DE102017126457A1 - Device for supporting a user in a workplace, method and computer program product - Google Patents

Abstract

A device (1) for supporting a user (2) at a workstation (3) is described. The device (1) has a data processing system (4) associated with the workstation (3), at least one output unit (10) present at the workstation (3) and connected to the data processing system (4), an input unit (9 and a plurality of at the workstation (3) existing work equipment (11) managed by the data processing equipment and usable to carry out the work carried out at the work station (3) The apparatus (1) is characterized by the fact that the data processing equipment (4) is designed to handle operations on the workstation (3) Workstation (3) involving the implements (3) to be detected by means of the implements (11) and / or the input unit (9), to recognize a situation on the basis of the detected events by means of a neuromorphic processing component (42), and to recognize the situation On the basis of the detected situation, a message is output to the user (2) Furthermore, a method for operating the device is described.

Description

• - einer dem Arbeitsplatz zugeordneten Datenverarbeitungsanlage,
• - wenigstens einer am Arbeitsplatz vorhandenen und mit der Datenverarbeitungsanlage verbundenen Ausgabeeinheit zur Widergabe von Information,
• - wenigstens einer am Arbeitsplatz vorhandenen und mit der Datenverarbeitungsanlage verbundenen Eingabeeinheit, die allein dem Zweck dient, Daten an die Datenverarbeitungsanlage zu übermitteln,
• - einer Vielzahl von am Arbeitsplatz vorhandenen und von der Datenverarbeitungsanlage verwalteten Arbeitsgeräten, die zur Ausführung der am Arbeitsplatz ausgeführten Tätigkeiten verwendbar sind.

The invention relates to a device for assisting a user in a workplace with:

• - a data processing system associated with the workstation,
• at least one output unit present at the workstation and connected to the data processing system for the purpose of displaying information,
• - at least one input unit located at the workstation and connected to the data processing system, the sole purpose of which is to transmit data to the data processing system,
• - a variety of work equipment available at the workplace and managed by the data processing equipment, which can be used to carry out activities carried out at the workplace.

The invention further relates to a method for operating the device and a computer program product.

From the CN 206058263 U A laboratory system is known in which data is collected at various laboratory workstations and passed through a network connection to a main terminal where the data is evaluated. The well-known laboratory system enables the central planning and evaluation of laboratory examinations from the main terminal and therefore enables the efficient use of laboratory resources.

When carrying out laboratory work or other activities of a scientific nature, the situation often arises that the user needs assistance, either because the processes are complex and therefore difficult to master, or because the user's current expertise is not sufficient to support them to be able to handle upcoming tasks. This situation will occur more and more with the advancement of technology.

Based on this prior art, the invention is therefore an object of the invention to provide an improved device for supporting users at workstations.

This object is achieved by a device, a method and a computer program product having the features of the independent claims. In dependent claims advantageous embodiments and developments are given.

In the apparatus, the data processing system is adapted to detect operations in the workplace that involve the implements by means of the implements and / or the input unit. These operations may be either actions performed by the user in the workplace or events related to the applications of the implements. The data processing system is further configured to recognize a situation by means of a neuromorphic processing component and to output a message to the user on the basis of the detected situation. The device is thus able to assist the user depending on the situation in an effective manner.

In one embodiment, the data processing system has at least one processing layer, each having at least one assembly in which a processing unit is formed with at least one neuromorphic processing component. The structuring into different processing layers enables the different processing layers to be trained separately for the respective tasks. In addition, the processing speed in the various processing layers may vary according to the tasks to be performed and be adapted to the respective tasks.

In another embodiment, the processing unit equipped with at least one neuromorphic processing component additionally includes at least one algorithmic processing component that interacts with the neuromorphic processing component. In particular, the algorithmic processing component may take on routine tasks that run according to fixed rules or take the place of the neuromorphic processing component as long as the neuromorphic processing component is not yet fully trained or if it fails.

To train the neuromorphic processing component, the assembly may be switched from an operating mode to a learning mode. In the learn mode, a compositor may apply training information to the neuromorphic processing component and an evaluator may generate a score in response to the response generated by the neuromorphic processing component and thereby apply the neuromorphic processing component. Based on this feedback, the neuromorphic processing component can be effectively trained.

The provided in each processing layer least one assembly can via a Communication channel Receive situation information from a subordinate module and send it to a higher-level module and receive instruction information from the higher-level module via another communication channel and send it to the subordinate module. In this way, the flow of information from the devices back through the data processing system and back to the devices.

The module can furthermore have a situation memory which is located upstream of the processing unit, in which situation information from the subordinate module can be stored, and / or have an instruction memory upstream of the processing unit, in which instruction information from the superordinate module can be stored, so that the situation information and instruction information of the processing unit persist available.

In this way, the compositor can generate the training information based on information taken from the situation store and / or the instruction store. In addition, the compositor can also generate the training information himself. The evaluator may then generate the score in response to the situation information and / or instruction information generated by the neuromorphic processing component and apply the score to the neuromorphic processing component.

In each assembly, the processing unit may be connected to a database in which processing rules for processing the situation information and / or the instruction information are stored. The processing rules stored in the database can be used to adapt and control the behavior of the module.

The processing unit will typically perform the processing of the situation information in accordance with the instruction information so as to allow the parent processing layer to influence the processing of the situation information and derive instructions for the subordinate processing layer.

In one embodiment, the data processing system has a device layer underlying the at least one processing layer that is configured to communicate with the work devices, the processing layer overlying the device layer automatically detecting newly added work devices based on a device catalog. Since the user does not have to act on his own to integrate the tools in the device, this represents a considerable work for the user.

In particular, the input unit can comprise a camera and the data processing system can be set up to generate data for recognizing the situation from the images supplied by the camera by means of image recognition. For example, in this embodiment, the user may also control the device by gestures.

Further, the camera, along with a subordinate image recognition, can also serve to detect newly added implements. The image recognition can be carried out in a processing layer for signal processing above the device layer, which is furthermore also set up to read out the data required for putting the newly added implement into operation from the device catalog maintained in a database on the basis of the detected situation.

In a further embodiment, the device may comprise, above the signal processing processing layer, a further processing layer for reactions and a processing layer for inferences, wherein the processing layer for reactions may produce faster control commands for the implements and / or the output devices than the inference processing layer are as with the processing device for conclusions. This ensures that the device can respond quickly to the user's actions and other workplace events when needed, while being able to draw complex conclusions from the situation information, even if more time is needed. In addition, it is possible to perform the training of the neuromorphic processing component separately for simple but fast reactions and complex but slow conclusions, which also facilitates troubleshooting.

The device may have an uppermost processing layer, to which several workstations are connected. This embodiment allows the coordination of different workplaces. In particular, the devices associated with the various workstations can be centrally monitored and controlled so that the messages to the user also take into account the workstation overlapping situation.

The output of the messages to the user may be audible or through the presentation of information on a visual display. In addition, the output unit of the device may also include a projector with which to project information for the user on the workstation, so that the user understands possible hints of the device easier and faster.

The device is particularly suitable for jobs that serve the material processing and / or material examination.

The advantages of various embodiments of the device listed here apply correspondingly to a method for operating the device and to a computer program product implementing the method.

• 1 eine Vorrichtung für einen Arbeitsplatz zur Materialbearbeitung und/oder zur Materialuntersuchung oder allgemein zur Durchführung von technischen, chemischen, physikalischen oder medizintechnischen Untersuchungen;
• 2 eine Darstellung der prinzipiellen Architektur der zugehörigen Datenverarbeitung;
• 3 einen Darstellung des Aufbaus einer Baugruppe der Architektur aus 2;
• 4 eine Darstellung der wechselnden Anteile von algorithmischer Logik und künstlicher Intelligenz in den Verarbeitungsschichten, in Abhängigkeit von der Abstraktionsebene bzw. Schicht;
• 5 eine Darstellung der Architektur aus 2 für eine Ausführungsform mit vier Verarbeitungsebenen;
• 6 eine Darstellung von möglichen Daten- und Informationsflüssen in der Ausführungsform aus 5.

Further advantages and features of the invention will become apparent from the following description, are explained in the embodiments of the invention with reference to the drawings in detail. Show it:

• 1 a device for a workstation for material processing and / or material examination or generally for carrying out technical, chemical, physical or medical technical examinations;
• 2 a representation of the basic architecture of the associated data processing;
• 3 a representation of the structure of an assembly of the architecture 2 ;
• 4 a representation of the varying proportions of algorithmic logic and artificial intelligence in the processing layers, depending on the level of abstraction or layer;
• 5 a representation of the architecture 2 for an embodiment with four processing levels;
• 6 an illustration of possible data and information flows in the embodiment 5 ,

1 shows a device 1 for the support of one or more human users 2 at a workplace 3 for material processing and / or material testing or generally for the performance of technical, chemical or physical investigations. The workplace 3 may be in a room where these examinations are carried out. It can be a laboratory, a technical center or an examination room. At this workplace 3 In particular, investigations in the field of natural sciences, medicine and technology should be carried out. There are usually various experiments, process controls, quality controls, tests and measurements performed. In the process, materials and products are examined and optional products are produced in small quantities.

Examples of this type of job 3 They can be: scientific laboratories such as chemical and pharmaceutical research facilities, technology as a link between laboratory and large-scale production in the chemical and pharmaceutical industry, laboratories for quality assurance in industry, examination and treatment rooms in medical practices or shops with scientific measuring points such as the offices of opticians or hearing care professionals , as well as microbiological and physical testing laboratories. All these jobs 3 can each be implemented stationary or movable.

Two features are for these types of jobs 3 typical:

Firstly, the activities carried out there have a scientific character, since measuring equipment must be operated and work is done in a defined environment. Typical of this type of work 3 is that the vast majority of all processes, for example more than 95%, are well-defined, that is to say, are or are documented to a large extent and have a low distribution of the processes themselves and the results. The processes may be a sequence of macro-process steps, which in turn are composed of micro-process steps. Examples of a macro process step are "Get Material" or "Titrate pH". Examples of micro process steps are "turn on device" or "move worker X from place A to place B".

Second, these activities have a manual share of the user 2 to carry out these well-defined activities.

Unlike this kind of jobs 3 are fully automatic systems, as they are used in large-scale production, or jobs with not exactly defined steps such as in a bakery or a psychotherapeutic practice.

The interaction of the device 1 with the user (s) 2 is so to the human user 2 adapted that device 1 from the users 2 not perceived as a machine. Cognitive immersion is said to be an essential feature of the device 1 his. This feature of the device 1 is achieved by the fact that the device 1 a data processing system 4 showing features of so-called artificial intelligence.

In terms of physical structure, this data processing system includes 4 Hardware components used to execute algorithmic logic or of artificial intelligence or a combination of both. These hardware components are using a network 5 interconnected. The network 5 may include a direct connection or a local or global data network, which may be wired or wireless as needed. Hardware components can be one or more computers 6 which may be, for example, a server and / or workstation, or a combination thereof, interconnected via the network 5 are connected. In addition, one or more workstation devices can also be used 7 the device 1 Include hardware components suitable for performing algorithmic logic or artificial intelligence. These components are then also part of the data processing system 4 consider. Other special hardware optimized for the implementation of artificial intelligence can also be part of the data processing system 4 his. On each of these hardware components can be logical components of the data processing system 4 come into play, which include algorithmic logic or artificial intelligence or a combination of both.

The data processing system 4 does not have to be a single computer 6 but can be on multiple machines 6 be distributed with one or more processors or special hardware. It is possible that the data processing system 4 Also includes computer capacities that are in one of the physical workstation devices 7 are set up. These workstation devices 7 can according to the concepts of the IoT (= Internet of Things) or the concepts of the Internet with the calculator 6 be connected.

In 1 are different examples of workplace devices 7 on a work surface 8th shown. In addition to input devices 9 and output devices 10 include the workplace devices 7 a set of work tools 11 with which for each job 3 typical process steps are performed.

For immediate communication with the user 2 can the data processing system 4 as input devices 9 one or more keyboards 12 with or without mouse and / or as acoustic input devices 9 one or more microphones 13 include. Furthermore, the user can 2 via a gesture recognition or via the evaluation of a camera 14 delivered nonverbally communicate with the system. Also, in non-verbal form, by means of an electroencephalography (EEG) measured and cognitively interpreted brain activity, a control of the device 1 respectively. As audiovisual output devices 10 can serve monitors 15 , Projectors 16 or speakers 17 , About these output devices 10 can recommend action to the user 2 be issued, for example, by the projectors 16 the on the job 3 existing and currently used tools 11 mark with a ray of light. For transmission of information from the device 1 to the user 2 can also be at the user 2 personal devices such as one in 1 glasses, not shown, with projection surface for augmented reality. Furthermore, for the transmission of information from the device 1 to the user 2 and by the user 2 to the device 1 also at the user 2 personal devices such as a smartphone or tactile feedback device such as smartwatches, fitness bracelets, or the like. In that respect, the input devices 9 and the output devices 10 also form a physical unit. Such input devices 9 and output devices 10 are typically via a wired or wireless local data connection to the data processing system 4 involved.

Are in a device 1 several users 2 Actively, the device can 1 these users 2 and the respective personal data environment, for example recipe prescriptions or records of the data, to the respective user 2 and assign its working context. For example, the user goes 2 B to one before by the user 2A used high-performance liquid chromatography unit, there are the corresponding default settings of user when 2B 2 B loaded.

Another example is the assignment of measurement data to the respective customer for an optician. In this case, the customer is the user 2 who passes through various measuring stations and has his virtual patient record in each context at each measuring station - until the spectacles have been fitted, where all the test results are displayed to the optician in a final consultation on a tablet computer. The same applies to the application of the device 1 in connection with the examination of patients in radiology, dentistry, or other medical fields.

In addition to the previously mentioned workstation devices 7 includes the device 1 other tools 11 that over the net 5 with the data processing system 4 can be connected. Some tools 11 can also without direct data connection to the data processing system 4 into the device 1 be involved, such as a conventional analog heater 18 , For this purpose, the device 1 that have one or more with the data processing system 4 connected cameras 14 has, by one of the Data processing system 4 executed image recognition such, with the data processing system 4 Data-unconnected implements 11 and capture and recognize their use.

The tools 11 may be different devices with or without sensors: for example, a conventional analog heater 18 its setting and use by means of the camera 14 visually recorded and evaluated cognitively, a modern thermometer 19 with up-to-date hardware to execute processing logic, either algorithmic or artificial intelligence, across the network 5 in the rest of the data processing system 4 is involved, one over the net 5 with the data processing system 4 connected power supply 20 , one over the net 5 with the data processing system 4 connected scales 21 with optional camera for goods recognition, for example by barcode, or via the network 5 with the data processing system 4 associated deduction 22 ,

Further for this type of jobs 3 typical tools 11 For example, glassware used for chemical experiments may include test tubes, an optician's instruments such as an optician's measuring glasses, various hand tools such as tweezers or spatulas, a dental prosthetic kit such as probes or mirrors, as well as classic tools. Every use of these manual tools and devices is made by means of the various cameras 14 recorded and evaluated cognitively, for example, according to movement, use and whereabouts. If necessary, the tools can 11 be provided with an optical or readable via a radio frequency identifier, the identification and thus the evaluation of the use of the tools 11 facilitated.

The connection of these workplace devices 7 can also, as already mentioned, according to the concepts of the IoT (= Internet of Things) or the concepts of the Internet. The devices can be connected with the help of converters, whereby a fieldbus is used and / or protocols such as KNX are used. No specifications are made for the topology of the fieldbus or for the protocol of the fieldbus (CPFx).

The exact number and positioning of the workstation devices 7 , their measurement and observation data from the data processing system 4 should be cognitively evaluated and optimized in such a way that the required information space is covered to a sufficient extent and as homogeneously as possible, ie without excessive ups and downs of information density. For example, the cameras required for situation detection are 14 to be attached to all the relevant jobs in the workplace 3 cover.

The data processing system 4 has an architecture whose basic structure is in 2 is shown. It generally includes a device layer 30 which the work tools 11 of the workplace 3 as data sources as well as the input devices 9 and output devices 10 for immediate interaction with the user 2 includes. In those workstation devices 7 from the device layer 30 , which comprise hardware components suitable for executing algorithmic logic or artificial intelligence, can each directly have an assembly MO.j (1≤j≤m ₀ , m ₀ = number of assemblies in the device layer 30 ), which are described below 3 will be described in more detail. In 2 this is for example with device G1 and G2 the case. The devices G1 and G2 each contain the modules M0.1 and M0.2 ; device Gm ₀ in 3 on the other hand, does not include such an assembly M0.m ₀ ,

Furthermore, the architecture comprises one or more processing layers 31 , in the 2 as a processing layer M1 until Mn are designated. The processing layers M1 to Mn comprise at least one assembly 32 , each with Mi.j (1≤i≤n, 1≤j≤m _i , m _i = number of assemblies 32 in the processing layer Mi) are designated. The internal structure of the assemblies Mi.j is described below with reference to 3 be explained.

Each processing layer Mi processes the data supplied from the underlying processing layer M (i-1), interprets them and synthesizes them into larger meaningful contexts. Any higher processing layer Wed. the processing layers M1 up to Mn can therefore possibly also with multiple instances of the modules 32 an adjacent lower processing layer M (i-1). For example, the processing layer bundles M1 communication with several devices, each with or without an assembly MO.j , from the device layer 30 , Analogously, the processing layer Wed. bundle the communication with a plurality of packages M (i-1) .j in the processing layer M (i-1).

Are in a specific system more of these processing layers M1 to Mn, these address different levels of abstraction of data processing and control. The processing layer M1 directly on the device layer 30 is the most concrete processing layer 31 , The processing layers based thereon M1 to Mn become increasingly abstract. With increasing distance to the device layer 30 thus increases the degree of abstraction in the Processing layer Mi processed or generated information.

Two adjacent processing layers Mi and M (i-1) and the lowest processing layer M1 and the device layer 30 are via communication channels 33 and 34 connected with each other. The white block arrows in 2 illustrate the communication channels 33 for measurement data or calculation results showing the lower processing layer 31 or the device layer 30 as documentation of a situation measured or interpreted by her to the next higher processing level 31 returns as inputs to their calculations. The higher processing layer 31 In turn, it provides computational results from the adjacent underlying processing layer 31 or device layer 30 also available as input, as instructions in the sense of control information. The communication channels 34 for instructions are in 2 illustrated by black block arrows.

Each assembly 32 in the device layer 30 and the processing layers 31 has the in 3 shown internal structure.

Each assembly 32 receives input data via the two communication channels 33 and 34 : a communication channel 33 for the measurement data or the interpreted situation from below from the next lower processing layer 31 or the device layer 30 , as well as a communication channel 34 for the instructions from the top of the next higher processing layer 31 , Through these instructions flow calculated results of a higher processing layer 31 in the calculation of results and thus decision-making of the immediately underlying processing layer 31 or the device layer immediately below it 30 on. The device layer 30 and the top processing layer 31 form special cases because they only have either an upper or a lower adjacent processing layer 31 feature. For i = 0 the module gets MO.j the input data from below directly from the meter Gj, on its hardware MO.j is performed. For i = n, and thus the topmost processing layer Mn, there is no further, overlying processing layer 31 , Accordingly, an empty stream of instructions enters from above.

The incoming data will be within the assembly 32 saved. The measurement data received from below from the processing layer M (i-1) or the interpreted situation becomes in a situation memory 35 persists. Accordingly, the instructions incoming from the top of the processing layer M (i + 1) are stored in an instruction memory 36 persists. The situation memory 35 and the instruction memory 36 can be either symbolic or subsymbolic nature. The situation memory 35 and the instruction memory 36 may also have separate memory areas for memory contents of symbolic and subsymbolic nature.

The third persistent store is the assembly 32 a knowledge database 37 , This is initially from outside the module 32 filled with explicit knowledge, for example in the form of rules governing basic physical laws.

Many application contexts require all processed data, calculations and changes in the data processing system 4 fully document and store to the operations of the data processing system 4 to make comprehensible. Therefore, the assembly includes 32 as the fourth persistent storage an auditable, permanent logbook 38 , This will be within the assembly 32 just described, but not read. It may, however, be from outside the assembly 32 be read out. In this logbook 38 logs the corresponding module 32 which are themselves part of device 1 is, automatically everything, for example, all inputs, calculations, outputs, feedback to the knowledge base 37 the assembly 32 , or changes of weights in a neural network of the assembly 32 , From the resulting data set potentially different results can be derived. As a technical representation of these audit-proof and unchangeable records are many technical formats and scenarios with private and public keys. Because of within the assemblies 32 Depending on the requirements of the field of application as well as with regard to the advancement of technology, the amount of data, possibly very large amounts of data, must be determined in the logbook 38 the assembly 32 should be saved.

Core of the assembly 32 is the processing unit 40 , This is realized by a combination of an algorithmic processing component 41 with algorithmic logic and another neuromorphic processing component 42 with artificial intelligence. The proportions of the two processing components change depending on the level of abstraction of the associated processing layer 31 , While at the processing unit 40 on the lower, ie the more concrete processing layers 31 the components of algorithmic logic predominate, are in the higher, ie the more abstract processing layers 31 , as in 3 illustrates the proportions of artificial intelligence in the foreground. Especially in the neuromorphic processing components 42 the processing unit 40 in which concepts and Artificial intelligence techniques are realized, and which include, for example, neural networks or a combination of different topologies of neural networks, in the weights and parameters of these neural networks, implicit knowledge of the processing unit 40 which influences the concrete course of the processing mechanism and thus also the results.

The algorithmic processing component 41 and the neuromorphic processing component 42 can be done in different ways in the processing unit 40 interact. In particular, they can be connected in any order and mixing in series, or work in parallel.

An embodiment for the in-series switching of the algorithmic processing component 41 and the neuromorphic processing component 42 put the thermometer 19 out 1 in which an algorithmic processing component 41 Detects if the measured value exceeds a certain threshold, in this case the measurement data of the thermometer 19 at its neuromorphic processing component 42 forwards, so that this recognizes the current situation from the course of the previous measurement data.

An embodiment for the in-series switching of the neuromorphic processing component 42 and the algorithmic processing component 41 , ie in reverse order to the previous embodiment, is on a processing layer Mi above the device layer 30 the analysis of measured data of different tools 11 through a neural network in the neuromorphic processing component 42 , which on this basis identifies and classifies the situation in which the subsection of device just analyzed 1 currently located. This classification result, ie the information about the detected situation, then becomes the algorithmic processing component 41 which then handles the predefined procedure for handling such situations.

An exemplary embodiment of an initially concurrent use of a first algorithmic processing component 41 and another neuromorphic processing component 42 with a subsequent, in-line, second algorithmic processing component 41 is a security mechanism in which a first algorithmic processing component 41 and the neuromorphic processing component 42 evaluate measurement and device data individually and on this basis individually classify the current situation. Based on this, a downstream second algorithmic processing component compares 41 whether the results obtained are consistent and how, if appropriate, further action is taken.

The embodiments described herein are to be considered as exemplary and have been used to better understand the interaction of the algorithmic processing component 41 and the neuromorphic processing component 42 selected with artificial intelligence. In principle, the algorithmic processing component 41 and the neuromorphic processing component 42 be combined in any way.

The processing unit 40 Processes inputs from three different sources: the input data stream E1 from the situation memory 35 , the input data stream E2 from the instruction memory 36 and the input data stream E3 from the knowledge base 37 ,

The processing unit 40 also generates four different types of outputs: in the communication channel 33 the output data stream A1 as the interpreted situation which is provided to the next higher processing layer M (i + 1) as input (or which remains idle if i = n and thus no higher processing layer 31 gives more); in the communication channel 34 an output data stream A2 with instructions for the next lower processing layer M (i-1) (or for the specific devices G1 to Gm ₀ or Gf, if i = 0); an output data stream A3 to return explicit knowledge to the knowledge base 37 ; and an output data stream A4 for the documentation of all inputs, processing steps, outputs and weighting changes that are in the logbook 38 is persisted.

For teaching the processing unit 40 includes the assembly 32 two other components, namely a compositor 43 and an evaluator 44 In the context of training the processing unit 40 can be used. The associated data paths are in 3 indicated by dashed and dotted lines.

The compositor 43 On the one hand information Sr from the situation memory 35 remove, reassemble and put back in the situation memory 35 play back as training data. On the other hand, the compositor 43 also in the sense of a simulator generate new information Sg and in the situation memory 35 import. Both apply mutatis mutandis to information Ir and Ig from the instruction memory 36 , From the compositor 43 in the situation memory 35 and the instruction memory 36 provided training data E1T and E2T are from the processing unit 40 processed in training mode and the two resulting from it Reply A1T and A2T to the evaluator 44 transmitted. The reviewer 44 The evaluation result generated is finally sent to the neuromorphic processing component 42 For example, to a neural network to be trained in the neuromorphic processing component 42 It replicates, on the one hand, the weights of the individual connections in this neural network and, on the other hand, the topology of the neural network by adding or removing signal paths (corresponding to the dendrites in a human brain) in the neural network.

The rating of the neuromorphic processing component 42 Results generated from the training data are based on goals and rating guidelines defined by rules. These determine which criteria of the training result should be rated as "good" - for example, minimizing costs or maximizing safety. The evaluation corresponds to an ontological-semantic mapping to this rule catalog. Includes a data processing system 4 several according to the blueprint in 3 implemented modules, so have the evaluators 44 in the individual modules each have their own, module-specific rules. The respective rules can also be found in the database 37 be filed.

The compositor 43 and the evaluator 44 are in the assembly 32 usually only active when the processing unit 40 the assembly 32 is trained. In productive operation of the processing unit 40 On the other hand, they are generally not active.

The device 1 knows the four states: "Ready to work", "Learn", "Maintenance" and "Off". The process of learning can be done throughout the device 1 take place only in the state "learning", ie in particular not when the device 1 in the state "ready to work" is. However, a second device 1 used to be for a first device 1 to take over the learning process, wherein the learning process on a copy of the respective situation memory 35 and instruction memory 36 and knowledge databases 37 the first device 1 takes place, so that only in a short phase of loading (state "maintenance") the first device 1 not available.

5 shows a concrete embodiment of the architecture of the device 1 with four processing layers 31 as well as the device layer 30 : the processing layer M1 for the signal processing, the processing layer M2 for reactions, the processing layer M3 for conclusions and the processing layer M4 for organization. This bundles the processing layer M4 if necessary, communication with several workstations and the processing of data from several other workplaces 3 , in the 2 are labeled "Workplace 2" to "Workplace k" and are also constructed on the same principle. These jobs 3 do not have to be the same. To the processing layer M4 So different jobs can be organized 3 be connected. The processing layer M4 for organization is to be dimensioned so that it is in the diversity of the connected jobs 3 justified complexity. In particular, those in the knowledge base 37 filed rules all types of connected jobs 3 account. Also the size and topology of the neuromorphic processing component 42 in the processing unit 40 the processing layer M4 for organization must match the size and complexity of the entire device 1 be chosen so that the processing layer M4 not undersized.

The device 1 with the data processing system 4 is synonymous only with the device layer 30 and the processing layer M1 in a simple way per workplace 3 run. The addition of additional processing layers M2 . M3 . M4 and optionally other processing layers 31 extends the performance of the device 1 , There are also other processing layers 31 over the processing layer M4 also conceivable, analogous to the basic architecture in 2 , The division into the processing layers exemplified here M1 to M4 leans on the cognitive experiential space of the human being and can be changed in its cutting-edge with recent findings.

For this exemplary embodiment of the data processing system 4 with four processing layers M1 to M4 includes the data processing system 4 next to the device layer 30 for these four processing layers 31 four assemblies 32 : an assembly 51 for signal processing in the processing layer M1 , an assembly 52 for reactions in the processing layer M2 , an assembly 53 for conclusions in the processing layer M3 and an assembly 54 for organization in the processing layer M4 , The assemblies 51 to 54 each have the in 3 shown construction.

The white block arrows represent that of the device layer 30 to the module 54 ascending information flow of measured or interpreted situations. They are with S1 to S5. The black block arrows represent the backflow of instructions that act as feedback from the assembly 54 descending down to the device layer 30 flow. You are with I5 to I1 designated. Because the assembly 54 the topmost processing layer M4 forms and no further module over it are the over S5 ongoing information flows are not further processed. For the same reason will be over I5 no instructions to the processing layer M4 because there is no parent processing layer 31 to the processing layer M4 gives.

Each of the four processing layers M1 to M4 as well as each of the assemblies MO.j in the device layer 30 has in their processing unit 40 Proportions based on machine learning or subsymbolic processing with the respective neuromorphic processing component 42 based. For the modules MO.j these are implemented directly within the respective device Gj. Each assembly 51 to 54 can this processing unit 40 implement locally or over the network 5 Services of the computer 6 take advantage of. The proportion of abstraction increases from the device layer 30 and the assemblies MO.j over the assembly 51 to the module 54 with the transition from symbolic to subsymbolic processing. The top two assemblies 53 and 54 are completely subsymbolic. According to this structure, in each of the assemblies MO.j and 51 to 54, a neuromorphic processing component 42 , especially in the form of a neural network.

The device layer 30 and the assemblies 51 to 54 perform the following tasks:

The device layer 30 serves the connection of the tools 11 as well as the input devices 9 for direct user interaction, such as the keyboard 12 , the camera 14 , or the microphone 13 , as well as the intended for issue output devices 10 such as the speaker 17 , the monitor 15 or the projector 16 , In the device layer 30 a connection to the conventional information technology can be present, for example via a field bus, and thus the connection to the technology of a conventional workstation be accomplished.

In the device layer 30 are over an unspecified number of input devices 9 , Output devices 10 as well as work tools 11 that provide sensors and actuators, the interfaces to the workplace 3 and its users 2 accomplished. These workstation devices 7 can have local storage, written to and read from the data. Furthermore, these workstation devices 7 as previously stated, includes hardware components suitable for executing algorithmic logic or artificial intelligence, or a combination of both.

The workplace devices 7 the device layer 30 automatically log on to the module 51 the processing layer M1 for signal processing forms. The knowledge database 37 the assembly 51 contains information about the attachable workstation devices 7 and their functionalities, in the sense of a service catalog.

The assembly 51 containing the processing layer M1 realized for signal processing, serves as a cognitive recording and playback layer in terms of audiovisual signal processing such as seeing and hearing with respect to the observation of the processes occurring in the workplace 3 take place, whereby automatic recognition services are provided, and not only via the visual and auditory input devices 9 and output devices 10 , Also the tools provided with sensors or actuators 11 are considered possibilities of sensory perception and communication with the user 2 used. In addition to the usual acquisition of measured values, which are also part of the device 1 is, cognitive evaluations are made about it. For example, for the thermometer 19 Results such as "temperature curve rises very fast" or "nothing happens for a long time" are generated. Through these in the neuromorphic processing component 42 carried out cognitive evaluation of the measured data of the thermometer 19 becomes the flow of information from the thermometer 19 to the data processing system 4 to a cognitive channel, analogous to the cognitive channel of visual perception. Information that needs to be merged to recognize a cognitive context is bundled into an associated cognitive channel.

The conversion of the measurement data and observation data from the device layer 30 in cognitive channels, with a pooling, separation and renaming of the communication with the devices G1 to Gm ₀ data used represents an abstraction in the processing layer M1 This facilitates processing in the overlying processing layer M2 and makes the device 1 regardless of changes in the device layer 30 , The information of the cognitive channels are transmitted through the flow of information S2 to the overlying module 52 the processing layer M2 passed on and there in the situation memory 35 the assembly 52 persists.

An important function of the assembly 51 is the self-recognition of the devices in the device layer 30 , in particular of the working tools 11 , That means: Whatever the device 1 newly added, automatically becomes part of the device 1 and does not have to be installed or configured. For this purpose, in the knowledge base 37 the assembly 51 a service catalog led, in all devices and device classes with their characteristics and possibilities are registered. The camera 14 with downstream image recognition, the detection of new in the device 1 introduced workplace devices 7 support.

The information flows from the recognized workstation devices 7 flow as inputs S1 .j in the assembly 51 the processing layer M1 for signal processing. The in the assembly 51 calculated instructions for the underlying device layer 30 flow as a technical instruction on the channels I1 .j in the corresponding devices Gj back.

The assembly 52 for reactions that the processing layer M2 realized for reactions, serves the predominantly rule-based and reflexive control with given process patterns and reaction patterns. In addition, this module offers 52 The ability to make simple inferences, complex dialogues, and mapping simple semantic context through machine learning. About the assembly 52 also takes place communication with other systems of information technology. In that respect, in the assembly 52 Interfaces in the classical information technology such as in the ERP (= Enterprise Resource Planning), the LIMS (= Laboratory Management Information System) or other conventional systems of information technology to be established. The information from these systems flows into the knowledge base 37 the assembly 52 on.

In the assembly 52 For reactions, an evaluation is carried out in the module 51 Information collected for signal processing. The vast majority of this control is algorithmic-static and is done using the associated algorithmic processing component 41 , Further, in the assembly 52 implemented so-called capabilities. These follow process descriptions for specific action contexts. These capabilities are mostly implemented on the basis of rules in fixed algorithms. These rules are in the knowledge base 37 the assembly 52 stored. However, shares of skills can also be found in the assembly 52 based on machine learning and the neuromorphic processing component 42 the assembly 52 use. In the assembly 52 are the primary security policies of the workplace 3 deposited, not by overlying processing layers 31 should be overridden.

About all objects in the workplace 3 is in the assembly 52 kept a list, which also includes the history of the objects in the workplace 3 documented. The history of an object indicates the sequence of activities associated with this object. The activities depict the relationships between the objects in the form of actions and dependencies between the objects. The activities can have a multi-level hierarchy: activity groups, macro activities, and micro activities. For example, a microactivity is a single handle in the parent macro activity "Weighing". The objects, together with the associated activities, form the action contexts that form the basis of the assembly 52 form implemented skills. Depending on the action contexts, the abilities generate responses that take the form of concrete cognitive information and action commands over the instruction channel I2 to the subassembly underneath 51 for signal processing M1 be passed on. This information is in the assembly 51 in their instruction memory 36 stored. The processing unit 40 from assembly 51 processes these instructions further, integrated with the measurement data and device data from the situation memory 35 the assembly 51 , and then passes these over the channels I1 to those in the device layer 30 into the device 1 integrated workstation devices 7 back. The output devices 10 such as the speaker 17 , the projector 16 or the monitor 15 , submit these instructions as action recommendations to the user 2 ,

Information about the available action contexts is provided by the module 52 about the flow of information S3 to the overlying module 53 and thus to the processing layer M3 for conclusions. In the assembly 53 This information is used as input in the situation memory 35 persists.

In the assembly 53 containing the processing layer M3 Realized for conclusions are based on the in-situation memory 35 , in the instruction memory 36 and the knowledge base 37 the assembly 53 conclusions are drawn and action instructions are generated. The generated instructions are sent via the channel for instructions I3 to the subassembly underneath 52 returned and there in their instruction memory 36 stored persistently. Conclusions on the situation, on the other hand, are as information flow to the overlying assembly 54 containing the processing layer M4 realized for organization, passed on and there in their situation memory 35 stored persistently.

Due to the information processing in the assembly 53 the device behaves 1 such that the user 2 does not have the impression of interacting with a machine. In this respect, the module generates 53 the identity, that is the essence of the device 1 , Due to the effect of the assembly 53 gets out of the device 1 so out with the user 2 communicates that the Pattern of behavior of the data processing system 4 do not differ from those of a human being. The assembly 53 is secured in its quality by a Turing test adapted to the particular application scenario (for example "laboratory", "optician", "operating room", etc.). The behavior of the assembly 53 is through the in his knowledge base 37 filed rules with determined. The knowledge database 37 the assembly 53 includes rules that govern the moral behavior of the data processing system 4 the operator 2 determine against. The creation of these rules therefore requires special care for ethical reasons.

The behavior patterns are learned through a combination of traditional algorithmic processing components 41 and neuromorphic processing components 42 , in particular a neural network of the assembly 53 , generated and based on concrete symbolic rules. The weights of the parameters that make up the neuromorphic processing component 42 control and thus the behavior of the assembly 53 can determine similar to a program code managed and thus from the outside in versions in the device 1 be invited.

According to the in 3 The structure shown creates the module 53 two output streams. The information in issue A1 to the "situation" concern the situation in the workplace 3 and the resulting in the assembly 53 taken assessments and conclusions drawn. They are crossing the canal S4 to the next highest module 54 transmitted to the organization and in their situation memory 35 persists. The output A2 with the instructions will go over the channel I3 to the subassembly underneath 52 transmitted and there in the instruction memory 36 stored. These instructions flow in the assembly 52 in their calculations of the reactions and trigger there appropriate actions and micro actions. These can trigger actions in the information technology systems that are connected to the module 52 are connected, for example, an order process for a declining material. Further actions are taken over the underlying layers including the device layer 30 as action recommendations to the user 2 communicated.

The assembly 54 realizes the processing layer M4 for organization. At this level, the concrete situation plays on the device 1 supported workplace 3 only a minor role. Rather, it is about the processing of information about the general situation. In the assembly 54 the data are evaluated by several workstations 3 to be delivered in 5 are referred to as "workstation 2" to "workstation k", the evaluation taking place in particular across workplaces. On this basis, the jobs 3 controlled according to superior criteria and resources planned and used together. This procedure can be applied to any combination of jobs 3 Respectively. In particular, these jobs need 3 not be the same. However, the information exchange with the underlying assembly 53 the processing layer M3 for conclusions, the individual jobs 3 are assigned, uniformly shaped.

Also the processing unit 40 the assembly 54 includes an algorithmic processing component 41 and a neuromorphic processing component 42 , the latter being predominant. In the neuromorphic processing component 42 Behaviors are trained on the basis of concrete symbolic rules. In this respect, the assembly behaves 54 according to learned content. Furthermore, the neuromorphic processing component 42 determine correlations that may be present in the input data via their evaluation mechanisms, even if these are not available in the training via the compositor 43 were predetermined. The evaluator 44 in the assembly 54 is based on a set of symbolic rules and laws.

The through the assembly 54 realized processing layer M4 for organization is to be understood as an analogy to the human community, the collective. In particular, the rules that the evaluator 44 the assembly 54 underlying, by the owners of different jobs 3 established. Each of these jobs 3 covers its own device layer 30 as well as assemblies 51 to 53 , The commercial billing of the assembly 54 is broken down accordingly over several owners.

The from the assembly 54 as an issue A1 calculated situation is over the channel S5 further propagated to the top. However, there is no next higher processing layer 31 in the data processing system 4 is present this information is not further used. The as output A2 Calculated instructions, on the other hand, will be applied to the individual jobs below 3 concretized and to these through the channels I4 forwarded.

It should be noted that the assembly 54 does not necessarily have to be present. The device 1 is already with a device layer 30 as well as the assembly 51 executable, with still very simple functionality. With addition of the module 52 and optionally module 53 and optionally module 54 extends the power spectrum of the device 1 corresponding.

6 illustrates by means of a concrete example, the flow of information processing in the device 1 with the architecture 5 , It illustrates 6 in an exemplary manner with reference to a concrete embodiment of the device 1 and on the basis of a concrete situation as well as a concrete process step in the workplace 3 between the processing layers 31 occurring information flows in their time sequence. The information on the chronology is in the first column.

At the in 6 illustrated embodiment, the device 1 via some input devices 9 , some output devices 10 and some tools 11 , The associated workplace 3 is located in a chemical laboratory. Accordingly, the thermometer 19 , a heater 18 (Hotplate, oven) and three cameras 14 (CAM1, CAM2, CAM3) as input devices 9 as well as a speaker 17 (LSP) and an e-mail connection as an output device 10 or IT service available. A user 2 be at work 3 ,

From the device layer 30 are temperature readings and the position of the detected user 2 in a time slice 1 to the processing layer M1 transmitted for signal processing.

In a time slice 2 takes place in the processing layer M1 a consolidation on the one hand of the different camera results, in order to obtain a stable statement about the position of the user 2 on the other hand, the readings of the thermometer 19 , An example is the thermometer 19 a cognitive interpretation, namely "rise" and "steep rise" generated. This information is about S2 to the processing layer M2 handed over for reactions.

In time slice 3 be in the processing layer M2 for reactions synthesizes the incoming information to a situation. This is already quite extensive in this small example, but still symbolic in nature. It should be noted that in the processing layer M2 already a feedback simple rule-based decisions about I2 to the processing layer M1 such as safety instructions or immediate reactions. However, this is not necessary in this example. The information about the situation is about S3 to the processing layer M3 forwarded for conclusions.

In time slice 4 gets out of the processing layer M3 for conclusions the interpretation of the situation is determined and as an instruction about I3 to the processing layer M2 transmitted down. In the example, the device concludes 1 from the situation that connections between the non-compliance with the recipe prescription (DOE 89 ), the objects in the laboratory (thermometer 19 , Heater 18 ) and the user 2 exist as an available actionable resource. This generates the reaction, which in this case consists of a simple action recommendation, which is about I3 to the processing layer M2 is transmitted.

In the processing layer M2 this reaction is converted into a concrete action via rules (time slice 5 ) and finally the processing layer M1 for signal processing to the device layer 30 pass, specifically to the speaker LSP, which in time slice 6 recommends that the operator correct the oven temperature.

Here, however, the method example is not yet finished, because even in the overlying processing layer M4 for the organization, the findings from the processing layer M3 interpreted, but more expansive: it falls due to the over the information channel S4 communicated the situation that there is a systematic error associated with the recipe DOE 89 stands. In the situation memory 35 the processing layer M4 Beyond this concrete situation, more and longer-term data are available to make this conclusion possible. The finding is made about the processing layers M3 and M2 (Time slices 5 . 6 and 7 ) to the device layer 30 down given. In addition, the findings from the processing layer M4 enriched with a caring safety note. This would have the processing layer M2 taken alone so can not be generated. As output channel for the correction of the cooking instruction DOE 89 is that in the situation memory 35 the processing layer M2 present information to the creator of the regulation DOE 89 addressed.

The use of the device 1 offers several advantages. The following are some selected examples.

Security:

At the workplaces 3 A variety of rules should be observed in the manner described here: no food, wearing safety goggles, correct handling of toxic substances. The device 1 observes the overall situation and gives safety instructions to the user if necessary 2 out. In case of danger help is required. Even if not a human user 2 at work 3 the workplace can be present 3 be controlled - much further than with conventional methods. Also granting access and access can be via the device 1 be handled.

efficiency:

Materials and working time can be better used. For example, the multiple execution of experiments is avoided. By learning the material consumption, stocks can be optimized with foresight.

Training:

The handling of workstation devices 7 , materials and the right handles when working in the workplace 3 can by means of the device 1 directly to the users 2 mediated. Upon a question during the work the device delivers 1 promptly the appropriate information - whether by reading aloud, by teaching video projected on the laboratory table or augmented with a pair of glasses with Projekttionsfläche for augmented reality.

Immediate evaluation of the work:

Through standardized and uniform data management across all workstation devices 7 and jobs 3 For example, mathematical evaluation is much easier and faster to achieve. This allows a more detailed and faster evaluation of results without having to perform many individual operations such as copying and reformatting measured values and observation data.

Automatic publishing:

In the scientific world, a break with previous publication practice is indicated by social media. In addition to the classic peer review, an evaluation is carried out according to "interest". In the course of this change, the nature of the experimental and data base, based on which statements are made and subsequently published, will also change with modern techniques.

The device 1 documents all facts and processing steps automatically in the audit-proof logbooks 38 from which the experimental and evaluating parts of each publication can be produced at any time (assuming access authorization by the role of "Auditor").

Important for a paradigm shift in science is the rejection of "only positive findings are published". This arose from the lack of resources in the early years of research, but no longer makes sense today. Since traditionally "unsuccessful = not published" applies, negative results are often not published. As a result, too many failed experiments are repeated again.

The publication of negative results will find its way into publication practice - not necessarily as a technical article, but always as a rapidly growing global raw data archive for the scientific community, with automatically generated content descriptions and a cognitive initial analysis on a machine basis. The device described here 1 enables this paradigm shift.

Finally, it should be noted again that the data processing system 1 can form a physical processing unit or can be distributed as a logical unit over several physical computing units. Parts of the data processing system can be realized in hardware and / or software, wherein in particular for the neuromorphic processing components neuromorphic circuits can be used, which realize a neural network in hardware.

It should also be noted that the operation of the device may be performed using a computer program product whose code is for the operation of the device described herein when the code is executed by a processor or a neuromorphic processing unit. In one embodiment, the code may be stored on a data medium readable by the data processing system, such as a floppy disk, a compact disc (CD) or a digital versatile disc (DVD) or the like. In other embodiments, the computer program product may include code stored on a data store of a server or a group of servers. In another embodiment, the carrier may also be an electrical carrier signal used to transfer the code from a server to a server by downloading.

Finally, it should be noted that features and properties that have been described in connection with a particular embodiment can also be combined with another embodiment, except where this is excluded for reasons of compatibility.

Finally, it should be noted that in the claims and in the description, the singular includes the plural unless the context indicates otherwise. In particular, when the indefinite article is used, it means both the singular and the plural.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• CN 206058263 U [0003]

Claims (19)

Device for supporting a user (2) at a workstation (3) with: - a data processing system (4) assigned to the workstation (3), - at least one output unit (10) present at the workstation (3) and connected to the data processing system (4) for the reproduction of information, - at least one input unit (9) present at the workstation (3) and connected to the data processing system (4) for the sole purpose of transmitting data to the data processing system (4); 3) existing working equipment (11) managed by the data processing equipment and usable to carry out the activities carried out at the workstation (3), characterized in that the data processing equipment (4) is arranged to: - operations at the workstation (3) which Involve implements (11) to be detected by means of the implements (11) and / or the input unit (9), - detected on the basis of a process by means of a neuromorphic processing component (42) to recognize a situation and - to issue a message to the user (2) based on the detected situation. Device after Claim 1 , characterized in that the data processing system (4) has at least one processing layer (31) each having at least one assembly (32) in which a processing unit (40) is formed with at least one neuromorphic processing component (42). Device after Claim 2 characterized in that the processing unit (40) equipped with at least one neuromorphic processing component (42) additionally comprises at least one algorithmic processing component (41) cooperating with the neuromorphic processing component (42). Device after Claim 2 or 3 characterized in that the assembly (32) is switchable from an operating mode to a learning mode in which a compositor (43) applies training information to the neuromorphic processing component (42) and an evaluator (44) in response to the neuromorphic processing component (44). 42) generates a score and thereby applies to the neuromorphic processing component (42). Device according to one of Claims 2 to 4 , characterized in that the at least one module (32) provided in each processing layer (31) receives situation information from a subordinate module (32) via a communication channel (33) and sends it to a higher-level module (32) and via a further communication channel (34 ) Receives instruction information from the parent assembly (32) and sends it to the subordinate assembly (32). Device after Claim 5 , characterized in that the assembly (32) has a situation memory (35) upstream of the processing unit (40), in which situation information from the subordinate assembly (32) can be stored, and / or an instruction memory (36) located upstream of the processing unit (40). in which instruction information from the superordinate module (32) can be stored. Device after Claim 5 and 6 characterized in that the compositor (43) generates the training information based on information taken from the situation memory (35) and / or the instruction memory (36) and / or generates it itself and the evaluator (44) in response to the the situation information and / or instruction information generated by the neuromorphic processing component (42) generates the evaluation and thereby applies to the neuromorphic processing component (42). Device according to one of Claims 2 to 7 , characterized in that the processing unit (40) is connected to a database (37) in which processing rules for the processing of the situation information and the instruction information are stored. Device according to one of Claims 5 to 8th characterized in that the processing unit (40) performs the processing of the situation information in response to the instruction information. Device according to one of Claims 2 to 9 , characterized in that the data processing system has a device layer (30) lying beneath the at least one processing layer (31) and adapted to exchange data with the work devices (11), the processing layer (31) lying above the device layer (30) ) for the signal processing newly added tools (11) recognizes automatically by means of a device catalog. Device according to one of Claims 1 to 10 , characterized in that the input unit comprises a camera (14) and the data processing system (4) adapted for this purpose from the images supplied by the camera (14) Image recognition to generate data to detect the situation. Device after Claim 8 . 10 and 11 characterized in that the processing layer (51) overlying the device layer (30) is adapted to recognize newly added implements (11) to the device by means of the camera (14) and the subordinate image recognition and to start up the device newly added implement (11) read out data required from the catalog of services stored in the database (37). Device after Claim 10 or 12 CHARACTERIZED IN THAT said device comprises, above said signal processing processing layer (51), a further processing layer (52) for responses and, further, a conclusion processing layer (53) having said processing layer for reactions (52) compared to said processing layer for Conclusions (53) can be generated for the implements (11) and / or the output devices (10) for faster control than with the inference processing device (53). Device according to one of Claims 2 to 13 , characterized in that the device has an uppermost processing layer (54) to which a plurality of workstations (3) are connected. Device according to one of Claims 1 to 14 , characterized in that the output unit comprises a projector (16), with which hints for the user (2) on the workstation (3) can be projected. Device according to one of Claims 1 to 15 characterized in that the implements (11) are used for material processing and / or material examination. Method for supporting a user (2) at a workstation (3) in which: - a data processing system (4) assigned to the workstation (3) is used, - at least one output unit present at the workstation (3) and connected to the data processing system (4) (10) is used for the reproduction of information, - at least one input (9) present at the workstation (3) and connected to the data processing system (4) is exclusively used to transmit data to the data processing system (4), - a plurality of work equipment (11) available at the workstation (3) which can be used to carry out the activities carried out at the workstation (3), are managed by the data processing installation (4), characterized in that - by the data processing installation (4) operations at the workstation (3) that involve the implements (11), are detected by the implements (11) and / or the input unit (9) that - you rch the data processing system (4) by means of a neuromorphic processing component (42) a situation is detected and that - on the basis of the detected situation, a message to the user (2) is output. Method according to Claim 16 , characterized in that the input unit (9), the output unit (10), the working devices (11) and the data processing system (4), according to one of Claims 2 to 16 is structured Computer program product, characterized in that the computer program product contains program code which when executed on a data processing system (4) according to the method Claim 17 or 18 implemented.

Images