Method for handling an object by means of a robotic arm, and device comprising a robotic arm for handling an object

Abstract

The invention relates to a method for handling an object (20) by means of a robotic arm (10), comprising the following steps: a. grasping at least one object (20) and measuring at least one measured value of at least one object characteristic of the object (20) by means of at least one sensor (31, 32); b. ascertaining at least one handling characteristic by retrieving the measured value from a database and / or by means of inference using a statistical model by inputting the measured value; c. creating a control data set for the robotic arm (10), taking into consideration the handling characteristic; and d. handling the object (20) by means of the robotic arm (10) according to the control data set. The invention further relates to a device for handling objects (20), the device comprising a robotic arm (10), at least one sensor (31, 32) and a computer system (40).

Description

[0001] 'Method for handling an object by means of a robot arm and device comprising a robot arm for handling an object'

[0002] The invention relates to a method for handling an object by means of a robot arm according to claim 1 and a device comprising a robot arm for handling an object according to claim 11.

[0003] Robots are known in the prior art which pick up an object using a robot arm and a suitable gripper, move it to a predetermined storage location, and place it there (so-called pick & place robots). It is important that the robot's access and the subsequent movements performed with the object are adapted to the object in such a way that the object is not damaged. In particular, excessive or insufficient gripping force can damage the object, causing it to be crushed or fall. The manner in which the object must be held to prevent damage is highly object-specific. As a result, the object must be identified beforehand before the handling parameters are selected. A disadvantage of the method known in the prior art, however, is that it only allows the handling of identifiable and previously known objects.

[0004] There is therefore a great need for a method and device for handling an object, where the object is not necessarily known and / or identified in advance. The object should be handled as reliably, safely, smoothly, and quickly as possible without delay, slowing down, or interruption. Furthermore, the method should be cost-effective to implement, and the device should be cost-effective to manufacture and operate, with a further focus on making them as compact and space-saving as possible. The object of the invention is to provide such a method and device, overcoming the disadvantages of the prior art while requiring the smallest possible space.

[0005] This object is achieved in a surprisingly simple but effective manner by a method for handling an object according to the teaching of the independent main claim 1 and a device comprising a robot arm for handling an object according to the teaching of claim 11.

[0006] According to the invention, a method for handling an object by means of a robot arm is proposed, the method comprising the following steps: a. detecting at least one object and measuring a measured value of at least one object property of the object by means of at least one sensor; b. determining at least one handling property by querying the measured value in a database and / or by inference with a statistical model upon input of the measured value; c. creating a control data record for the robot arm taking into account the handling property; and d. handling the object by means of the robot arm according to the control data record.

[0007] Within the scope of the invention, it was recognized that the handling of unknown objects regularly fails because, if the object is not identified, no handling properties can be assigned to the object. Thus, the handling of the object fails because it is not known how the unidentified object is to be handled. Within the scope of the invention, it was recognized that for the assignment of handling properties, the precise identification of the object is not necessary; rather, object properties allow conclusions to be drawn about handling properties. The method enables such conclusions to be drawn reliably.

[0008] In order for the object to be able to be handled at all, at least one object in step a. must first come into the range of action of the robot arm and be detected by it. This means that the object must be made available in a way that initially enables the sensor to measure the measured value and later enables the robot arm to handle it. Detecting the object triggers the measurement by the sensor to determine the measured value of the object property. The sensor determines the measured value by measuring. It is conceivable that the sensor's measured value includes several identical object properties of several objects simultaneously or sequentially. For example, the sensor detects the total weight of several objects simultaneously or sequentially.It is further conceivable for the sensor to be part of a measuring device with a data processing unit, and for the measuring device to obtain the measured value by calculation in the data processing unit from the data determined by the sensor. Preferably, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, or 20 objects are detected simultaneously or consecutively. Further preferably, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, or 20 identical or different measured values ​​are measured. Even more preferably, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, or 20 sensors record at least one measured value. The term "object property" refers to a state variable and / or a characteristic attributed to the object, in particular a specific marking that allows conclusions to be drawn about the identity, type, and / or safe handling of the object.In particular, the term "object property" encompasses the interaction and / or cooperation with other objects and / or the environment. It also encompasses characteristics that are closely related to the object but are not necessarily and / or necessarily inherent in it. The term "measured value" refers to the quantization, expression, and / or digitization of an object property.

[0009] In step b., the measured value measured in step a. is queried in a database and / or entered into a statistical model. At least one handling property is uniquely stored in the database for any number of measured values. If the queried measured value is stored in the database, the object is identified and the database returns at least one handling property. Preferably, the type of object is also stored. Additionally or alternatively, the measured value is entered into a statistical model. The statistical model was trained before and / or is trained during the execution of the method using training data sets. The statistical model infers at least one, preferably an optimal and / or correct, handling property from the measured value.It is conceivable that the statistical model infers the at least one, preferably optimal and / or correct, handling property from the fact that all recorded and weighted measured values ​​have the highest agreement with the training measured values ​​of a specific other object and therefore the at least one handling property stored for this specific other object is the at least one, preferably optimal and / or correct, handling property. Even more preferably, it is conceivable that the statistical model infers the at least one, preferably optimal and / or correct, handling property from the fact that the individual recorded and weighted measured values ​​have the highest agreement with individual training measured values ​​and the handling property stored for the individual training measured values ​​is the at least one optimal and / or correct handling property.In other words, the inference of the statistical model corresponds to at least one handling property of a specific other object or the inference is formed from different handling properties of different objects. The advantage of the first approach is that, in particular, several handling properties are guaranteed to be coordinated with one another. The advantage of the second approach is that, particularly when there are several handling properties, these are more individually and therefore potentially more optimally and / or more accurately coordinated with the unknown detected object. In other words, for a measured value not stored in the database, an inference can nevertheless be made using the statistical model regarding at least one, preferably optimal and / or correct, handling property, without the object being or having to be identified.It is therefore irrelevant for the success and / or feasibility of the method what the object ultimately is; the object can still be assigned at least one handling property. Even more preferably, in step b. at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 or 20 identical or different handling properties are determined. It is particularly preferred if in step a. at least one first sensor measures at least one first measured value and at least one second sensor measures at least a second measured value, wherein in step b. at least one first handling property is determined by querying the first measured value in a database and at least one second handling property is determined by inference with a statistical model upon input of the second measured value. This makes it particularly easy to utilize the advantages of both procedures.In particular, it has been recognized that handling properties that are based solely on the design of the object are better retrieved from a database, whereas handling properties that result from the interaction of the object with the environment can be better determined using a statistical model. The way in which the database is created is arbitrary. It can be a database specifically created for the process. However, it is also conceivable that the database is generally accessible and / or intended for other processes. In this case, the database serves the general networking and digital accessibility of objects, and the process uses the stored information.

[0010] The term “optimal and / or correct handling property” refers to a handling property which a knowledgeable and skilled person skilled in the art who is familiar with the detected object would classify as optimal and / or correct in the given environment in which the method is carried out. Furthermore, it is obvious to a person skilled in the art that the measured value depends on the type of sensor used and, depending on the type of sensor used, a measured value of one embodiment can be a handling property of another embodiment of the method according to the invention. For example, in one embodiment the sensor is part of a lidar and the measured value is therefore the position of the object, which is simultaneously the object property of the recording location, while in another embodiment the sensor is a camera and the measured value is therefore an image from which the object property of the recording location is inferred using the statistical model. In step c.A control data set for the robot arm is created, taking the handling properties into account. In addition to the stored handling properties, it is also conceivable that further information outside the object is taken into account. The information can preferably relate to the robot arm, in particular a status of the robot arm, the environment in which the method is executed, in particular a status of the environment, and / or the user, in particular specifications of the user using the method. The control data set specifies the movement sequences that the robot arm should execute.

[0011] In step d, the robot arm handles the object according to the control data set. This means that the robot arm executes the movements specified by the control data set.

[0012] The method according to the invention makes it possible to automatically handle an identified or unidentified object by a robot arm in a simple, fast, and easily adaptable manner. Therefore, the corresponding work processes of the robot arm are not prevented, delayed, or interrupted when an object that cannot be identified under the given circumstances is encountered. The handling of the identified or unidentifiable object by the robot arm is carried out smoothly, reliably, safely, and quickly.

[0013] Advantageous further developments of the invention, which can be implemented individually or in combination, are presented in the subclaims.

[0014] In an advantageous development of the method, in step a. the measured object property is a property inherent in the object, a property influenced by the environment and / or a property assigned to the object. A property inherent in the object is in particular, but by no means exclusively, the volume, the contours, the dimensions, the fill level, the content, the weight, the state of aggregation, the hardness, the strength, the density, the roughness, the composition, the refractive index, the temperature and / or the color. A property influenced by the environment is in particular, but by no means exclusively, the position, the orientation, the concealment, in particular the concealment by at least one other object, the temperature and / or the type of delivery, in particular the packaging, an indication on the packaging, the sender, enclosed information and / or the type of container containing the object.A property assigned to the object is in particular, but by no means exclusively, an identification number, an identification name, a product number, a product name, an article number or an article name. The property assigned to the object can conceivably be a property assigned to the object arbitrarily, randomly or systematically. In principle, the object property can be anything, as long as the measured value representing it allows an inference to at least one handling property and / or a retrieval of at least one handling property in a database, preferably with identification of the object. The object properties listed above enable the acquisition of a measured value representing them, which is particularly suitable for being compared with a database and / or for allowing inferences to an optimal and / or correct handling property.In particular, the properties assigned to the object are particularly suitable for comparison with a database. Object properties inherent in the object are particularly suitable for comparing the measured value with a statistical model.

[0015] Advantageously, it is further conceivable for the handling characteristic in step b to be a grip strength, a point of engagement, a pick-up location, a pick-up orientation, a concealment, a deposit orientation, a transport orientation, a deposit location, a maximum acceleration, a maximum speed, and / or the minimum lifting height. In principle, the handling characteristic can be any desired characteristic, as long as it meets the inventive requirements for the most reliable, safe, and rapid handling possible. In particular, the handling characteristic should prevent damage and / or deterioration of the object during rapid handling.

[0016] Furthermore, it is advantageously conceivable that in step b, the inference is performed using the statistical model if the query for the measured value in the database fails. Within the scope of the invention, it has been recognized that handling by means of a control data set, which has been created based on at least one handling property uniquely assigned to the detected object, is more reliable, better, and safer, at least for certain handling properties known to those skilled in the art. Therefore, it is preferable that a unique assignment of the detected object with at least one handling property assigned to it is preferable to an inference based on probabilities.Therefore, it is advantageous to first query the measured value in the database in order to obtain at least one handling property associated with the object and, only if this fails, to make an inference using the statistical model to obtain at least one statistically suitable handling property.

[0017] In a preferred alternative, it is conceivable that in step b. at least one further measured value is recorded by means of at least one further sensor and the inference is carried out by means of the statistical model by entering the further measured value if the query for the measured value in the database fails. In other words, in step a. at least a first measured value of at least a first object property is recorded by means of at least a first sensor, and in step b. an attempt is made to determine at least one handling property by querying the first measured value in the database. If the first measured value is not present in the database and / or cannot be retrieved, the query fails and a step a2. is subsequently carried out which comprises measuring at least a second measured value of at least a second object property of the object by means of at least a second sensor. Subsequently, in a step b2.at least one handling property is determined by means of inference with a statistical model upon input of the second measured value. In this way, it is possible, on the one hand, to utilize the advantage offered by handling properties specifically stored for an object and, at the same time, to have a procedure for objects whose first measured value cannot be found in the database. In particular, it has been recognized that, as described elsewhere, some object properties, in particular those properties assigned to the object randomly, arbitrarily and / or systematically, are particularly well suited to querying with a database, but have little or no direct connection to a handling property, so that on their basis no statistically reliable statement can be made regarding a correct and / or optimal handling property.Therefore, the measurement of at least one second measured value, which has a statistically relevant relationship to at least one handling property and is therefore suitable as a basis for a statistically sound inference using a statistical model, is advantageous.

[0018] Furthermore, it is advantageously conceivable that in step b. the database and / or the statistical model is expanded and / or updated by the measured value measured in step a. and / or the further measured value measured in step b. Within the scope of the invention, it has been recognized that the physical characteristic and thus the measured value of an object property of an object can change temporarily or permanently. The reason for this is arbitrary; possible causes include, for example, a new design of the object or a new article number assigned to the object. Furthermore, it has been recognized that optimal and / or correct handling properties can be inferred using the statistical model, in particular when the statistical model is trained with a large number of typical and / or realistic data (training data).The measured values ​​of objects that arise during ongoing operation represent such realistic data that are particularly suitable for training the statistical model. By expanding and / or updating the database and / or the statistical model, the procedure is continuously improved and kept up to date during the ongoing process.

[0019] It is furthermore preferably conceivable for the handling to comprise grasping, picking up, lifting, transporting, sorting, feeding, and / or shaking. In principle, the ultimate handling of the object is irrelevant, as long as the handling is the optimal and / or correct handling characteristic according to the invention. The aforementioned handling steps are common handling steps according to which a robot arm handles an object. The term "feeding" refers to making the object available to a machine that further handles and / or processes the object in a form suitable for the machine.

[0020] It is furthermore advantageously conceivable for step d. to further comprise evaluating the handling and obtaining an evaluation, and optionally displaying the evaluation. This evaluation makes it possible to assess the quality of the handling. Advantageously, it is furthermore possible to expand and / or update the handling properties in the database and / or the statistical model by means of the evaluation in order to obtain the advantages described elsewhere. In this way, the handling can be continuously optimized during ongoing operation. In particular, the statistical model can be continuously trained, improved and / or adapted by the evaluation, so that for new objects that cannot be queried in the database, the optimal and / or correct handling property can be inferred with a higher probability when the method according to the invention is repeated using the static model.Updating and / or expanding the handling property in the database is also conceivable, enabling damage-free and / or error-free handling of future objects with the same measured value as the object being evaluated. It is conceivable to perform the evaluation using the existing sensors or using sensors specifically installed for evaluation.

[0021] In a further embodiment, it is conceivable that the control data record is updated and / or expanded in step d., taking the evaluation into account. This makes it possible to detect any possible incorrect and / or hazardous handling of the object during handling, and to prevent deterioration and / or damage to the object during handling by adapting the control data record.

[0022] In a particularly preferred development of the invention, it is conceivable that the object is a medical sample or a container with a medical sample, the handling property in step b. is a storage location, wherein the storage location is the position of a feed device of a medical analysis device, and the handling in step b. comprises feeding to the feed device. In principle, it is arbitrary which type of object and which type of handling is carried out with the method according to the invention. However, within the scope of the invention, it has been recognized that the method is particularly suitable for handling medical samples or containers of medical samples.This is due to the fact that medical samples, although they comprise a limited number of types of objects which differ fundamentally in their properties, may, under certain circumstances, have a range of measured values ​​for object properties that cannot be completely and directly assigned, so that a complete and comprehensive database can only be created with great effort. Furthermore, it has been recognized that different types of medical samples must be fed to different analysis devices. Based on the object properties, it can be determined which type of medical sample is present or which type of medical analysis device the medical sample should be fed to. The method according to the invention therefore enables the quick, easy, and safe handling of medical samples.It is also conceivable that the location of the object is a delivery zone for objects to be handled for the first time or the delivery facility of the medical analysis device for objects that have already been handled once or several times.

[0023] It is assumed that the definitions and descriptions of the above terms apply to all aspects described below in this description, unless otherwise stated.

[0024] According to the invention, a device for handling objects comprising a robot arm, at least one sensor and a computer system is further proposed, wherein the computer system comprises a computer-readable storage medium comprising a database and / or a statistical model for carrying out step b of a method described elsewhere and carried out by the device and a data processing device for carrying out steps b and c of the method described elsewhere. By means of the device according to the invention, the advantageous method can be carried out in addition to the advantages described there. Particularly preferably, the device comprises at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 or 20 identical or different sensors.

[0025] It is also conceivable for the sensor to be a light sensor, a force-torque sensor, or a pressure sensor, or to be part of a device, in particular a barcode scanner, an RFID scanner, an image capture device, a lidar, a radar, a tactile sensor, or a measuring device, in particular a spectrometer, an ultrasound measuring device, an X-ray device, a refractometer, a thermometer, a moisture meter, and / or a scale. The type of sensor used to measure the measured value is fundamentally irrelevant, as long as it satisfies the inventive claims and can detect a measured value of an object property suitable according to the invention. Which object property is suitable according to the invention depends to a large extent on the expected objects and thus on the intended use of the device.The aforementioned sensors and / or devices are particularly suitable for one or more specific applications. The corresponding applications and the sensors and / or devices suitable for them will be apparent to a person skilled in the art.

[0026] In a further development of the invention, it is conceivable for the robot arm to be single-axis or multi-axis. A single-axis robot has the advantage of being easier to control and monitor. A multi-axis robot, on the other hand, can reach a larger area of ​​accessible surfaces, which serve, for example, as a storage location, and / or influence the storage orientation of the object. Particularly preferably, the robot arm is guided on a rail. Furthermore, it is conceivable for the robot arm to comprise a gripping device with at least one finger and / or at least one suction cup. The gripping device facilitates the secure gripping and transport of the object.

[0027] It is further conceivable for the computer system to comprise an interface and / or a network interface. The network can be an intranet comprising at least one other computer system and / or the internet. The interface and / or the network interface make it possible to make additional user information described elsewhere available to the computer system for consideration when generating the control instructions. Likewise, entries and / or updates in the database, in particular handling properties for specific objects, can be made using the interface and / or the network interface. It is further conceivable for the database to be located partially or completely outside the computer system and for the computer system to be connected to the database via the network interface.The manner in which the entries and / or updates for the database are generated is fundamentally arbitrary. It is preferably conceivable for the entries and / or updates to be made using the measured values ​​recorded by the computer system, as described elsewhere. Even more preferably, it is conceivable for the database to be updateable and / or expandable by at least one other identical or different computer system which is also capable of recording measured data, and / or by being stored during production and / or recording of the object. In particular, this can be the database described elsewhere which serves for general networking and digital accessibility. Furthermore, it is possible to provide training data sets for training the statistical model using the interface components and / or the Internet interface.This enables safe, correct and / or optimal handling of individual objects in advance.

[0028] In a further development of the invention, it is conceivable that the objects to be handled are medical samples and / or containers containing a medical sample, and that the device comprises at least one medical analysis device. As described elsewhere, the method according to the invention is particularly well suited for handling medical samples and containers containing a medical sample. Therefore, the device is also particularly well suited for handling medical samples or containers containing a medical sample.In order to be able to make full use of the advantages of the device according to the invention, it is advantageous if the device comprises at least one medical analysis device, preferably at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 or 20 identically or differently designed analysis devices, to which the medical sample or the container with the medical sample can be fed during handling.

[0029] In a further development, it is conceivable that the medical analysis device is a polymerase chain reaction (PCR) test device, a real-time PCR test device, a sequence analysis device, a blood analysis device, in particular a blood glucose meter or a lactate meter, a blood coagulation meter, a blood alcohol meter, a refractometer, a pH meter, a urine analysis device, and / or a photomicroscope. In principle, the type of medical analysis device is irrelevant. The aforementioned analysis devices are widespread and frequently used in the medical field. The respective advantages and areas of application of the analysis devices are known to those skilled in the art.

[0030] Further details, features, and advantages of the invention will become apparent from the following description of the preferred embodiment in conjunction with the subclaims. The respective features can be implemented individually or in combination with one another. The invention is not limited to the embodiment. The embodiment is illustrated schematically in the figures.

[0031] In detail:

[0032] Fig. 1 is a schematic side view of a device according to the invention; and

[0033] Fig. 2 is a schematic plan view of the device according to the invention.

[0034] Fig. 1 shows a schematic side view of a device according to the invention. The device is used in a medical laboratory for analyzing medical samples. The device comprises a robot arm 10. In Fig. 1, the robot arm is biaxial, i.e. it has two axes 12. The robot arm also comprises a gripping device 11 with which it can grasp objects 20. On the left in the image, several differently designed objects 20, which are different medical samples in containers, are ready for handling in a package 22. A first sensor 31 is provided for their identification. In the present case, the first sensor 31 is part of an RFID scanner. The objects 20 have a corresponding RFID chip with an identification number identifying the object 20. The first sensor 31 measures a first measured value in the form of the identification number from the chip.Furthermore, the first sensor 31 is connected to a computer system 40. If the first sensor 31 measures the identification number of an RFID chip, it reports the measured identification number to the computer system 40. The computer system 40 compares the identification number with a database. The database contains first handling instructions corresponding to the identification number, namely grip strength, point of attack, placement location, placement orientation, maximum speed and maximum acceleration for creating control data sets for the robot arm 10. The device further comprises a second sensor 32, which is a camera. The second sensor 32 detects the objects 20 and transmits a second measured value in the form of an image to the computer system 40. Second handling properties, namely an occlusion, a recording location and a recording orientation of the individual objects 20, are inferred using a statistical model.Furthermore, the position of the robot arm 10 is also provided to the computer system 40 as additional information. Should the RFID chip be lost before provision, it is possible to determine initial handling instructions from the second measured value, also using the statistical model, after querying the measured value in the database logically fails. From the handling properties and the additional information, a data processing device included in the computer system 40 creates a control data record for controlling the robot arm 10 and transmits it to a control device 13 of the robot arm 10. The computer system 40 has an interface section 41 into which a user can also enter additional information relating to the handling properties. This information is taken into account by the computer system 40 when creating the control data record.The computer system 40 further comprises an internet interface 42. The internet interface 42 makes it possible to provide training data sets for the statistical model or to transmit additional information to the computer system regarding the handling. The additional information can be taken into account by the computer system 40 when creating the control data set or stored in the database as a handling property for later consideration. The additional information, which is provided via the internet interface 42 and stored in the database with the identification number, contains, in particular, the storage location, which is a feed device for an analysis device (not shown in Fig. 1).For this purpose, a user specifies the type of sample which provides information on the desired examination, for example “blood”, or directly the desired analysis device, by specifying the identification number. The container with the RFID chip and the medical sample is then provided to the device according to the invention and handled according to the method described above.

[0035] The statistical model can also be trained during operation. For this purpose, the second sensor 32 continuously records an image while the object 20 is being handled by the robot arm 10. The handling is then evaluated using the image, and an evaluation is obtained. The statistical model is further trained using the evaluation. Fig. 2 shows a plan view of the device according to the invention. The robot arm 10 is arranged between a preparation area 21 with various packages 22 and a storage area 23. The packages 22 contain objects (not shown in Fig. 2) which are detected by the first sensor 31. The storage area 23 contains three feed devices 24 for various medical analysis devices, which are not shown in Fig. 2. In the present case, these feed devices are carriers for medical sample containers.The robot arm 10 handles the various objects according to the control data set by removing them from the packaging 22 and feeding them to the designated feed devices 24. Furthermore, the storage area 23 has a sorting device 25 into which the robot arm 10 sorts damaged sample containers according to the control data set.

Claims

Claims: A method for handling an object (20) using a robot arm (10) comprising the following steps: a. Detecting at least one object (20) and measuring at least one measured value of at least one object property of the object (20) using at least one sensor (31, 32); b. Determining at least one handling property by querying the measured value in a database and / or by inference with a statistical model using the measured value as input; c. Creating a control data set for the robot arm (10) taking the handling property into account; and d. Handling the object (20) using the robot arm (10) according to the control data set. Method according to claim 1, wherein the object property in step a. is a property inherent in object (20), a property influenced by the environment and / or a property assigned to object (20).A method according to claim 1 or 2, wherein the handling characteristic in step b is a grip strength, a point of attack, a recording location, a recording orientation, a concealment, a storage orientation, a transport orientation, a storage location, a maximum acceleration, a maximum speed, and / or the minimum lifting height. A method according to any one of the preceding claims, wherein in step b, inference is made using the statistical model if the query of the measured value in the database fails. A method according to any one of claims 1 to 3, wherein in step b. at least one further measured value is acquired by means of at least one further sensor (32) and the inference is made using the statistical model with input of the further measured value if the query of the measured value in the database fails. A method according to any one of the preceding claims, wherein in step b. the database and / or the statistical model is extended and / or updated with the measured value measured in step a. and / or the further measured value measured in step b. A method according to any one of the preceding claims, wherein handling comprises grasping, picking up, lifting, transporting, sorting, feeding and / or shaking. A method according to any one of the preceding claims, wherein step d. further comprises: Evaluating the handling and obtaining an evaluation. Method according to claim 8, wherein step d. further comprises: Updating and / or extending the control data set taking the evaluation into account. Method according to one of the preceding claims, wherein the object (20) is a medical sample or a container with a medical sample, the handling feature in step b is a storage location, wherein the storage location is the position of a feeding device (24) to at least one medical analysis device. The device is characterized in that it is characterized in that the handling in step d. includes the feeding to the feeding device (24).

1. Device for handling objects (20) comprising a robot arm (10), at least one sensor (31, 32) and a computer system (40), characterized in that the device uses the method according to one of claims 1 to 10, and that the computer system (40) comprises a computer-readable storage medium comprising the database and / or the statistical model for executing step b. and a data processing device for executing steps b. and c.Device according to claim 1, characterized in that the sensor (3, 32) is a light sensor, a force-torque sensor, or a pressure sensor, or is part of a device, in particular a barcode scanner, an RFID scanner, an image acquisition device, a lidar, a radar, a tactile sensor, or a measuring device, in particular a spectrometer, an ultrasonic measuring device, an X-ray device, a refractometer, a thermometer, a humidity measuring device, and / or a scale. Device according to claim 1 or 12, characterized in that the robot arm (10) is uniaxial or multiaxial. Device according to any one of claims 1 to 13, characterized in that the computer system (40) comprises an interface component (41) and / or a network interface (42). Device according to any one of claims 1 to 14, characterized in that the objects (20) to be handled are medical samples and / or containers with a medical sample and the device comprises at least one medical analyzer. Device according to claim 15, characterized in that the medical analyzer is a polymerase chain reaction (PCR) test device, a real-time PCR test device, a sequence analyzer, a blood analyzer, in particular a blood glucose meter or a lactate meter, a blood coagulation meter, a blood alcohol meter, a refractometer, a pH meter, a urinalysis device and / or a photomicroscope.

Images