Controlling technical systems using artificial intelligence computing units

Abstract

The invention relates to a computer-implemented method for controlling a technical system, comprising the following steps: - reading in (S1) hardware configuration parameters of a control unit and a value of a real-time requirement of the control unit, - reading in (S2) hardware configuration parameters of an artificial intelligence computing unit (NPU), - reading in (S3) a control application for controlling the technical system, wherein the control application is configured to generate input values for the control unit based on artificial intelligence, - determining (S4) a processing time of the control application for executing the control application on the artificial intelligence computing unit taking into account the hardware configuration parameters of the control unit and the hardware configuration parameters of the artificial intelligence computing unit, - checking (S5) the determined processing time in accordance with the value of the real-time requirement of the control unit and outputting the result of the check, and - outputting (S6) the control application based on the result of the check for controlling the technical system.

Description

Technical Field

[0001] The invention relates to a computer-implemented method for controlling a technical system, a device for controlling a technical system, and a control system comprising an artificial intelligence (AI) computing unit. Background Art

[0002] In order to accelerate specific calculations, such as deep learning inference using artificial neural networks, specialized system-on-a-chip solutions, so-called AI accelerators, can be used. These AI accelerators can also be called artificial intelligence (AI) computing units or neural network processors (Neural Processor) or neural network processing units (NPU (Neural Processing Unit) for short).

[0003] Since such AI-based algorithms are increasingly also being used in industrial automation, such specific AI computing units can also be integrated into industrial control devices for controlling industrial facilities or machines.

[0004] EP 3 657 277 A1 describes an expansion unit for an automation device of an industrial system, wherein the expansion unit is configured to perform a data evaluation based on artificial intelligence.

[0005] However, in contrast to the typical application areas of AI accelerators, algorithms in industrial automation often have to comply with rigid real-time requirements, such as the maximum cycle time of a programmable logic controller (PLC), which is typically not set by such special artificial intelligence computing units, making integration into control processes with less stringent real-time requirements possible. Summary of the Invention

[0006] Therefore, the object of the present invention is to enable the integration of artificial intelligence computing units / AI accelerators into real-time-critical control processes.

[0007] This object is achieved by the measures described in the independent claims. Advantageous developments of the invention are described in the dependent claims. According to a first aspect, the invention relates to a computer-implemented method for controlling a technical system, the method comprising the following method steps:

[0008] - read in the hardware configuration parameters of the control unit and the values required by the control unit in real time,

[0009] -Read in the hardware configuration parameters of the artificial intelligence computing unit,

[0010] - reading in a control application for controlling the technical system, wherein the control application is designed to generate input values for the control unit based on artificial intelligence,

[0011] determining a processing time of the control application for executing the control application on the artificial intelligence computing unit, taking into account hardware configuration parameters of the control unit and hardware configuration parameters of the artificial intelligence computing unit;

[0012] - checking the determined processing time against the value of the real-time requirement of the control unit and outputting the checking result,

[0013] as well as

[0014] - Output control applications based on the inspection results to control technical systems.

[0015] In particular, the method can be at least partially computer-assisted or computer-implemented. In the context of the present invention, "computer-assisted" can be understood, for example, to mean the implementation of a method in which, in particular, a processor performs at least one method step of the method. In the context of the present invention, a processor can be understood, for example, as a machine or an electronic circuit. A processor can be, in particular, a central processing unit (CPU), a microprocessor or microcontroller, for example, an application-specific integrated circuit (possibly combined with a memory unit for storing program instructions), or a digital signal processor, etc. The processor can also be, for example, an IC (Integrated Circuit), in particular an FPGA (Field Programmable Gate Array) or an ASIC (Application-Specific Integrated Circuit), or a DSP (Digital Signal Processor) or a Graphics Processing Unit (GPU). A processor can also be understood as a virtualized processor, a virtual machine, or a soft CPU. For example, it may also involve a programmable processor that is equipped with configuration steps for executing the aforementioned method according to the present invention, or is configured using configuration steps so that: the programmable processor implements the method, components, modules or other aspects and / or sub-aspects of the present invention according to the features of the present invention.

[0016] Unless otherwise indicated in the following description, the terms "execute," "calculate," "computer-aided," "calculate," "ascertain," "generate," "configure," "reconstruct," etc. preferably refer to actions and / or processes and / or processing steps that modify data and / or generate data and / or convert data into other data, wherein the data can be represented or exist as physical variables, for example, as electrical impulses. In particular, the term "computer" is to be interpreted as broadly as possible to cover, in particular, all electronic devices with data processing capabilities. Computers can therefore include, for example, personal computers, servers, programmable logic controllers (PLCs), handheld computer systems, pocket PCs, mobile radios, and other communication devices that can process data in a computer-aided manner, processors, and other electronic devices for data processing.

[0017] In the context of the present invention, "providing" (especially in relation to data and / or information) may be understood to mean providing, for example, in a computer-assisted manner. For example, providing occurs via an interface, such as, for example, via a network interface, a communication interface or an interface to a storage unit. Via such an interface, for example, corresponding data and / or information can be transmitted and / or sent and / or retrieved and / or received during the provision.

[0018] In the context of the present invention, “hardware configuration parameters of the control unit” may be understood to mean, for example, the PLC type, the type and parameters of, for example, the backplane bus, the number and type of components, and the like.

[0019] In the context of the present invention, "real-time requirement" can be understood as the following period or time point of a process for a control unit: during this period or at this time point, the process is to be carried out. Therefore, "real-time required value" can be such a condition in a time unit.

[0020] In the context of the present invention, a "control application" may be understood in particular to mean a software application / application suitable for controlling a technical system, that is, for example, a software application / application that provides at least one input value to a control unit of the technical system. The control application is configured to generate input values for the control unit based on artificial intelligence. In other words, in this case, it may be an AI-based control application. For example, the artificial intelligence may be implemented as an artificial neural network, so that when the control application is read in, in particular, a certain number and type of input nodes and output nodes of the artificial neural network are read in.

[0021] In the context of the present invention, an "artificial intelligence computing unit" may be understood to mean, in particular, an AI accelerator or a neural network processing unit. Such a computing unit is preferably one that is particularly suitable for performing calculations based on artificial intelligence, that is, a dedicated AI computing unit or a computing unit suitable for executing AI-based applications. In other words, the computing unit may be assigned in particular to AI calculations, such as, for example, calculations using artificial neural networks.

[0022] In the context of the present invention, a “hardware configuration of an artificial intelligence computing unit” may be understood to mean, in particular, information about hardware components of the computing unit / NPU component, such as, for example, a processor clock, a memory clock, and / or a bus bandwidth.

[0023] In particular, “processing time” may also be understood to mean the processing duration, the total computing time, the total execution duration or the execution time with respect to the execution of the control application on the AI computing unit.

[0024] The present invention makes it possible to check whether a control application meets the real-time requirements of a control unit. In particular, this involves checking whether real-time requirements are adhered to when executing an AI-based control application on an AI computing unit. Consequently, the present invention has the advantage of determining whether a control application executed on an AI computing unit / AI accelerator can be used or is suitable for controlling a system with strict real-time requirements. This is particularly true with respect to the corresponding hardware.

[0025] In one implementation, processing time may include:

[0026] - the transmission time for data transmission between the control unit and the artificial intelligence computing unit,

[0027] and / or

[0028] -Execution time of the application on the AI computing unit.

[0029] In one embodiment, the execution time can be determined by means of a computer-aided simulation of the execution of the control application on the artificial intelligence computing unit.

[0030] For this purpose, for example, CPU simulation can be used. Cache behavior can also be taken into account.

[0031] In an alternative embodiment, the execution time can be determined by executing the control application on the artificial intelligence computing unit.

[0032] It is also possible to determine the processing time using direct measurements. This allows each characteristic of a specific computing unit to be taken into account.

[0033] In one embodiment, based on the inspection results,

[0034] - the value of the real-time requirement of the control unit can be adapted to the determined processing time, and / or

[0035] - By modifying the control application, the execution time of the control application can be adapted to the value of the real-time requirement.

[0036] This enables either a revision of the real-time standard and / or an adaptation of the control application to the corresponding hardware.

[0037] In one embodiment, the control application can be modified iteratively until the determined execution time of the control application meets the value of the real-time requirement of the control unit.

[0038] In particular, the control application can be modified with the aid of the optimization method, wherein the control application is adapted iteratively until the execution time is minimized.

[0039] In one embodiment, the control application may include an artificial neural network, and the nodes of the neural network may be iteratively reduced until the determined execution time of the control application meets a value of the real-time requirement.

[0040] This allows, for example, to remove nodes with low weights from an artificial neural network. This can be determined, for example, using sensitivity analysis. As in sensitivity analysis, the extent to which nodes in a neural network contribute to the calculated output can be studied by stimulating them with input signals. Nodes with low influence can be ignored.

[0041] In one embodiment, a control application can be executed on the artificial intelligence computing unit, thereby generating input values for the control unit and controlling the technical system based on these input values.

[0042] In particular, in the case of a positive check result, ie when the processing time meets the real-time requirements, the control application can be executed.

[0043] According to a second aspect, the invention relates to a device for controlling a technical system, comprising:

[0044] a first interface which is configured to read in hardware configuration parameters of the control unit and values required by the control unit in real time,

[0045] a second interface, which is configured to read in hardware configuration parameters of the artificial intelligence computing unit,

[0046] a third interface, which is configured to read in a control application for controlling the technical system, wherein the control application is configured to generate input values for the control unit based on artificial intelligence,

[0047] an analysis module, which is configured to determine a processing time of a control application for executing the control application on the artificial intelligence computing unit to generate input values for the control unit, taking into account hardware configuration parameters of the control unit and hardware configuration parameters of the artificial intelligence computing unit;

[0048] a checking module which is designed to check the determined processing time against a value required in real time by the control unit and to output a result of the checking;

[0049] Know

[0050] An output module is designed to output a control application for controlling the technical system as a function of the test result.

[0051] In particular, the device and / or at least one of its interfaces or modules can be implemented in hardware and / or software. Preferably, the device comprises at least one processor. The device is preferably coupled to a control unit and an artificial intelligence computing unit of the technical system.

[0052] In one embodiment, the device may include a simulation module that is designed to determine the execution time by means of a computer-aided simulation of the execution of the control application on the artificial intelligence computing unit.

[0053] "Simulation module" can also be understood as a simulation environment.

[0054] In one embodiment, the device may include an optimization module that is designed to iteratively modify the control application until the determined execution time of the control application meets a value for a real-time requirement of the control unit.

[0055] Preferably, the optimization module includes an optimization method by which the control application can be iteratively modified so that the execution time of the control application is minimized. Preferably, the optimization module is coupled to at least the analysis module and the checking module so that the control application can be iteratively modified and checked until the execution time meets the real-time requirements.

[0056] According to another aspect, the present invention relates to a control system comprising:

[0057] - a device according to the invention,

[0058] an artificial intelligence computing unit, which is designed to execute a control application for controlling the technical system, wherein the control application is designed to generate input values for the control unit based on artificial intelligence,

[0059] A control unit which is designed to control the technical system on the basis of input values generated by the control application.

[0060] Furthermore, the present invention relates to a computer program product which can be directly loaded into a programmable computer and comprises program code portions which, when the program is executed by a computer, cause the computer to carry out the steps of the method according to the invention.

[0061] The computer program product can be provided or supplied, for example, on a storage medium, such as a memory card, a USB stick, a CD-ROM, a DVD, a non-volatile / persistent storage medium (English: Non-transitory storage medium), or can also be provided or supplied in the form of a file that can be downloaded from a server in a network. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In the drawings, exemplary embodiments of the method according to the invention, the device according to the invention and the control system according to the invention are shown and explained in more detail in the following description. In these figures:

[0063] Figure 1 A first embodiment of the method according to the invention is shown;

[0064] Figure 2 A second embodiment of the method according to the invention is shown; and

[0065] Figure 3 Embodiments of a device according to the invention and of a control system according to the invention are shown.

[0066] Parts that correspond to one another are provided with the same reference numerals in all the figures. DETAILED DESCRIPTION

[0067] In particular, the following exemplary embodiments merely show exemplary implementation possibilities, ie, in particular how these implementations according to the teaching of the present invention may appear, since it is not possible to mention all of these implementation possibilities and it is not convenient or necessary to mention all of them for understanding the present invention.

[0068] In particular, all possibilities for implementing the invention which are common in the prior art are also known to the (relevant) person skilled in the art who is aware of the / the method claim, so that in particular they do not need to be disclosed separately in the description.

[0069] Figure 1An exemplary embodiment of a method for controlling a technical system according to the present invention is shown as a flow chart. In this case, the technical system may be, for example, an industrial facility, such as a production facility, an automation system, or a robotic system. In particular, the technical system can be controlled using a control application based on a machine learning method / machine learning algorithm, wherein the control application outputs input values to a control unit of the technical system. For this purpose, the control unit is coupled to an artificial intelligence computing unit, wherein the control application executes on the artificial intelligence computing unit.

[0070] The control of a technical system may have real-time requirements that must be met when the control application is executed on the artificial intelligence computing unit. This can be checked using the method described below so that the control application for controlling the technical system can be released and / or executed.

[0071] In the first step S1 of the method, the hardware configuration or hardware specifications of the control unit are acquired. To this end, hardware configuration parameters of the control unit, such as the PLC type, memory size, memory speed, and / or bus bandwidth, are read in. Furthermore, at least one value for the real-time requirement of the control process, such as an upper time limit, is read in. This real-time requirement can be determined, for example, based on the hardware configuration.

[0072] In a next step S2 , hardware configuration parameters of the artificial intelligence computing unit are read in, such as, for example, type, computing power, NPU clock / size, internal bus speed and / or bandwidth.

[0073] Furthermore, information about the firmware and / or software of the artificial intelligence computing unit and / or the control unit can also be read in.

[0074] In the next step S3, a control application suitable for controlling the technical system is read in. This control application is designed to generate and output input values for the control unit based on artificial intelligence (e.g., an artificial neural network). In other words, the control application is based on artificial intelligence. Furthermore, the control application is designed to be executed on an artificial intelligence computing unit.

[0075] In a next step S4, a processing time, preferably a maximum or minimum processing time, for executing the control application on the artificial intelligence computing unit is determined, wherein hardware configuration parameters of the artificial intelligence computing unit and hardware configuration parameters of the control unit are taken into account, i.e., the calculation of the processing time is determined specifically for these hardware configuration parameters. Optionally, the firmware and / or software of the artificial intelligence computing unit and / or the control unit may also be taken into account.

[0076] The processing time here includes the transmission time for data transmission between the control unit and the artificial intelligence computing unit and / or the execution time for executing the application on the artificial intelligence computing unit. In other words, the processing time is preferably the sum of the transmission time and the execution time. In particular, the processing time describes the time required to execute the control application and transmit the output input values to the control unit.

[0077] In particular, the execution time can be determined by computer-aided simulation of the execution of the control application on the artificial intelligence computing unit (step S4a), wherein the artificial intelligence computing unit is depicted / modeled based on hardware configuration parameters. Alternatively, the execution time can also be determined by directly executing the control application on the actual artificial intelligence computing unit (step S4b).

[0078] The maximum transmission time between the functional blocks of the control unit and the NPU components preferably includes the request and response time between these components, for example, the total transmission time on the backplane bus (including the maximum latency from the bus request to the bus grant in the bus arbitration control), the maximum clock skew when switching between different clock domains of the hardware, and the maximum reaction time of the interrupt service routines of the control system and the NPU components.

[0079] The maximum execution time on the NPU component includes, for example, the maximum execution time of an Atom code segment (without incoming / outgoing branches). These Atom code segments can be calculated or simulated, for example, with the help of a static worst-case execution time analysis (StaticWorstCaseExecutionTimeAnalysis). Here, at the beginning of the code segment, all caches must be assumed to be "cold", that is, they must not contain any required cache lines (Cache Lines). Possible calculations can be performed in parallel on different resources. Therefore, in the case of full parallelization, the longest path in the data flow graph of the control application or the artificial intelligence of the control application represents the upper limit of the execution time.

[0080] In the next step S5, it is checked whether the determined processing time meets the real-time requirement of the control unit. To this end, the determined processing time is compared with at least one value of the real-time requirement.

[0081] If the processing time meets the real-time requirements of the control unit (i.e., path Y), a control application for controlling the technical system is output (steps S6 and S10). To this end, the control application can be set up and executed on the artificial intelligence computing unit, where it generates input values for the control unit. The control unit can then control the technical system based on these input values (step S10).

[0082] If the processing time does not meet the real-time requirements of the control unit (i.e., path N), the real-time requirements can be adapted, for example, in step S7. For example, the upper limit of the processing time can be modified accordingly. The processing time can then be rechecked based on the modified real-time requirements. This allows the control system of the technical system to be adapted so that a predefined AI-based control application can be used.

[0083] Figure 2 A further embodiment of the method according to the invention for controlling a technical system is shown. Steps S1 to S6 and S10 are Figure 1 These steps correspond to .

[0084] If the check of the processing time (step S5) shows that the processing time does not meet the real-time requirements of the control unit of the technical system, i.e. path N, then in addition to adapting to the real-time requirements ( Figure 1 In addition to or instead of step S7 in the adaptation real-time requirements ( Figure 1 In step S7 of the process, the control application may be modified (step S8) so that the execution time of the control application is adapted to the value of the real-time requirement.

[0085] To this end, the control application can be iteratively adapted (step S9) until the resulting execution time meets the real-time requirement. For example, the control application can include an artificial neural network, so that the nodes of the neural network are iteratively reduced until the determined execution time of the control application meets the value of the real-time requirement. This can be supported, for example, by an optimization method, in which the artificial neural network is changed / adapted / reduced over time, for example by iteratively removing or adding nodes, until the execution time determined for this corresponds at most to the maximum cycle time of the control unit. For example, a sensitivity analysis can be performed on the corresponding nodes in order to eliminate those with low weights.

[0086] Figure 3 An embodiment of a device 100 according to the invention for controlling a technical system TS and a control system 200 according to the invention for controlling a technical system TS is shown. For example, the device 100 can be coupled to the control system 200 or can be integrated in the control system 200. The device 100 is preferably configured to execute the Figure 1 and / or Figure 2 The steps of the method shown in .

[0087] A control system 200 is shown, comprising the device 100, an artificial intelligence computing unit NPU (e.g., a neural network processing unit), and a control unit PLC. The technical system may be, for example, an image-controlled robotic system that can be controlled based on a machine learning algorithm. The control system 200 preferably includes at least one processor CPU that centrally processes data from the neural network processing unit NPU and the control unit PLC.

[0088] Device 100 includes a first interface 101 and a second interface 102. The first interface 101 is configured to read in hardware configuration parameters HW1 of a control unit PLC and a value of the control unit PLC's real-time requirement RT. The second interface 102 is configured to read in hardware configuration parameters HW2 of an artificial intelligence computing unit NPU. Device 100 also includes a third interface 103 configured to read in a control application APP for controlling a technical system. The control application APP is configured to generate input values for the control unit based on artificial intelligence. Device 100 further includes an analysis module 104 configured to determine a processing time T of the control application, taking into account hardware configuration parameters HW1 of the control unit and hardware configuration parameters HW2 of the artificial intelligence computing unit. This processing time T is the time required to execute the control application APP on the artificial intelligence computing unit NPU to generate input values for the control unit. The device further includes a checking module 105 and an output module 106. The checking module 105 is configured to check the determined processing time T against the real-time requirement of the control unit and output a check result CR. The output module 106 is configured to output a control application APP for controlling the technical system TS if the check result is positive. This check, for example, includes comparing the processing time with the real-time requirement. In particular, the check result is positive if the processing time T of the control application APP meets the real-time requirement RT of the control unit PLC.

[0089] The processing time T is preferably determined using a computer-assisted simulation, such as a CPU simulation, of the execution of the control application APP on the neural network processing unit NPU. To this end, the device 100 may include a simulation module 107 configured to perform such a computer-assisted simulation. The simulation module 107 is preferably coupled to the analysis module 104. For the calculation, the hardware configuration parameters HW2 of the neural network processing unit NPU are transmitted to the simulation module 107, in particular, so that a computer-assisted simulation specific to the neural network processing unit NPU can be performed.

[0090] If the control application APP meets the real-time requirements RT of the control unit, it is executed on the neural network processing unit NPU and supplies the control unit PLC with input values CTL for controlling the technical system TS.

[0091] If the check module 105 outputs a negative check result, that is, if the processing time T of the control application APP does not meet the real-time requirement RT, the execution time of the control application can be optimized with the aid of the optional optimization module 108 until the real-time requirement RT is met. The optimization module 108 is configured to iteratively modify the control application APP until the determined execution time of the control application meets the value of the real-time requirement of the control unit. For example, the machine learning model underlying the control application APP can be reduced by, for example, removing nodes of an artificial neural network.

[0092] Within the scope of the invention, all described and / or illustrated features can be advantageously combined with one another. The invention is not restricted to the described exemplary embodiments.

Claims

1. A computer-implemented method for controlling a technical system (TS), comprising the following steps: - reading in the hardware configuration parameters (HW1) of the control unit (PLC) and the value of the real-time requirement (RT) of said control unit (PLC), - Read in the hardware configuration parameters (HW2) of the artificial intelligence computing unit (NPU), - reading in a control application (APP) for controlling the technical system, wherein the control application is configured to generate input values (CTL) for the control unit based on artificial intelligence, - determining a processing time of the control application for executing the control application on the artificial intelligence computing unit, taking into account the hardware configuration parameters of the control unit and the hardware configuration parameters of the artificial intelligence computing unit, wherein the processing time includes an execution time for executing the application on the artificial intelligence computing unit, - checking the determined processing time (T) according to the value of the real-time requirement (RT) of the control unit and outputting the result of the checking, as well as - outputting the control application to control the technical system based on the inspection result, wherein the execution time of the control application is adapted to the value of the real-time requirement by modifying the control application, wherein the control application is modified iteratively until the determined execution time of the control application meets the value of the real-time requirement of the control unit, The control application comprises an artificial neural network, and the nodes of the neural network are iteratively reduced until the determined execution time of the control application meets the value of the real-time requirement.

2. The computer-implemented method of claim 1 , wherein: The processing time (T) includes: - Transmission time for data transmission between the control unit and the artificial intelligence computing unit.

3. The computer-implemented method of claim 2, wherein: The execution time is determined by means of a computer-aided simulation of the execution of the control application on the artificial intelligence computing unit.

4. The computer-implemented method of claim 2, wherein: The execution time is determined by executing the control application on the artificial intelligence computing unit.

5. The computer-implemented method according to any one of claims 1 to 4, wherein: According to the inspection results, - The value of the real-time requirement of the control unit is adapted to the determined processing time.

6. The computer-implemented method according to any one of claims 1 to 4, wherein: The control application (APP) is executed on the artificial intelligence computing unit (NPU), thereby generating the input values (CTL) for the control unit (PLC) and controlling the technical system based on the input values.

7. A device (100) for controlling a technical system (TS), comprising: a first interface (101) which is configured to read in hardware configuration parameters (HW1) of a control unit (PLC) and the value of a real-time requirement (RT) of the control unit (PLC), a second interface (102) configured to read in hardware configuration parameters (HW2) of the artificial intelligence computing unit (NPU), a third interface (103) configured to read in a control application (APP) for controlling the technical system, wherein the control application (APP) is configured to generate input values for the control unit (PLC) using artificial intelligence, an analysis module (104) configured to determine a processing time (T) of the control application for executing the control application on the artificial intelligence computing unit to generate the input value for the control unit, taking into account the hardware configuration parameters (HW1) of the control unit and the hardware configuration parameters (HW2) of the artificial intelligence computing unit, wherein the processing time includes an execution time for executing the application on the artificial intelligence computing unit, a checking module (105) which is designed to check the determined processing time (T) with respect to the value of the real-time requirement of the control unit and to output a checking result (CR), an output module (106) which is configured to output the control application for controlling the technical system based on the check result, and - an optimization module (108) which is designed to iteratively modify the control application until the determined execution time of the control application (APP) meets the value of the real-time requirement (RT) of the control unit, wherein the control application comprises an artificial neural network and iteratively reduce the nodes of the neural network until the determined execution time of the control application meets the value of the real-time requirement.

8. The device according to claim 7, comprising a simulation module (107), which is configured to determine the execution time by means of a computer-aided simulation of the execution of the control application (APP) on the artificial intelligence computing unit.

9. A control system (200), comprising: - an apparatus (100) according to any one of claims 7 to 8, an artificial intelligence computing unit (NPU) which is configured to execute a control application (APP) for controlling the technical system, wherein the control application is configured to generate input values (CTL) for the control unit on the basis of artificial intelligence, The control unit (PLC) is configured to control the technical system based on the input values generated by the control application. 10 . A computer program product which can be directly loaded into a programmable computer, the computer program product comprising program code portions which are suitable for executing the steps of the method according to claim 1 .

Images