BEVERAGE MAKER AND METHOD FOR OPERATING A BEVERAGE MAKER

Description

[0001] A beverage maker and a method for operating a beverage maker are provided. The beverage maker contains a category 1 sensor that delivers only a single measured value S1 at a specific time, a category 2 sensor that delivers multiple measured values ​​S2 at a specific time, a data memory, and a control unit. The control unit is configured to compare the measured values ​​S2 with target values ​​stored in the data memory and, based on this comparison, to trigger or not trigger at least one event. If the comparison does not trigger an event, the control unit records at least one measured value S1, wherein the measured value S1 is used to assign the measured values ​​S2 to at least one event and to store these measured values ​​S2 as target values ​​for the assigned event in the data memory.

[0002] Machines, such as beverage makers, are primarily designed to receive commands from users and translate them into a result. The commands (e.g., in the form of entire command chains) are therefore processed by the machine and completed with a result (expected by the user). Complex tasks can no longer be easily solved by machines in this way, as it is very time-consuming to capture, define, and program the complex tasks and their constraints into command chains. This process is time-consuming, carries a certain risk of error, and is associated with high costs.

[0003] The term "artificial intelligence" in relation to a computer is commonly understood to mean that the computer is built or programmed to solve problems independently. The creation of an artificial neural network involves creating a network of nodes and connections at the software level. The network evolves in the software by deleting, adding, or re-weighting connections or nodes.

[0004] Machine learning is based on algorithms and code. In machine learning, the machine acquires data, learns from the data, and makes appropriate decisions. The machine learning algorithm can be represented as a flowchart or decision tree. The individual branches of the flowchart or decision tree are weighted by the data collected by the machine.

[0005] Supervised machine learning is based on a predetermined output to be learned, the results of which are known (as of March 22, 2018: https: / / de.wikipedia.org / wiki / %C3%9Cberwachtes_Lernen). The results of the learning process can be compared with the known, correct results, i.e., "supervised." Typically, a learning step looks like this: 1. Creating the input; 2. Processing the input (propagation); 3. Comparing the output with the desired value (error); 4. Reducing the error by modifying the weighting (e.g., with backpropagation).

[0006] In unsupervised machine learning, no target values ​​are known in advance. The machine attempts to recognize and interpret patterns.

[0007] In backpropagation, the algorithm runs in the following phases (as of March 22, 2018: https: / / de.wi-kipedia.org / wiki / Backpropagation): 1. An input pattern is applied and propagated forward through the network; 2. The network's output is compared with the desired output, and the difference between the two values ​​is considered the network's error; and 3. The error is then propagated back through the output layer to the input layer, with the weights of the connections in the network changing depending on their influence on the error. This guarantees an approximation of the desired output when the input is applied again.

[0008] Deep learning is a branch of machine learning that involves analyzing vast amounts of data to identify and derive patterns.

[0009] Artificial intelligence is particularly advantageous when a machine is required to perform a specific process based on a complex detection pattern of a sensor (e.g. a collection of several measuring points instead of a single measured value).

[0010] Sensors can basically be divided into two different categories, namely 1st category sensors and 2nd category sensors.

[0011] Sensors in the first category only provide a single measured value at a specific time, such as a specific temperature (temperature sensor), a specific pressure (pressure sensor), a specific speed (speed sensor), a specific flow rate (flow rate sensor) or a specific electrical capacitance. If this category of sensors measures continuously, individual measured values ​​are obtained as a function of time. The result of the measurement can be displayed in an xy diagram, where x represents the time and y represents the individual measured value. An example of this would be the touchscreen of a beverage maker, which triggers a specific event, such as the dispensing of a specific beverage, when a specific pressure is exceeded (pressure-sensitive touchscreen) or when a specific capacitance is detected (capacitive touchscreen) on a specific field of the touchscreen.

[0012] Sensors in the second category provide not just a single, but at least two different measured values ​​at a specific time, such as the frequency and amplitude of a longitudinal oscillation (e.g. acoustic sensor or vibration sensor), the frequency and amplitude of a transverse oscillation (e.g. light sensor, IR sensor or radar sensor), the direction and strength of a magnetic field (magnetic field sensor) or the electrical conductivity, pH value and sugar concentration of a specific liquid (taste sensor). If this category of sensors measures continuously, at least two different measured values ​​are obtained depending on the time. The measurement result can be displayed in an xyz diagram for two different measured values, where x represents the time, y the first measured value and z the second measured value.

[0013] Electrical devices (e.g., beverage makers such as coffee machines) prefer to use Category 1 sensors, as these sensors provide the measured values ​​without requiring significant computational effort. For example, a resistance-based temperature sensor outputs a specific resistance at a specific time, and this specific resistance is used to determine a specific temperature.

[0014] Electrical devices contain software that can specify how the device should react when a specific measured value is measured at a specific time. For example, a device action can be triggered when a specific threshold is exceeded. Generally speaking, the software stores a table in which measured values ​​are assigned to a specific device state or a specific device action (i.e., a specific device control command). In this way, the device can be controlled solely using a single measured value from a sensor, e.g., a specific action can be performed.

[0015] Electrical devices can also be controlled with Category 2 sensors. In this case, too, tables or patterns can be created in which an assignment of specific measured values ​​(patterns) recorded by the sensor to one or more machine actions is coded. However, the effort required to create the assignment(s) is significantly greater than with the (simpler) Category 1 sensors. The reason for the increased effort is that in order to command the device to perform a specific action, at least two measured values ​​must each have a specific, predefined value ("and" operation of the y-value and z-value at time x), thus increasing the complexity of the control system.

[0016] It is assumed that the increasing demands on electrical devices (in particular: beverage makers such as coffee machines), i.e. complex control processes in electrical devices, can only be met with sensors of the 2nd category in the future.

[0017] EP 1 647 951 A1 discloses a beverage maker having a control device which has a characteristic value of a container in its long-term data memory, wherein the control device can control a flow rate of liquid into the container.

[0018] EP 2 279 683 A1 discloses a method for adjusting a technical parameter by a user-controlled neural network controlling the parameter to influence the extraction of consumable substances during the production of foodstuffs, and a vending machine for implementing the method.

[0019] WO 2004 / 024615 A1 discloses a dispensing device for beverages with an identification device, wherein the identification device is suitable for identifying different types of containers and for emitting an identification signal describing the container, wherein the dispensing device is further provided with a valve device which, on the basis of the identification signal, can control at least one feed device of the dispensing device in order to fill the container with a predetermined amount of food.

[0020] US 2011 / 283888 A1 discloses a beverage maker comprising a water reservoir, a boiler, a brewing unit, a user interface, a control unit and a coffee container, wherein the coffee container has a capacitive sensor for detecting a residual amount of coffee in the coffee container and wherein the control unit is programmed to provide a user via the user interface with at least one item of information about the amount of coffee in the coffee container.

[0021] EP 0 527 567 A2 discloses a method for controlling an object, wherein signals are derived from a single sensor measuring a single property of the controlled object and used by a set sampling unit to derive a plurality of control characteristics for the object, wherein the plurality of control characteristics are processed by a neural network to derive a basic control signal, and properties derived from a further sensor and a local state detector are analyzed by a fuzzy logic system and used to modify the basic control signal.

[0022] It is therefore the object of the present invention to provide a beverage maker that can learn and perform complex control processes in a simple manner and with minimal effort for a user.

[0023] The invention is achieved by a beverage maker having the features of claim 1 and a method having the features of claim 8. The dependent claims show advantageous developments.

[0024] According to the invention, a beverage maker is provided, comprising a) at least one sensor of a 1st category, wherein a sensor of the 1st category is understood to be a sensor that delivers only a single measured value S1 at a specific time; b) at least one sensor of a 2nd category, wherein a sensor of the 2nd category is understood to be a sensor that delivers multiple measured values ​​S2 at a specific time; c) a data memory; and d) a control unit, wherein the control unit is configured to compare the measured values ​​S2 with target values ​​for the measured values ​​S2 that are stored in the data memory and code for at least one specific event, and based on this comparison to automatically trigger or not trigger at least one event in the beverage maker; wherein the control unit of the beverage maker is further configured to record and use at least one measured value S1 to assign certain measured values ​​S2, which do not trigger an event after a comparison with target values ​​for the measured values ​​S2, to at least one event, and to store these measured values ​​S2 as target values ​​for the at least one event on the data memory, characterized in that the beverage maker is further configured to assign measured values ​​S2 of the at least one sensor of the 2nd category, based on a confirmation via the at least one sensor of the 1st category by a user, to certain, regularly occurring events, wherein the assignment is carried out according to the principle of machine learning, and wherein the storage of the measured values ​​S2 for the events is carried out automatically on the data memory.

[0025] The beverage maker according to the invention is thus suitable for teaching one or more category 2 sensors contained in the beverage maker to respond to certain regularly occurring events with the aid of one or more category 1 sensors contained in the beverage maker. An event here is, for example, the dispensing of espresso by the beverage maker confirmed by the user via the at least one category 1 sensor when the at least one category 2 sensor detects an espresso cup at a specific location on the beverage maker (e.g., under a brewing group). The user can confirm this by quickly and easily pressing a button. The measured values ​​S2 for this event are automatically saved on the data memory.

[0026] The advantage of "learning" the beverage maker via at least one sensor of the 1st category is that the effort required for the user of the beverage maker to learn is very low (e.g. a simple push of a button).

[0027] If, after learning, an espresso cup is again placed within the detection range of at least one Category 2 sensor, the comparison by the beverage maker's control unit of the pattern detected by the Category 2 sensor with the pattern stored in the data memory causes the beverage maker to automatically trigger the learned event (i.e., without further user input). The learned, automatically triggered event can, for example, be the display of a "yes / no" query for a specific beverage (e.g., espresso), the display of a timer until a specific beverage (e.g., espresso) is dispensed, the display or highlighting of a specific selection of beverages (e.g., espresso, ristretto, espresso macchiato), and / or the immediate dispensing of an espresso.If the learned event involves the immediate dispensing of an espresso by the beverage maker, the beverage maker will automatically dispense an espresso. In other words, a repeated confirmation or selection of the espresso dispensing by the user is no longer necessary. If the learned event involves the display or highlighting of a specific selection of beverages, the beverage maker will automatically display or highlight the specific selection of beverages. The user can then select a specific beverage from this selection. The selection can be made via a category 1 sensor, for example by pressing a button on the beverage maker or touching a field on a touchscreen on the beverage maker, or via a category 2 sensor, for example by performing a gesture and / or facial expression (e.g. eye control) in front of a camera on the beverage maker.

[0028] The beverage maker according to the invention therefore assigns measured values ​​from sensors of the 2nd category based on a confirmation via a sensor of the 1st.

[0029] It assigns a category to specific events by a user, stores this assignment in its data storage, and reacts accordingly in the future if a previously occurred (known) event occurs again. This assignment is based on the principle of machine learning.

[0030] As the frequency of events during operation of the beverage maker increases, so does the beverage maker's wealth of experience. Therefore, as the beverage maker is exposed to many events, the probability increases that a specific detection pattern is already known to the second-category sensor and will thus trigger an event even without activating a first-category sensor. In other words, the connection between the patterns detected by the second-category sensor and the associated events becomes stronger. Accordingly, the importance of the first-category sensors increasingly decreases and, in the idealized case of a "fully trained" beverage maker, no longer plays any role at all. The triggered event can also be, for example, that a user interface of the beverage maker is displayed in a certain form, i.e., differently than before the event was triggered.A specific beverage selection can be displayed or highlighted on the user interface. Selection of a specific beverage within this selection can then be achieved, for example, by pressing a button (1st category sensor) or by using a specific gesture (2nd category sensor).

[0031] In a preferred embodiment, the beverage maker according to the invention contains a user interface, preferably in the form of a touchscreen. The user interface can be configured to display a triggered event. Furthermore, the user interface can be configured to display a countdown until the implementation of a triggered event. In addition, the user interface can be configured to display a (specific) beverage selection. Furthermore, the user interface can be configured to display information about measured values ​​S2 of the at least one sensor of a second category, preferably information selected from the group consisting of the height of a beverage container, the fill level of a beverage container, the fill level of a bean container of the beverage maker, the empty grinding of a grinder of the beverage maker, and combinations thereof.Furthermore, the operator interface can be configured to display target values ​​for S2 and to confirm and / or delete them at the request of a user.

[0032] The control unit of the beverage maker can be configured to trigger the at least one event encoded in the target values ​​for the measured values ​​S2 of the sensor of the 2nd category if the measured values ​​S2 can be assigned to target values ​​for the measured values ​​S2 with a probability of >50%, preferably ≥60%, particularly preferably ≥70%, very particularly preferably ≥80%, in particular ≥90%.

[0033] In a preferred embodiment, the control unit of the beverage maker is configured to use at least one measured value S1 if the measured values ​​S2 can be assigned to target values ​​for the measured values ​​S2 with a probability of only ≤50%, preferably ≤40%, particularly preferably ≤30%, very particularly preferably ≤20%, in particular ≤10%.

[0034] The at least one sensor of the first category can be configured to provide a one-dimensional measured value S1 at a specific time. The one-dimensional measured value can be selected from the group consisting of temperature, pressure, speed, flow velocity, electrical capacitance, direction of movement, distance, time, and mass.

[0035] The at least one sensor of the first category can be selected from the group consisting of temperature sensors, pressure sensors, speed sensors, flow rate sensors, electrical current sensors, and capacitive sensors. Preferably, the at least one sensor of the first category is a pressure sensor and / or capacitive sensor. Most preferably, the sensor of the first category is a touchscreen of the beverage maker.

[0036] The at least one sensor of the 2nd category can be configured to deliver multi-dimensional measured values ​​S2 at a specific point in time, preferably at least three-dimensional measured values ​​S2, particularly preferably at least four-dimensional measured values ​​S2. One dimension of the multi-dimensional measured value S2 can be selected from the group consisting of frequency of an oscillation (e.g. wavelength of light, ie light color), amplitude of an oscillation (e.g. amplitude of light, ie light intensity), polarization of an oscillation (e.g. polarization of light), direction of propagation of an oscillation (e.g. direction of propagation of light), direction of propagation of lines of force (e.g. direction of propagation of magnetic field lines), strength of lines of force (e.g. strength of magnetic field lines). A further dimension of the multi-dimensional measured value S2 can also be selected from the above-mentioned group, wherein the further dimension is then different from the (first) dimension.

[0037] The at least one sensor of the 2nd category can be selected from the group consisting of sensors for detecting longitudinal vibrations, sensors for detecting transverse vibrations, sensors for detecting magnetic fields, sensors for detecting electric fields, chemical sensors, electrical sensors, and combinations thereof. The sensor of the 2nd category is preferably selected from the group consisting of sound sensors, vibration sensors, light sensors (e.g., a camera), IR sensors, radar sensors, magnetic field sensors, conductivity sensors, pH sensors, sensors for detecting a sugar concentration, and combinations thereof. The sensor of the 2nd category is preferably a radar sensor, a light sensor (e.g., a line scan camera or matrix camera), a sound sensor (e.g., an acoustic sensor), and combinations thereof.

[0038] Radar sensors are particularly preferred sensors in category 2. They have the advantage of providing very precise information about the properties of specific beverage containers, thus enabling fine and precise differentiation between different beverage containers.

[0039] Light sensors, such as line scan cameras or matrix cameras, are also particularly preferred sensors in the second category. They provide a value y and a value z over a time x in the matrix y / z, where the value y represents, for example, a specific light intensity (brightness) and the value z represents, for example, a specific light frequency (color). With ToF sensors, a spatial distance from the sensor can be detected as an additional dimension. Like radar sensors, light sensors are suitable for providing information about the properties of certain beverage containers.

[0040] Sound sensors (such as acoustic sensors) are also particularly preferred sensors of the 2nd category. Sound sensors are suitable, for example, for detecting grinding noises from a grinder (e.g. for coffee beans) of the beverage maker and preventing the grinder from running dry. For this function, the sound sensor is preferably mounted near a grinder of the beverage maker. The advantage of using such a sound sensor as a 2nd category sensor is that, via the sound curve recorded by this sensor, a sound pattern shortly before the grinder runs dry can be used to indicate to the user that the grinder is about to run dry. By predicting that the grinder is about to run dry, a user can be prompted to refill a bean container and thus prevent the bean container from emptying while a beverage (e.g. coffee) is being dispensed. This means that the provision of a beverage (e.g.coffee) can be carried out without interruption and there is no downtime at the beverage maker. Furthermore, the problem is known that once the grinder has run out of ground beans, it requires a certain amount of lead time (i.e. a certain amount of grinding time) to refill the bean container with (ground) beans. On the one hand, this extends the time until a beverage is ready after a beverage dispensing has been initiated. On the other hand, there are fluctuations in the dosage with regard to the amount of ground beans used to prepare the beverage and thus a certain variability in the composition of the beverage dispensed after the grinder has run out of idle. The use of the sound sensor means that the grinder can be prevented from running out of grounds right from the start. As a result, the beverage maker can always provide a beverage with a consistent product composition without any downtime and without any delays in dispensing.

[0041] In a preferred embodiment, the at least one specific event (which can be triggered at the beverage maker) is selected from the group consisting of i) Interaction of the beverage maker with a user, preferably selected from the group consisting of issuing a request from the beverage maker to the user, displaying a specific beverage selection, hiding a specific beverage selection and combinations thereof, wherein the request is particularly preferably selected from the group consisting of refilling coffee beans, emptying the grounds container, filling the water tank, filling the milk tank, starting a maintenance program (e.g.Starting a cleaning program, starting a descaling program and / or starting a rinsing program) and combinations thereof; ii) maintenance of the beverage maker, preferably selected from the group consisting of detecting and / or evaluating errors in processes of the beverage maker, detecting and / or evaluating errors in components of the beverage maker, determining a service interval of the beverage maker, triggering a service call of the beverage maker, setting and / or adjusting a setting of the beverage maker; iii) controlling at least one component of the beverage maker, preferably dispensing a specific beverage via the beverage maker, moving a beverage outlet of the beverage maker, carrying out a maintenance program of the beverage maker and combinations thereof; and iv) combinations thereof.

[0042] Furthermore, according to the invention, a method for operating a beverage maker is provided, comprising the steps a) Recording several measured values ​​S2 from at least one sensor of a 2nd category, where a sensor of the 2nd category is understood to be a sensor that delivers several measured values ​​S2 at a specific time; b) Comparison of the measured values ​​S2 with target values ​​for the measured values ​​S2, which are stored in a data memory of the beverage maker and code for at least one specific event, by a control unit of the beverage maker; c) Automatic triggering or not triggering of at least one event in the beverage maker based on this comparison; and d) Recording at least one measured value S1 from at least one sensor of a 1st category if no event is triggered, where a sensor of the 1st category is understood to be a sensor that delivers only a single measured value S1 at a specific time; wherein, in the case where no event is triggered after the comparison with target values ​​for the measured values ​​S2, at least one measured value S1 from at least one sensor of a 1st category of the beverage maker is used by the control unit to assign at least one event to the measured values ​​S2 which do not trigger an event after a comparison with target values ​​for the measured values ​​S2, and these measured values ​​S2 are stored as target values ​​for the at least one event on the data memory, characterized in that, via the beverage maker, measured values ​​S2 of the at least one sensor of the 2nd category are assigned to certain, regularly occurring events based on a confirmation via the at least one sensor of the 1st category by a user, wherein the assignment is carried out according to the principle of machine learning, and wherein the storage of the measured values ​​S2 for the events is carried out automatically on the data memory.

[0043] The beverage maker used in the method according to the invention can include a user interface, preferably a touchscreen. A triggered event is preferably displayed via the user interface. Furthermore, a countdown until the implementation of a triggered event can be displayed via the user interface. In addition, a (specific) beverage selection can be displayed via the user interface. Furthermore, information about measured values ​​S2 of the at least one sensor of a second category can be displayed via the user interface, preferably information selected from the group consisting of the height of a beverage container, the fill level of a beverage container, the fill level of a bean container of the beverage maker, the empty grinding of a grinder of the beverage maker, and combinations thereof. Furthermore, target values ​​for S2 can be displayed via the user interface and can be confirmed and / or deleted at the user's request.

[0044] In a preferred embodiment of the method, the at least one event encoded in the target values ​​for the measured values ​​S2 is triggered if the measured values ​​S2 can be assigned to target values ​​for the measured values ​​S2 with a probability of >50%, preferably ≥60%, particularly preferably ≥70%, very particularly preferably ≥80%, in particular ≥90%.

[0045] In the method according to the invention, for example, the at least one measured value S1 can be recorded and used if the measured values ​​S2 can be assigned to target values ​​for the measured values ​​S2 with a probability of only ≤50%, preferably ≤40%, particularly preferably ≤30%, very particularly preferably ≤20%, in particular ≤10%.

[0046] In a preferred embodiment of the method, the at least one sensor of the first category provides a one-dimensional measured value S1 at a specific time. The one-dimensional measured value can be selected from the group consisting of temperature, pressure, speed, flow velocity, electrical capacitance, direction of movement, distance, time, and mass.

[0047] The Category 1 sensor used in the method may have the same properties as the Category 1 sensor of the beverage maker described above.

[0048] It is further preferred if at least one sensor of the 2nd category is used in the method, which delivers multi-dimensional measured values ​​S2 at a specific point in time, preferably at least three-dimensional measured values ​​S2, in particular at least four-dimensional measured values ​​S2. One dimension of the multi-dimensional measured value can be selected from the group consisting of frequency of an oscillation (e.g. wavelength of light, ie light color), amplitude of an oscillation (e.g. amplitude of light, ie light intensity), polarization of an oscillation (e.g. polarization of light), direction of propagation of an oscillation (e.g. direction of propagation of light), direction of propagation of lines of force (e.g. direction of propagation of magnetic field lines), strength of lines of force (e.g. strength of magnetic field lines). A further dimension of the multi-dimensional measured value can likewise be selected from the above-mentioned group.

[0049] The second category sensor used in the method may have the same properties as the second category sensor of the beverage maker described above.

[0050] The method may be characterized in that the at least one specific event is selected from the group consisting of i) Interaction of the beverage maker with a user, preferably selected from the group consisting of issuing a request from the beverage maker to the user, displaying a specific beverage selection, hiding a specific beverage selection, and combinations thereof, wherein the request is particularly preferably selected from the group consisting of refilling coffee beans, emptying the grounds container, filling the water tank, filling the milk tank, starting a maintenance program, and combinations thereof; ii) Maintenance of the beverage maker, preferably selected from the group consisting of detecting and / or evaluating errors in processes of the beverage maker, detecting and / or evaluating errors in components of the beverage maker, determining a service interval of the beverage maker, triggering a service call of the beverage maker, setting and / or adjusting a setting of the beverage maker;iii) controlling at least one component of the beverage maker, preferably dispensing a specific beverage via the beverage maker, moving a beverage outlet of the beverage maker, performing a maintenance program of the beverage maker, and combinations thereof; and iv) combinations thereof.

[0051] The subject matter of the invention will be explained in more detail with reference to the following figures and examples, without wishing to restrict it to the specific embodiments shown here.

[0052] Figure 1shows the functional mechanism of a beverage maker according to the invention. If an object to be analyzed is detected by the sensor of the 2nd category 2, a plurality of measured values ​​4 (a so-called pattern) are sent from the sensor of the 2nd category 2 to the control unit 6 of the beverage maker. The control unit 6 accesses the data memory 5 and checks whether the detected measured values ​​4 are already known with a certain probability, i.e. whether they are stored in the data memory 5 and assigned there to a specific event. If this question 7 can be answered in the affirmative, the beverage maker triggers a specific event 8. If this question is answered in the negative, the beverage maker waits for a single measured value 3 from the sensor of the 1st category 1. After receiving the measured value 3 from the sensor of the 1st category 1, the measured values ​​4 detected by the sensor of the 2nd category 2 are assigned to at least one specific event 8 in the beverage maker.On the data storage 5, no pattern 4 is originally associated with an event 8. A measured value 3 from the sensor of the 1st category 1 is then used to connect a specific event 8 with the two measured values ​​4 from the sensor of the 2nd category 2 and to store this connection on the data storage 5 (= machine learning).

[0053] Figure 2 shows the functional mechanism of a beverage maker according to the invention with a radar sensor 2 for detecting several measured values ​​12, 12' from a container 10, 10', which is placed by a user under a beverage dispensing unit 11 of the beverage maker.

[0054] In Figure 2A the vessel is a latte macchiato glass 10, which generates a frequency spectrum 12 characteristic of this glass 10 from the radar sensor 2. In the case of Figure 2AThis characteristic frequency spectrum 12 is already stored in the data memory 5 of the beverage maker (see √ symbol), so that the beverage maker recognizes the latte macchiato glass 10 as such and, without further input from the user (e.g., via interaction with a sensor of the 1st category 1), carries out an event linked to this glass 10 (e.g., dispensing a latte macchiato via the beverage dispensing unit 11 of the beverage maker or displaying or highlighting a specific beverage selection). If the event is a display or highlighting of a specific beverage selection (e.g., an enlarged representation of specific beverages on a user interface of the beverage maker), the user can specifically select and dispense a specific beverage from the beverage selection by pressing a button (actuating a sensor of the 1st category) or by using a gesture, a facial expression, or an acoustic announcement (actuating a sensor of the 2nd category).

[0055] The same applies to the Figure 2B situation described. In Figure 2B Instead of the latte macchiato glass 10, an espresso cup 10' is placed under the beverage dispensing unit 11 of the beverage maker. Since the characteristic frequency spectrum 12' detected by the radar sensor 2 for the espresso cup 10' is already stored in the data memory 5 (see √ symbol), the beverage maker recognizes the espresso cup 10' as such and, without further input from the user (e.g., via interaction with a sensor of the 1st category 1), performs an event linked to this cup 10' (e.g., dispensing an espresso via the beverage dispensing unit 11 of the beverage maker).

[0056] In Figure 2CIn contrast, a situation is shown in which the pattern 12 detected by the radar sensor 2 for the latte macchiato glass 10 is not yet stored in the data storage 5 of the beverage maker or bears no similarity to patterns stored there (see X symbol). This causes the beverage maker to use a measured value detected by the sensor of category 1 (here: a touchscreen 1 on the beverage maker) (here: an impedance change that changes when a user touches a latte macchiato glass 10 displayed on the touchscreen) to assign the detected pattern 12 to the event "dispensing latte macchiato" and to save this assignment in the data storage 5.

[0057] Figure 3shows an example of multiple measured values ​​from a Category 2 sensor. This Category 2 sensor is a two-dimensional sensor that records the intensity (y-axis = 1st dimension) of electromagnetic radiation of different frequencies I, II, and III (z-axis = 2nd dimension) over time (x-axis). Example 1 - Beverage maker with electromagnetic sensor for cup detection

[0058] The beverage maker is a coffee machine with a Category 2 sensor built in. This sensor is either a radar sensor (e.g., a 60 GHz radar sensor from Infineon Technologies AG, sensitivity: sub-mm range) or an image sensor.

[0059] First, a glass or cup is placed under the beverage maker. In the case of a radar sensor, the first measured value is the frequency of the detected radar radiation, and the second measured value is the amplitude of the detected radar radiation. In the case of an image sensor, the frequency of the detected light (the color of the light) and the amplitude of the detected light (the brightness of the light) are recorded.

[0060] Since no target values ​​are specified in the permanent memory of the beverage maker, there is no assignment of the measured values ​​(i.e. the pattern) to a specific event.

[0061] This beverage maker features a Category 1 sensor, such as a touchpad. The operator input interface displays various beverages that can be dispensed by the beverage maker.

[0062] If the user has placed a latte macchiato glass under the beverage maker because they want to have one prepared, they can now select the latte macchiato dispensing option by pressing the touchpad. The signal from the category 1 sensor (touchpad) is then used to link the signal recorded by the radar sensor or the image sensor with the latte macchiato dispensing option and save this link to the permanent data memory. The next time the same latte macchiato glass is placed under the beverage maker, the latte macchiato will be dispensed without confirmation via the touchpad, because the beverage maker has assigned the pattern of the specific latte macchiato glass recorded by the category 2 sensor to the latte macchiato dispensing option.

[0063] Since the shape of latte macchiato glasses can vary within a series or between different manufacturers, the assignment becomes better the more often a latte macchiato glass is placed under the beverage maker (increase in shape information). Even with repeated use of a particular glass, accuracy increases, as even category 2 sensors can exhibit measurement errors, the significance of which decreases with each further measurement repetition. At some point, operator input via the touchpad is no longer necessary, as the machine clearly assigns the recorded measured values ​​to a latte macchiato glass. As a result, the user interface of the beverage maker (e.g., display or touchscreen) can only display the beverage that, based on a comparison with the data memory, matched the last dispenses with the same pattern S2 and the corresponding button press of S1 (here: a latte macchiato).The user then has the choice of selecting the drink (here: only a latte macchiato displayed on the user interface) or, if necessary, canceling with another command.

[0064] Of course, the beverage maker's control unit can be configured so that the beverage maker offers the user a drink selection after the cup is placed under the glass and / or locks certain drinks. If no alternative drinks are available for a latte macchiato glass, the control unit can be configured to automatically start dispensing the latte macchiato. This can be done, for example, by displaying "Drink will start automatically in X seconds." To cancel, press Y," where X stands for seconds and Y represents a button to be pressed on the beverage maker.

[0065] For example, a radar sensor functions as a Category 2 sensor in such a way that a portion of the radar waves is reflected directly from the surface of the object being examined (e.g., a coffee cup), while another portion penetrates the object before being reflected, slows down due to the object's higher density compared to the surrounding air, and then exits the object again. The amount of reflected rays and the time-of-flight differences between the reflected rays and the emitted rays are characteristic of different objects and materials.

[0066] Furthermore, additional sensors in the beverage maker can collect further information about the glass or cup and assign it to the sensor signals. The following sensors are possible: Collision detection sensor to determine the height of the vessel; flow sensor to protect against vessel overflow; pressure sensor to interrupt the beverage dispensing and adjust the volume dispensing to the user's habits. Example 2 - Beverage maker with acoustic sensor for bean empty notification

[0067] This beverage maker is a coffee maker and contains a Category 2 acoustic sensor. The acoustic sensor "listens" to various events or errors. For example, the grinder of a coffee machine emits a characteristic noise when grinding coffee beans. This noise can be recorded by the acoustic sensor and linked to the event that the grinder should stop the grinding process. If the acoustic sensor detects the characteristic noise again after this "learning process," the grinder can be stopped before the grinder runs out of coffee beans.

[0068] In the current state of the art, for example, only a current sensor provides information about whether a grinder motor is running idle, i.e., no longer grinding coffee beans. In this case, the level of the electrical current detected by the current sensor decreases. The noise pattern emitted by the grinder can then be recorded and, for example, learned as early as 2 to 3 seconds before the current limit is reached. This measure allows the beverage maker to detect a condition shortly before the grinder "runs empty" and can shut it down in time (before detecting the lower current value). The advantage of this is that product preparation is not interrupted by refilling beans: When the grinder is running dry, the grinder and grinder outlet are empty, and after filling with coffee beans, a certain amount of pre-running time (grinding) is required to fill the system with beans and ground coffee. Only then is the coffee grounds correctly fed into the brewing chamber.This results in a loss of time and fluctuations in the dosage of the ground coffee.

[0069] The sound of refilling beans could also be detected and a specific event could be associated with this process. List of reference symbols

[0070] 1:Sensor of the 1st category (e.g. touch display); 2:Sensor of the 2nd category (e.g. radar sensor); 3:Single measured value S1 of the sensor of the 1st category (e.g. a discrete pressure value of a pressure sensor); 4:Several measured values ​​S2 of the sensor of the 2nd category (e.g. spatially resolved discrete frequencies and discrete intensities of electromagnetic radiation); 5:Data memory; 6:Control unit; 7:Determination of whether measured values ​​4 are present in data memory 5; 8:Event triggered by the beverage maker; 9:Assignment of measured values ​​4 to at least one event; 10:Latte macchiato glass; 10':Espresso cup; 11:Drink dispensing unit of the beverage maker; 12:Frequency spectrum of a latte macchiato glass 10; 12':Frequency spectrum of an espresso cup 10'; I:1. Frequency of electromagnetic radiation; II: 1. Frequency of electromagnetic radiation; III: 3. Frequency of electromagnetic radiation;

Claims

1. A beverage maker, comprising a) at least one sensor of a 1st category (1), wherein a sensor of the 1st category (1) is understood to mean a sensor that provides only a single measured value S1 (3) at a certain point in time; b) at least one sensor of a 2nd category (2), wherein a sensor of the 2nd category (2) is understood to mean a sensor that provides several measured values S2 (4) at a certain point in time; c) a data storage (5); d) a control unit (6), wherein the control unit (6) is configured to compare the measured values S2 (4) with target values for the measured values S2, which are stored on the data storage (5) and encode for at least one specific event (8), and, on the basis of this comparison, to automatically trigger or not trigger at least one event (8) in the beverage maker; wherein the control unit (6) of the beverage maker is further configured to detect and use at least one measured value S1 (3) to assign certain measured values S2 (4), which do not trigger an event (8) after being compared with target values for the measured values S2, to at least one event (8), and to store these measured values S2 (4) on the data storage (5) as target values for the at least one event (8), characterised in that the beverage maker is further configured to assign measured values S2 (4) of the at least one sensor of the 2nd category (2), on the basis of a user's confirmation via the at least one sensor of the 1st category (1), to certain regularly occurring events (8), wherein the assignment is performed according to machine learning principles, and wherein the measured values S2 (4) for the events (8) are automatically stored on the data storage (5).

2. A beverage maker according to the preceding claim, characterised in that the beverage maker includes an operator interface, preferably a touch screen, the operator interface preferably being configured to i) display a triggered event (8); and / or ii) to display a countdown until the implementation of a triggered event (8); and / or iii) to display a selection of beverages; iv) to display information on measured values S2 (4) of at least one sensor of a 2nd category (2), preferably information selected from the group consisting of a height of a beverage container, a fill level of a beverage container, a fill level of a beverage maker's bean container, an empty running of a beverage maker's grinder and combinations thereof; and / or v) to display target values for S2 and, at a user's request, confirm and / or delete them.

3. Beverage maker according to any one of the preceding claims, characterised in that the control unit (6) is configured to trigger the at least one event (8) encoded in the target values for the measured values S2 if the measured values S2 (4) can be assigned to target values for the measured values S2 (4) with a probability of respectively >50%, preferably ≥60%, particularly preferably ≥70%, very particularly preferably ≥80%, especially ≥90%.

4. Beverage maker according to any one of the preceding claims, characterised in that the control unit (6) is configured to use at least one measured value S1 (3) if the measured values S2 (4) can be assigned to target values for the measured values S2 with a probability of respectively only ≤50%, preferably ≤40%, particularly preferably ≤30%, very particularly preferably ≤20%, in particular ≤10%.

5. Beverage maker according to any one of the preceding claims, characterised in that the at least one sensor of the 1st category (1) is configured to provide only a one-dimensional measured value S1 (3) at a certain point in time.

6. Beverage maker according to any one of the preceding claims, characterised in that the at least one sensor of the 2nd category (2) is configured to provide multi-dimensional measured values S2 (4) at a certain point in time, preferably at least three-dimensional measured values S2 (4), particularly preferably at least four-dimensional measured values S2 (4).

7. Beverage maker according to any of the preceding claims, characterised in that the at least one certain event (8) is selected from the group consisting of an interaction between the beverage maker and a user, a beverage maker maintenance, a control of at least one of the beverage maker's components and combinations thereof.

8. A method of operating a beverage maker, said method comprising the steps a) detecting several measured values S2 from at least one sensor of a 2nd category (2), wherein a sensor of the 2nd category (2) is understood to mean a sensor that provides several measured values S2 (4) at a certain point in time; b) comparing the measured values S2 (4) with target values for the measured values S2, which are stored on a data storage (5) of the beverage maker and encode for at least one specific event (8), by a control unit (6) of the beverage maker; c) automatically triggering or not triggering at least one event (8) in the beverage maker on the basis of said comparison; and d) detecting at least one measured value S1 (3) from at least one sensor of a 1st category (1) if no event (8) is triggered, wherein a sensor of the 1st category (1) is understood to mean a sensor that provides only a single measured value S1 (3) at a certain point in time; wherein, in a case in which no event (8) is triggered after the comparison with target values for the measured values S2, at least one measured value S1 (3) from at least one sensor of a 1st category (1) of the beverage maker is used by the control unit (6) to assign at least one event (8) to the measured values S2 which do not trigger an event (8) after a comparison with target values for the measured values S2, and these measured values S2 (4) are stored on the data storage (5) as target values for the at least one event (8), characterised in that, via the beverage maker, measured values S2 (4) of the at least one sensor of the 2nd category (2), on the basis of a user's confirmation via the at least one sensor of the 1st category (1), are assigned to certain regularly occurring events (8), whereby the assignment is based on the machine learning principles, and wherein the measured values S2 (4) for the events (8) are automatically stored on the data storage (5).

9. A method according to claim 8, characterised in that the beverage maker includes an operator interface, preferably a touch screen, via which the operator interface preferably i) displays a triggered event (8); and / or ii) displays a countdown until an implementation of a triggered event (8); and / or iii) displays a selection of beverages; iv) displays information on measured values S2 (4) of at least one sensor of a 2nd category (2), preferably information selected from the group consisting of a height of a beverage container, a fill level of a beverage container, a fill level of a beverage maker's bean container, an emtpy running of a beverage maker's grinder and combinations thereof; and / or v) displays target values for S2 and, at the user's request, confirm and / or delete them.

10. A method according to claim 8 or 9, characterised in that the at least one event (8) encoded in the target values for the measured values S2 is triggered if the measured values S2 (4) can be assigned to target values for the measured values S2 (4) with a probability of respectively >50%, preferably ≥60%, particularly preferably ≥70%, very particularly preferably ≥80%, in particular ≥90%.

11. A method according to one of the claims 8 to 10, characterised in that at least one measured value S1 (3) is detected and used if the measured values S2 (4) can be assigned to target values for the measured values S2 with a probability of respectively only ≤50%, preferably ≤40%, particularly preferably ≤30%, very particularly preferably ≤20%, in particular ≤10%.

12. A method according to any one of claims 8 to 11, characterised in that the at least one sensor of the 1st category (1) provides a one-dimensional measured value S1 (3) at a certain point in time.

13. A method according to any one of claims 8 to 12, characterised in that the at least one sensor of the 2nd category (2) provides multi-dimensional measurement values S2 (4) at a certain point in time, preferably at least three-dimensional measurement values S2 (4), in particular at least four-dimensional measurement values S2 (4).

14. A method according to any one of claims 8 to 13, characterised in that the at least one certain event (8) is selected from the group consisting of an interaction between the beverage maker and a user, a maintenance of the beverage maker, a control of at least one component of the beverage maker, and combinations thereof.