 Holding Robot

[0064] 30 Method

[0065] S10 Determine Gripping Actual Data

[0066] S12 Determine Gripping Position

[0067] S14 Select the Determined Gripping Position

[0068] S16 Grip Object

[0069] S18 Store Gripping Success Situation

[0070] S20 Determine Holding Actual Data

[0071] S22 Determine Holding-Gripping Position

[0072] S24 Grip at the Holding-Gripping Position

[0073] S26 Apply Force

[0074] S30 Simulate

[0075] S32 Optimize

[0076] S40 Control / Move the Robot.

Claims

1. A method (30) for automatically optimizing parameters of a robot-assisted grasping process, comprising - determining grasping actual data (S10), wherein the grasping actual data (S10) describes at least one grasping success situation of a grasping position on an object grasped by a grasping robot; and / or - determining holding actual data (S20), wherein the holding actual data (S20) describes at least one parameter related to the grasping process, in particular a parameter related to grasping, in particular a force on an object grasped by the grasping robot; - simulating (S30), in particular replicating at least one grasping process, in particular a grasping, in a simulation environment based on the determined grasping actual data (S10) and / or the determined holding actual data (S20); - optimizing (S32) the parameters based on data, in particular optimizing the parameters based on the determined grasping actual data (S10) and / or based on the determined holding actual data (S20) to determine the optimized parameters.

2. The method (30) according to the preceding claim, characterized in that, The data-based optimization (S32) includes automatically adjusting parameters, in particular automatically adjusting the quality of the simulation, the dynamic parameters of the simulation and / or the forces of the simulation, in particular friction.

3. The method (30) according to any one of the preceding claims, characterized in that, The data-based optimization (S32), in particular the optimization of the parameters, is based on a cost function of the grasping actual data (S10) and the grasping simulation data and / or the holding actual data (S20) and the holding simulation data, wherein the grasping simulation data describes at least one simulated grasping parameter, in particular the grasping success situation of a grasping position in the simulation, and wherein the holding simulation data describes at least one simulated holding parameter, in particular the force of the simulated grasping.

4. The method (30) according to any one of the preceding claims, characterized in that, The optimization is performed by means of a gradient-free optimizer.

5. The method (30) according to any one of the preceding claims, characterized in that, The determining of the grasping actual data (S10) includes: - randomly placing an object to be grasped; - determining (S12) a plurality of grasping positions on the object to be grasped, in particular collision-free and / or graspable grasping positions; - selecting (S14) one of the determined grasping positions; - grasping (S16) the object based on the selected grasping position, in particular using a grasping robot; - storing (S18) the grasping success situation at the selected grasping position; - repeating the above steps until a predetermined termination condition is met and / or until a predetermined number of repetitions, in particular to obtain actual data, especially a pre-sorted set of grasps, more particularly a pre-sorted set of grasps according to the grasping success situation.

6. The method (30) according to the preceding claim, characterized in that, The selection (S14) is based on the selection frequency of the determined grasping positions, and in particular selects the determined grasping position with the lowest selection frequency.

7. The method (30) according to claim 5 or 6 above, characterized in that, The random placement is performed by the grasping robot, in particular, the object is moved by the grasping robot to a random position and then dropped.

8. The method (30) according to any one of the preceding claims, characterized in that, The determining of the holding actual data (S20) includes: - Determine a holding - grasping position (S22), wherein the holding - grasping position is arranged on an object grasped by the grasping robot such that, via the holding - grasping position, a force can be exerted on the grasped object that acts in a direction opposite to the force on the grasped object, in particular opposite to the grasping direction of the grasping robot. - Grasp (S24), in particular hold, the object at the determined holding - grasping position; - Exert a force (S26) on the object that acts, in particular, in a direction opposite to the grasping direction of the grasping robot.

9. The method (30) according to the preceding claim, characterized in that, The force exertion includes gradually increasing the force in predetermined steps, in particular until the grasping robot loses the object and / or until a maximum force is exceeded.

10. The method (30) according to any one of the preceding claims, characterized in that, Determining grasping actual data (S10) and / or holding actual data (S20) includes photographing the object by means of an imaging device, in particular determining the pose of the object, in particular the pose varying over time.

11. A method (30) for controlling a robot to perform a robot - assisted grasping process, comprising: - Based on the optimized parameters according to any one of the preceding claims 1 to 10, determine control data, - Based on the optimized control data, control and / or move (S40) the robot to perform a robot - assisted grasping process.

12. The method (30) according to any one of the preceding claims, characterized in that, Transmit the determined optimized parameters to, in particular apply them to, a grasping application with multiple objects in a container.

13. A system (1) for operating and / or monitoring at least one robot, in particular a grasping robot, the system being designed to perform the method (30) according to any one of the preceding claims.

14. A computer program or computer program product, wherein, The computer program or computer program product contains instructions, in particular instructions stored on a computer - readable and / or non - volatile storage medium, which, when executed by one or more computers or by the system (1) according to claim 13, cause the one or more computers or the system (1) to perform the method (30) according to any one of claims 1 to 12.

Images