Transfer of object gripped by robot from receiving pose to target pose

Abstract

A method comprising the steps of: determining (S60) a prediction of an attempt to transfer an object (41, 42,..., 45) grasped by a robot (10) from a receiving pose to a target pose based on grasping data generated by a grasping process in which the robot grasps the object in the receiving pose; and at least one of the following two steps: providing (S50) the capture data, and / or controlling and / or monitoring (S70) the transfer attempt based on the determined prediction result. The invention also relates to a system or a computer program (product).

Description

Technical Field

[0001] The present invention relates to a method for controlling and / or monitoring an attempt to transfer an object gripped by a robot from a receiving pose to a target pose and / or for determining an expected result of the transfer attempt, as well as a system or a computer program or a computer program product for performing the method described herein. Summary of the Invention

[0002] An object of one embodiment of the present invention is to improve the transfer of an object gripped by a robot from a receiving position to a target position and / or to better predict the outcome of such a transfer attempt.

[0003] In particular, the object of the invention is achieved by a method having the features of claim 1 or 4. Claims 12 and 13 claim a system or a computer program or a computer program product for carrying out the method described herein. The dependent claims relate to advantageous developments.

[0004] According to one embodiment of the invention, an attempt is made to transfer an object gripped by the robot from a receiving pose to a target pose by means of (a corresponding movement of) the robot, which is referred to herein as transfer (attempt).

[0005] In one embodiment, the robot has at least three, preferably at least six, and in an extended embodiment at least seven joints or (motion) axes; in one embodiment, the robot has a robot arm, the robot arm has at least three, preferably at least six, and in an extended embodiment at least seven joints or (motion) axes, wherein the robot (arm) preferably has a drive for adjusting the joints or (motion) axes, preferably an electric (dynamic) drive, and / or an end effector for temporarily or non-destructively releasably grasping an object, which is generally referred to as a gripper in this document, in one embodiment, is a mechanical (action) gripper, a magnetic (action) gripper, preferably an electromagnetic (action) gripper and / or a pneumatic (action) gripper or a suction cup gripper, and the robot (arm) uses the gripper to (respectively) grasp the object grasped by the robot, in one embodiment, mechanically, preferably frictionally and / or form-fittingly, magnetically, preferably electromagnetically, and / or pneumatically or using negative pressure to achieve gripping, or is designed or used for this purpose. Accordingly, in one embodiment, the object gripped by the robot comprises an object gripped by the robot (arm / gripper) in such a manner, and / or in one embodiment, the gripping of the object by the robot comprises gripping mechanically, preferably frictionally and / or form-fittingly, magnetically, preferably electromagnetically, and / or pneumatically, or using negative pressure, which can in particular be such gripping. In one embodiment, activation of the gripper comprises a closing movement of the gripper and / or generation of an (electro)magnetic gripping force and / or a gripping force based on negative pressure, which can in particular be generated thereby.

[0006] For such robot-assisted gripping processes and subsequent transfers, the present invention is particularly advantageous due to boundary conditions, in particular objects that are often difficult and / or unreliable to grip (by robot), the need for rapid transfer movements of the robot, etc.

[0007] In one embodiment, a pose in the sense of the present invention comprises a one-dimensional, two-dimensional or three-dimensional position and / or a one-dimensional, two-dimensional or three-dimensional orientation. In one embodiment, the pose of an object, in particular an object gripped by a robot, is the pose of the object relative to the robot, in an extended embodiment relative to its gripper (in its gripper), or relative to the environment. In one embodiment, the pose of the robot is determined by the position of its joints or (motion) axes and / or the position and / or orientation of its gripper. In one embodiment, the received pose and / or the target pose is in particular predetermined, in an extended embodiment before the transfer attempt is performed.

[0008] According to one embodiment of the present invention, the predicted or (by means of computational technology) predicted result of the transfer attempt is determined based on or depending on data, which are based on or depending on the process of grasping an object by the robot in a received posture (during the grasping process), in particular the process of grasping an object actually attempted and preferably detected by sensor technology, which are correspondingly referred to as grasping data in this article.

[0009] This is based on the recognition that during the (grasping) process of grasping an object, different parameters such as force, image, etc. have values ​​that allow a (more) precise and / or (more) reliable prediction or forecast of the transfer result, in particular whether the transfer is successful or not, and based on this recognition, these parameter values ​​are detected and used for the prediction or forecast.

[0010] The result predicted in this way can be used particularly advantageously, in particular for controlling and / or monitoring transfer attempts.

[0011] Accordingly, according to one embodiment of the present invention, the transfer attempt (for which the result is predicted or the predicted result is determined) is controlled and / or monitored based on the determined prediction result.

[0012] Therefore, according to one embodiment of the present invention, the method according to the present invention comprises the following steps:

[0013] - determining a predicted outcome of attempting to transfer an object grasped by the robot from a received pose to a target pose based on grasping data generated based on the robot grasping the object at the received pose;

[0014] And the following steps:

[0015] - providing the grasping data (generated based on the grasping process of the robot grasping the object in the receiving position);

[0016] And / or the following steps:

[0017] - Controlling and / or monitoring the transfer attempt based on the determined prediction result.

[0018] Providing within the meaning of the present invention may in particular include, in particular consist of, determining, in particular detecting, storing, sending and / or receiving, retrieving or loading (from a memory), and / or processing captured data.

[0019] In one embodiment, controlling and / or monitoring the transfer attempt includes interrupting the attempt, in an extension, releasing the object (held by the robot) before reaching the target position, in particular instead of continuing to travel along the transfer trajectory of the robot that is predefined for the transfer, in one embodiment.

[0020] - the robot drops the object, or

[0021] The object is brought to a standstill (by the robot) in a rest position, wherein in one embodiment, the rest position is determined based on the determined prediction result.

[0022] In one embodiment, by interrupting the transfer (attempt) in this way, unnecessary movements of the robot can be omitted, thereby saving, in particular, energy and / or (process or cycle) time.

[0023] In one embodiment, the (probability of success of) a renewed transfer attempt can be increased by stopping the object in a specific position, in particular by stopping the object in a position in which it can subsequently be grasped well or better by the robot.

[0024] In one embodiment, the transfer trajectory of the robot includes the path of the robot, preferably its gripper, and in an embodiment also the speed when traveling over this trajectory. In one embodiment, a transfer attempt within the meaning of the present invention includes an attempt to travel over this transfer trajectory or transfer path with the robot.

[0025] Additionally or in addition to interrupting the transfer (attempt), in one embodiment, controlling and / or monitoring the transfer attempt also includes (starting, in particular executing) one or more countermeasures, which in one embodiment are

[0026] - changing the robot's grip on the object, preferably during the transfer attempt; and / or

[0027] - preferably during the transfer attempt, changing the transfer trajectory of the robot pre-set for the transfer;

[0028] In one extension,

[0029] - changing, preferably at least temporarily increasing, the gripping force with which the robot grips the object; and / or

[0030] - Changing the pose of the object, which is preferably still being gripped, relative to the robot, in particular its gripper.

[0031] Additionally or alternatively, in one embodiment, controlling and / or monitoring the transfer attempt further comprises:

[0032] - changing the robot's grip on the object, in particular the same object, during a subsequent, in particular repeated, transfer attempt; and / or

[0033] - changing a predefined transfer trajectory of the robot during or for a subsequent, in particular a renewed, transfer attempt;

[0034] In one extension,

[0035] - changing, preferably increasing, the gripping force with which the robot grips the object in a subsequent, in particular a renewed, transfer attempt; and / or

[0036] During a subsequent, in particular repeated, transfer attempt, the position of the object relative to the robot, in particular its gripper, is changed.

[0037] Thus, in one embodiment, it is possible to advantageously react to predicted problems during the current transfer attempt, in particular predicted failures, in particular predicted errors or erroneous attempts, and / or to increase the probability of success of the current and / or subsequent transfer attempts, for example, by causing the robot to re-grasp and / or grasp more tightly, etc., in the current or subsequent transfer attempt, and / or by causing the robot to drive (more) slowly along the planned path or re-plan the path when a problem is predicted, etc.

[0038] Additionally or alternatively, in one embodiment, controlling and / or monitoring the transfer attempt also includes controlling processes associated with the transfer, in particular the performed transfer, in particular controlling further processing of the object after the transfer. Thus, in one embodiment, such transfer-related processes can be improved.

[0039] In one embodiment, a prediction result is determined by processing the provided captured data with the aid of data processing based at least in part on machine learning, in an extended embodiment with the aid of at least one machine learning model or machine learning algorithm, which preferably correlates the captured data and the prediction result or maps the captured data to the prediction result, and in one embodiment the prediction result is determined with the aid of at least one artificial neural network. In one embodiment, the artificial neural network has a 2D convolutional layer, which is particularly advantageous for (data) processing of captured data including image data. Additionally or alternatively, in one embodiment, the artificial neural network includes a 1D convolutional layer, which is particularly advantageous for (data) processing of (data) captured data including time series.

[0040] By means of (this) at least partial machine learning or data processing based on machine learning, the outcome of a robot-assisted transfer attempt can be predicted particularly well, in particular (more) precisely, (more) reliably and / or for different and / or complex processes based on the gripping data of the robot's gripping process of the object.

[0041] In one embodiment, the method comprises the following steps:

[0042] -supply

[0043] - learning data, which are generated based on gripping processes of objects gripped by robots, preferably robots of the same type, and

[0044] Result data, in particular annotations or evaluations, generated based on the robot's attempt to transfer an object, in particular based on a preferably automated annotation or evaluation of this attempt; and

[0045] - Based on the learning data and the result data, training of the data processing is performed, in particular machine learning or training, preferably supervised learning or training.

[0046] Thus, based on the gripping data of the gripping process of the robot gripping an object, the result of a robot-assisted transfer attempt can be predicted particularly well, in particular (more) precisely, (more) reliably and / or for various and / or complex processes.

[0047] In one embodiment, the process of grasping an object in the sense of the present invention includes:

[0048] - driving the robot towards the object, and / or

[0049] - bringing the robot, in particular its gripper, into contact with the object, and / or

[0050] - activating a gripper of the robot (for gripping in particular mechanically, preferably with a friction fit and / or a form fit, magnetically, preferably electromagnetically, and / or pneumatically or by means of negative pressure), and / or

[0051] After activating the gripper or while the gripper remains in the activated position, the robot performs a retraction movement, in particular together with the gripped object, and in one embodiment, moves to the starting position of the transfer trajectory while the gripper is activated.

[0052] In particular, such a grasping process may provide parameter values ​​that allow a (more) precise and / or (more) reliable prediction or forecast of the outcome of the transfer, in particular the success or failure of the transfer.

[0053] In one embodiment, the gripping data and / or learning data and / or result data are (respectively) determined by one or more sensors on the robot side or fixed to the robot, in an extended embodiment by one or more drive sensors and / or one or more joint sensors and / or one or more gripper sensors.

[0054] Advantageously, such sensors are usually already present for control and / or monitoring. Additionally or alternatively, in one embodiment, such sensors can be used to determine or provide advantageous data even when relocating the robot and / or in confined environments.

[0055] Additionally or alternatively, in one embodiment, the gripping data and / or learning data and / or result data are (respectively) determined by one or more (robot) external sensors, in particular environment-side sensors, preferably by one or more 2D cameras and / or one or more 3D cameras.

[0056] Such sensors can generally determine or provide particularly informative data that has hitherto not been utilized in robot-assisted gripping and transfer processes and / or (also) allow a (more) precise and / or (more) reliable prediction or forecast of the transfer result, in particular the success or failure of the transfer.

[0057] Additionally or alternatively, in one embodiment, the gripping data and / or learning data and / or result data are (respectively) determined by at least one data model of the robot and / or at least one data model of the object, in an extended embodiment by a CAD data model or the like.

[0058] Such a data model may, for example, allow or improve the determination of the pose of an object, in particular its pose relative to a robot (gripper), in particular by increasing the accuracy and / or reducing the computational effort.

[0059] Additionally or alternatively, in one embodiment, the gripping data and / or the learning data and / or the result data (respectively) depend on a force, in one embodiment, a driving force and / or a contact force and / or a gripping force, which can in particular be specified or described. For a more compact description, antiparallel force couples or torques are also collectively referred to herein as forces.

[0060] Additionally or alternatively, in one embodiment, the gripping data and / or the learning data and / or the result data (respectively) depend on a pose of the robot and / or the object, and such a pose and / or its time derivatives, in particular the velocity and / or acceleration of the robot or the object, can be specified or described. In a further embodiment, the gripping data and / or the learning data and / or the result data (respectively) include images or image data of the robot and / or the object.

[0061] In particular, such force data, posture data or image data may allow a (more) precise and / or (more) reliable prediction or forecast of the transfer result, in particular whether the transfer was successful or not.

[0062] Additionally or alternatively, in one embodiment, the gripping data and / or learning data depend on values ​​detected by one or more sensors during the gripping process of the robot gripping the object, in a particularly preferred embodiment, on

[0063] - while the robot is driving towards an object or receiving a pose, and / or

[0064] - during contact between the robot, in particular its gripper, and / or

[0065] - during activation of a gripper of the robot (for gripping in particular mechanically, preferably with a friction fit and / or a form fit, magnetically, preferably electromagnetically, and / or pneumatically or by means of negative pressure), and / or

[0066] - while the gripper remains active (to grip an object), and / or

[0067] - during a retraction movement of the robot after activating the gripper or while the gripper remains in an activated position, in particular together with the (possibly) gripped object, in an extended embodiment during travel to the starting position of the transfer trajectory, preferably values ​​detected (respectively) by one or more sensors.

[0068] Accordingly, in one embodiment, the method according to the invention comprises the following steps: providing gripping data generated by a gripping process of the robot gripping an object in a receiving position, the gripping data being dependent on values ​​detected during the gripping process, preferably by one or more sensors, and in an extension, further comprising the following steps: detecting values ​​during the gripping process, preferably by one or more sensors, and providing gripping data based on the detected values, wherein these values ​​are particularly preferably

[0069] - while the robot is driving towards an object or receiving a pose, and / or

[0070] - during contact between the robot, in particular its gripper, and / or

[0071] - during activation of the gripper of the robot (for gripping in particular mechanically, preferably with a friction fit and / or with a form fit, magnetically, preferably electromagnetically, and / or pneumatically or by means of negative pressure), and / or

[0072] - while the gripper remains active (to grip an object), and / or

[0073] - during a retraction movement of the robot after activation of the gripper or while the gripper remains activated, in particular together with the (possibly) gripped object, in an embodiment, during movement to the starting position of the transfer trajectory,

[0074] Be detected.

[0075] As previously mentioned, this embodiment is based on the recognition that during the gripping process, parameters such as forces and scenes can have numerical values, such as force values, image data, etc., which allow for a (more) precise and / or (more) reliable prediction or forecast of the transfer result, in particular whether the transfer was successful, and based on this recognition, these parameter values ​​are detected and used for the prediction or forecast. This is particularly evident when detecting forces, in particular driving forces and / or contact forces, when the object is in contact and / or the gripper is activated and / or in an activated gripper state. These forces accordingly constitute particularly advantageous numerical values ​​or gripper data, but the invention is not limited thereto. Similarly, image data or posture data, etc., generated in particular when the object is in contact and / or the gripper is activated and / or in an activated gripper state, can also achieve particularly good predictions.

[0076] Additionally or alternatively, in one embodiment, a predicted outcome is determined before or during a transfer attempt.

[0077] In one embodiment, a reaction to the prediction can thus be particularly advantageous.

[0078] In one embodiment, the captured data and / or the learning data (respectively) include image data and / or time series.

[0079] Such data in particular allow a (more) precise and / or (more) reliable prediction or forecast of the outcome of the transfer, in particular the success or failure of the transfer.

[0080] In one embodiment, the captured data and / or the learning data (respectively) do not include image data and / or time series.

[0081] In one embodiment, machine learning or training (of data processing) can thereby be improved, in particular accelerated and / or simplified.

[0082] In one embodiment, the determined prediction result and / or outcome data depend on whether the subject was lost during the transfer attempt (Verlust), which may in particular lead to a prediction or assessment of a poor or unfavorable outcome. In general, the determined prediction result within the meaning of the present invention may include, in particular, the success of the transfer attempt.

[0083] Additionally or alternatively, in one embodiment, the determined prediction result and / or result data depend on the posture of the object at the end of the transfer attempt and / or a predetermined value range (Wertebereich), in particular when the deviation between the predicted posture or the actual posture and the theoretical posture or the target posture is within a predetermined allowable error range, or when the determined prediction result can provide the deviation or the degree of deviation, in one embodiment in a continuous form or in a two-level or multi-level discrete form, then it can be predicted or marked as a better or good result.

[0084] Therefore, in an advantageous and particularly simple embodiment, when training data processing, if the object is not (any longer) grasped by the robot after a (training) transfer attempt, the (training) transfer attempt can be evaluated or labeled as failed or bad; if the object is (still) grasped by the robot after a (training) transfer attempt, the (training) transfer attempt can be evaluated or labeled as successful or good.

[0085] Typically, in one embodiment (binary): the result that can be predicted is "(expected) success" or "(expected) failure", or it is determined as the predicted result.

[0086] In another embodiment that is also particularly advantageous, for example when training data processing, the (training) transfer attempt can be evaluated or labeled based on the size of the deviation between the theoretical or target pose of the object and the pose achieved after the (training) transfer. In one extended embodiment, the smaller the deviation, the better. In another extended embodiment, once the deviation is outside a pre-set allowable error range, it is evaluated or labeled as failed or bad.

[0087] Accordingly, here it can also be, for example, a (binary) "(expected) success" or "(expected) failure", or the deviation between the theoretical or target posture of the object and the (expected) achieved posture can be predicted as a result or determined as a prediction result, in particular as a deviation in a graded discrete form or a continuous, if necessary scaled, form.

[0088] According to one embodiment of the present invention, a system is provided, which is designed in particular in terms of hardware and / or software, in particular programming, for carrying out the method described herein and / or comprises one or preferably more of the following devices:

[0089] - means for determining a predicted outcome of an attempt to transfer an object held by the robot from the receiving pose to a target pose (transfer attempt) based on gripping data of a gripping process of the robot gripping an object in the receiving pose;

[0090] - means for controlling and / or monitoring said transfer attempt based on the determined prediction result, in particular

[0091] - means for interrupting the attempt, in one embodiment by delivering the object before reaching the target position, in a further embodiment by stopping the object in a determined stopping position;

[0092] - means for initiating, in particular executing, a countermeasure, in particular for changing the robot's grip on the object and / or for changing a transfer trajectory of the robot predefined for the transfer; and / or

[0093] - means for changing the robot's grip on the object and / or the pre-set robot's transfer trajectory during subsequent transfer attempts; and / or

[0094] - To control the processes related to the transfer, in particular the further processing of the object after the transfer.

[0095] step processing device;

[0096] - means for providing captured data;

[0097] - data processing (means) based at least in part on machine learning, in particular at least one artificial neural network, for determining a prediction result by processing the provided scraped data;

[0098] - means for providing learning data generated based on the grasping process of the robot grasping the object and data based on the results of attempts to transfer the object by the robot;

[0099] - means for training the data processing (device) based on these learning data and result data;

[0100] - at least one robot-side sensor for determining grasping data, learning data and / or result data;

[0101] - at least one external sensor for determining grasping data, learning data and / or result data;

[0102] at least one data model for the robot and / or the object, for determining gripping data, learning data and / or result data;

[0103] - means for providing values ​​detected during a gripping process of the robot gripping an object for determining gripping data, learning data and / or result data, in particular means for detecting these values.

[0104] Systems and / or devices within the meaning of the present invention can be designed in hardware and / or software technology and, in particular, include at least one processing unit, in particular a digital processing unit, in particular a microprocessor unit (CPU), a graphics card (GPU), etc., preferably with a data or signal connection to a storage system and / or a bus system; and / or one or more programs or program modules. The processing unit can be designed to process instructions implemented as a program stored in the storage system, acquire input signals from a data bus, and / or transmit output signals to the data bus. The storage system can include one or more, in particular different, storage media, in particular optical, magnetic, solid, and / or other non-volatile media. A program can be provided that embodies or executes the methods described herein, enabling the processing unit to perform the steps of such methods and thereby control and / or monitor an attempt to transfer an object gripped by the robot from a receiving position to a target position, or to determine a predicted outcome of such an attempt. In one embodiment, the computer program product can include, in particular, a computer-readable, non-volatile storage medium for storing the program or instructions, or a storage medium on which the program or instructions are stored. In one embodiment, executing a program or instruction by a system or controller, in particular a computer or a plurality of computer arrays, causes the system or controller, in particular one or more computers, to perform the method described herein or one or more steps thereof, or the program or instruction is designed for this purpose.

[0105] In one embodiment, one or more, in particular all, steps of the method are fully or partially implemented by a computer, or one or more, in particular all, steps of the method are fully or partially automated, in particular executed by the system or its device.

[0106] In one embodiment, the system comprises a robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0107] Further advantages and features are given by the dependent claims and the exemplary embodiments. To this end, a partial schematic diagram is shown:

[0108] Figure 1 A system according to an embodiment of the present invention; and

[0109] Figure 2 This is a method according to one embodiment of the present invention. DETAILED DESCRIPTION

[0110] Figure 1 A system according to an embodiment of the present invention is shown, which includes a robot 10 , a data processing device 20 and cameras (sensors) 31 , 32 .

[0111] The robot 10 uses its gripper 11 to successively grasp the objects 41 , 42 , . . . arranged in the container 40 and transports them to the storage surface 400 or attempts to do so.

[0112] In the training phase, first, during the grasping process of the robot 10 grasping an object, the camera 31 and / or one or more robot-side sensors are used to detect measurement values ​​and, if necessary, further process them (e.g., filter them) and store them as learning data ( Figure 2 : step S10), wherein the sensor is particularly advantageously shown as a contact sensor or contact force sensor 12 on the gripper 11. Additionally or alternatively, the driving force can also be detected and, if necessary, stored as learning data after further processing (e.g. filtering, etc.). Preferably, the learning data are based on (measured) values, at least some of which are detected during the approach to the object and / or contact between the object and the gripper, during activation of the gripper and / or while the gripper remains activated, in particular during a retraction movement of the robot from a takeover position of the transfer trajectory or transfer path to a starting position.

[0113] After the crawl process, a transfer (attempt) is performed accordingly ( Figure 2 : Step S20), evaluate the result and store the result as result data ( Figure 2 : Step S30), for example, by means of a camera 32 and image processing, quantifying the deviation of the position reached by the corresponding object at the end of the transfer from the desired theoretical position or target position, and / or by means of a sensor 12 detecting a possible loss of the object during the transfer (attempt), and evaluating the corresponding transfer (attempt) as successful or unsuccessful or as a success or failure.

[0114] In step S40 , the learning data and result data thus obtained are used to train a data processing at least partially based on machine learning, in this embodiment, to train at least one artificial neural network 21 .

[0115] Next, in the operation phase, similarly to the learning data, the grasping data ( Figure 2 :Step S50).

[0116] The data processing device 20 then uses the trained artificial neural network 21 to determine the predicted outcome of the transfer attempt after the gripping process ( Figure 2 :Step S60).

[0117] Based on the prediction result, the data processing device 20 controls and / or monitors the transfer attempt ( Figure 2 :Step S70).

[0118] For example, if a failure is predicted based on the gripping data generated during the gripping process of the object in the receiving position by the robot, the data processing device can cause the object gripped in step 50 to fall in the center of the container 40, or stop it in a (particularly) (more) favorable position for gripping and then re-grip the object, in one embodiment, specifically grab the stopped object and make another transfer attempt. Similarly, in step S70, the data processing device 20 can also increase the gripping force for gripping the object and / or change the gripping method (Griff) during the transfer attempt. Additionally or alternatively, in step S70, processes related to the transfer can also be controlled based on the determined prediction result, such as transporting the object after the transfer.

[0119] Figure 1 Shown by way of example are the receiving poses of objects 41 , 42 and 43 , the target pose of an object 45 , and the transfer trajectory or transfer path of an object 44 held by the robot 10 or its gripper 11 .

[0120] While exemplary embodiments have been described above, it should be noted that many variations are possible.

[0121] Therefore, for the sake of simplicity, the above description uses the same container 40 and the objects 41, 42, ... therein as an example to explain the data processing or training of the artificial neural network 21 and the subsequent control or monitoring. Preferably, a separate training phase can be carried out beforehand using other objects and / or other robots, which are preferably identical or similar to the objects whose transfer is to be controlled or monitored in the subsequent operating phase or the robot gripper 11 used therein, wherein the data processing or artificial neural network 21 can also advantageously be further trained in this operating phase.

[0122] It should also be noted that the exemplary embodiment is merely an example and does not constitute any limitation on the scope of protection, application, or configuration. On the contrary, the foregoing description provides a person skilled in the art with a method for modifying at least one of the exemplary embodiments, wherein various modifications, particularly with respect to the function and arrangement of the components, are possible without departing from the scope of protection of the present invention, such as may be obtained according to the claims and their equivalent feature combinations.

[0123] Reference Signs List

[0124] 10. Robot

[0125] 11. Grab

[0126] 12 Sensors

[0127] 20 Data processing device

[0128] 21 Artificial neural network (data processing device)

[0129] 31, 32 Camera (Sensor)

[0130] 40 containers

[0131] 41–45 objects

[0132] 400 storage surface.

Claims

1. A method for controlling and / or monitoring an attempt to transfer an object (41, 42, ..., 45) gripped by a robot (10) from a receiving pose to a target pose, the method comprising the following steps: - determining (S60) a predicted result of a transfer attempt based on grasping data generated based on a grasping process of the robot grasping the object in the receiving pose; as well as - controlling and / or monitoring (S70) the transfer attempt based on the determined prediction result.

2. The method according to claim 1, characterized in that Controlling and / or monitoring the transfer attempt includes: - interrupting the attempt, in particular delivering the object before the target position is reached, in particular stopping the object in a determined stop position; - countermeasures, in particular changing the grip of the robot on the object and / or changing the transfer trajectory of the robot predefined for the transfer; and / or - changing the robot's grip on the object and / or changing a pre-set transfer trajectory of the robot during a subsequent transfer attempt; and / or - controlling processes associated with the transfer, in particular further processing of the object after the transfer.

3. The method according to any one of the preceding claims, characterized in that The method comprises: providing the captured data (S50), and / or determining the prediction result according to any one of the following claims.

4. A method for determining a predicted outcome of attempting to transfer an object (41, 42, ..., 45) grasped by a robot (10) from a receiving pose to a target pose, the method comprising the steps of: - providing (S50) grasping data generated based on a grasping process of the robot grasping the object in the receiving posture; as well as - Determining (S60) a predicted outcome of the transfer attempt based on the provided fetch data.

5. The method according to the preceding claim, characterized in that The prediction result is determined by processing the provided captured data with the aid of data processing based at least in part on machine learning, in particular with the aid of at least one artificial neural network (21).

6. Method according to the preceding claim, characterized in that The following steps are also included: - providing (S10) learning data generated based on a grasping process of the robot grasping the object, and providing (S30) result data generated based on an attempt to transfer the object by the robot; and - Training (S40) the data processing based on the learning data and the result data.

7. The method according to any one of the preceding claims, characterized in that The grasping process of grasping an object includes: making the robot move towards and / or contact the object, and / or activating the gripper of the robot, and / or the robot's retreat movement after activating the gripper, in particular moving towards the starting position of the transfer trajectory.

8. The method according to any one of the preceding claims, characterized in that The captured data, learning data and / or result data are obtained by - at least one robot-side sensor (12), - at least one external sensor (31, 32), and / or - at least one data model of the robot and / or the object to determine, and / or depend on - forces, in particular driving forces, contact forces and / or gripping forces, and / or - the pose of the robot and / or object; and / or The gripping data and / or learning data depend on values ​​detected during a gripping process of an object by the robot; and / or The prediction result is determined before and during the transfer attempt.

9. The method according to any one of the preceding claims, characterized in that The captured data and / or learning data include image data and / or time series.

10. The method according to any one of the preceding claims 1 to 8, characterized in that The captured data and / or learning data do not include image data and / or time series.

11. The method according to any one of the preceding claims, characterized in that The determined prediction result and / or the result data depend on the loss of the object during the transfer attempt and / or on the position of the object at the end of the transfer attempt and / or on a predefined value range.

12. A system designed to perform the method according to any one of the preceding claims and / or comprising: - means for determining a predicted outcome of an attempt to transfer an object (41, 42, ..., 45) grasped by the robot (10) from a receiving pose to a target pose based on grasping data generated by the robot during a grasping process of grasping the object in the receiving pose; as well as - means for providing said captured data, and / or means for controlling and / or monitoring transfer attempts based on the determined prediction result.

13. A computer program or computer program product, wherein: The computer program or computer program product comprises instructions, in particular instructions stored on a computer-readable and / or non-volatile storage medium, which, when executed by one or more computers or a system according to claim 12 , cause the one or more computers or the system to perform the method according to any one of claims 1 to 11 .

Images