Method and control unit for locking user interface of household appliance

Abstract

A control unit (101) for controlling a user interface (110) of a household appliance (100) is described, where the user interface (110) comprises a set of control elements (111). The control unit (101) is configured to be able to detect a sequence (200) of actuations (201) of the set of control elements (111) and to classify the detected sequence (200) of actuations (201) as a normal sequence or an abnormal sequence. Furthermore, the control unit (101) is configured to be able to cause the user interface (110) to be locked if the sequence (200) of detected actuations (201) is classified as an abnormal sequence.

Description

Technical Field

[0001] The present invention relates to a method and a corresponding control unit, which are intended to control a user interface of a household appliance. Background Art

[0002] Household appliances such as washing machines, dishwashers or refrigerators usually comprise a user interface having one or more control elements which can be actuated by a user of the appliance for setting parameters of the appliance, for selecting an operating program of the appliance and / or for starting operation of a selected operating program of the appliance.

[0003] The user interface may display a parental lock function that can be activated by the user to lock one or more control elements of the user interface, thereby protecting the user interface from user input by children or animals. A dedicated control element may be used to activate the child lock function, wherein providing a dedicated control element generally increases the cost of the user interface. Furthermore, it may happen that the user forgets to activate the child lock function, thereby failing to protect the appliance from user input caused by a child. Summary of the invention

[0004] The technical problem of improving the efficiency and comfort of use of a child lock function of a user interface of a household appliance is solved herein. The technical problem is solved by each independent claim. Preferred examples are described in the dependent claims.

[0005] According to one aspect, a control unit for controlling a user interface of a household appliance (e.g., a freezer, a dishwasher, an oven, a washing machine, a dryer, etc.) is described, wherein the user interface includes a set of control elements (e.g., one or more buttons, one or more knobs, one or more sliders, etc.). The user interface may include a touch screen with one or more virtual control elements.

[0006] The control unit is configured to detect a sequence of actuations of the set of control elements. In particular, one or more actuations of one or more different control elements may be detected in a sequential manner. The sequence of actuations may indicate the one or more control elements that have been actuated, and the order in which the one or more control elements have been actuated. The sequence of actuations may include multiple actuations of the same control element. Alternatively or additionally, the sequence of actuations may include actuations of different control elements.

[0007] The control unit is also configured to be able to classify the detected actuation sequence as a normal sequence or an abnormal sequence. A normal sequence may be a sequence that is caused by a user with a relatively high probability (e.g., 5% or higher) in the operating environment of the household appliance. On the other hand, an abnormal sequence may be a sequence that is caused by a user with a relatively low probability (e.g., less than 5%) in the operating environment of the household appliance.

[0008] The control unit may be configured to classify the detected sequence of actuations using a classification unit, wherein the classification unit may have been pre-trained using training data indicating sequences of normal and / or abnormal actuations. The training data may have been collected during (normal) operation of the household appliance (and / or household appliances of the same type). The training data may have been collected in a backend server of the household appliance (possibly including training data from a plurality of different household appliances of the same type and / or model).

[0009] In many cases, it may not be necessary to identify a specific user based on the sequence of detected actuations. It is sufficient to determine whether the sequence of detected actuations is normal or abnormal. Typically, the actuations that are frequently performed will depend on the type and / or model of household appliances. Many adults will use household appliances of a certain type and / or model in a very similar way. Typically, their input and / or use behavior will not change significantly due to individual or personal characteristics. On the contrary, each adult who wants to input a reasonable command to a household appliance may act in a very similar way. Therefore, for the child lock function, it may be sufficient to collect the sequence of detected actuations of field household appliances (e.g., household appliances that have been sold and are now located at the end user) in the central data lake of the household appliance manufacturer. Therefore, for example, the sequence of actuations of a user of a certain household appliance detected can be transmitted to the server or cloud service of the household appliance manufacturer. The data lake stored on the server or in the cloud service can be analyzed by the type and / or model of the household appliance to find out the normal use behavior of this type and / or model of household appliances. Based on this knowledge, it is easy to distinguish between normal and random behavior of children, where the child has no intention to deliberately control the household appliance, but touches the user interface of the household appliance at will. Therefore, in some preferred embodiments, the control unit is configured to be able to classify the sequence of detected actuations using a classification unit, wherein the classification unit has been pre-trained with training data indicating a sequence of normal and / or abnormal actuations, wherein the training data has been collected in a back-end server from a plurality of different household appliances of the same type and / or model as the household appliance. The back-end server can be understood as a server or cloud service or any other information technology infrastructure, which can be controlled by the manufacturer of the household appliance. This method can allow the use of swarm intelligence and / or collaborative filtering to help distinguish normal and abnormal usage behaviors of a certain type and / or model of household appliances.

[0010] Thus, the (machine-learned) classification unit may be used to determine whether a detected sequence of actuations is a normal sequence (which is typically used for normal operation of a household appliance) or an abnormal sequence (which is typically not used for normal operation of a household appliance). The classification unit may for example comprise a trained artificial neural network.

[0011] The classifier may have been trained for a specific user or household (based on training data for the specific user and / or household). Thus, user and / or household specific classifiers may be used, thereby improving the quality of the classification. Alternatively or additionally, the classifier may have been trained for a specific type or model of household appliances. Thus, model and / or type specific classifiers may be used, thereby improving the quality of the classification.

[0012] Furthermore, the control unit is configured to enable the user interface to be locked if the detected sequence of actuations is classified as an abnormal sequence. If the detected sequence of actuations is classified as an abnormal sequence, the user interface may be automatically locked (without any further user intervention). As a result of locking the user interface, the touch screen of the user interface may be blackened and / or turned off. Alternatively or additionally, any further actuation of one or more control elements of the user interface may be ignored.

[0013] On the other hand, if the detected sequence of actuations is classified as a normal sequence, the control unit may be configured to operate the user interface and / or the household appliance in accordance with the detected sequence of actuations.

[0014] Thus, a control unit is described which is configured to monitor the actuation of control elements caused by a user of a household appliance in order to detect an abnormal sequence of actuations (possibly caused by a child or an animal) and to (automatically) place the user interface in a locked state if an abnormal sequence of actuations is detected. Thus, a reliable and comfortable locking function is provided. Furthermore, stress on the user interface caused by abnormal use can be reduced, thereby increasing the service life of the user interface.

[0015] The control unit can be configured to repeatedly append the detected sequence of actuations with the newly detected actuation (of the control element). In addition, the control unit can be configured to classify the appended sequence of actuations as a normal sequence or an abnormal sequence, and if the appended sequence of actuations is classified as an abnormal sequence, the user interface is locked.

[0016] In particular, initially, a sequence comprising only a single actuation of a control element may be detected and classified. When one or more further actuations of one or more control elements are caused, the sequence may be appended with one or more newly detected actuations. As the sequence of detected actuations becomes longer and longer, the likelihood that the detected sequence is an abnormal sequence may increase (e.g., once the sequence comprises 2 or more, or 3 or more, actuations). Once the sequence of detected actuations is classified as an abnormal sequence, the user interface may be locked (thereby ignoring any further actuations of one or more control elements of the user interface).

[0017] Therefore, the control unit can be configured to repeatedly analyze the sequence of actuations because the sequence of actuations is caused by the user. Once the sequence is detected as an abnormal sequence, the user interface can be automatically locked. Therefore, the quality and speed of the automatic locking function can be improved.

[0018] The control unit may be configured to determine one or more features of the transition between two continuous actuations of a sequence of one or more actuations and / or detected actuations. The one or more features may include the duration of the actuation, the intensity of the actuation, and / or the time interval of the transition between two continuous actuations. The control unit may be configured to classify the sequence of the actuation detected as a normal sequence or an abnormal sequence according to the one or more features. By considering one or more features of the transition between a single actuation and / or the actuation, the reliability of the classification may be improved, thereby improving the quality of the automatic locking function.

[0019] The control unit can be configured to be able to compare the detected sequence of actuations with a set of reference sequences (e.g., with 10 or more than ten, or 50 or more than fifty, or 100 or more than one hundred reference sequences). The set of reference sequences may include sequences of actuations that are each considered to be normal sequences. The set of reference sequences may have been predefined based on training data (which indicates sequences of actuations that have been caused during normal operation of the household appliance and / or household appliances of the same type). The set of reference sequences may have been defined by a human expert. The set of reference sequences may be stored on a back-end server and / or a storage unit of the household appliance.

[0020] Based on the comparison with the set of reference sequences, the detected sequence of actuations can be classified as a normal sequence or an abnormal sequence. In particular, if the detected sequence of actuations corresponds to a reference sequence included in the set of reference sequences, the detected sequence of actuations can be classified as a normal sequence. Alternatively or additionally, if the detected sequence of actuations does not correspond to any of the reference sequences included in the set of reference sequences, the detected sequence of actuations can be classified as an abnormal sequence. By utilizing a set of reference sequences, the robustness and reliability of the classification and automatic locking functions can be improved.

[0021] The set of reference sequences may be ordered according to the total number of actuations contained in the respective reference sequence. In other words, the set of reference sequences may include a plurality of subsets of reference sequences corresponding to a plurality of different total numbers of actuations, wherein each subset includes reference sequences having a particular total number of actuations. For example, subsets of reference sequences may be provided for a total number of 1, 2, 3, 4, or more than four actuations.

[0022] The control unit may be configured to determine the total number of actuations contained in the detected sequence of actuations. Furthermore, the control unit may be configured to identify a subset of reference sequences having the same total number of actuations as the detected sequence of actuations. The detected sequence of actuations may then be compared with one or more reference sequences (exclusively) contained in the identified subset of reference sequences in order to classify the detected sequence of actuations as a normal sequence or an abnormal sequence. Thus, the computational complexity of the automatic locking function may be reduced.

[0023] The control unit can be configured to determine the abnormal score of the sequence of the detected actuation based on a predetermined probability model of the actuation of a set of control elements of the user interface. For each of a set of previous actuations, the predetermined probability model can indicate the probability value of each of a plurality of different subsequent actuations after the corresponding previous actuation. For example, the probability model can include a matrix, wherein one dimension (e.g., different rows) indicates a plurality of different previous actuations (of the corresponding plurality of control elements), and another dimension (e.g., different columns) indicates a plurality of different subsequent actuations (of the corresponding plurality of control elements). The matrix elements of a specific previous actuation and a specific subsequent actuation can indicate the transition probability of a situation where a specific subsequent actuation (directly) follows a specific previous actuation. The probability model can have been determined based on training data.

[0024] The control unit may be configured to be able to classify the sequence of detected actuations as a normal sequence or an abnormal sequence according to the anomaly score. In particular, the control unit may be configured to be able to select an actuation (as a previous actuation) from the sequence of detected actuations. In addition, a subsequent actuation that (directly) follows the selected actuation can be determined from the sequence of detected actuations, and a probability model can be used to determine the probability value of the determined subsequent actuation (and the selected previous actuation). The anomaly score of the sequence of detected actuations can then be determined in an accurate manner based on the determined probability value.

[0025] The control unit can be configured to determine the abnormal score of the detected sequence of actuations based on a predetermined abnormal index, wherein the abnormal index can indicate the probability that the sequence of actuations including actuations with corresponding index numbers is an abnormal sequence for each of the multiple index numbers of actuations in the sequence of actuations. The index number can be an integer starting from 1 (for the first actuation), 2 (for the second actuation), 3 (for the third actuation), etc. The abnormal index will usually increase with the increase of the index number (especially when the index number is relatively high, such as greater than 2). The abnormal index may have been determined based on training data of actuations that have been caused during the operation of the household appliance. By considering the abnormal index, the accuracy of the classification can be further improved.

[0026] The control unit may be configured to iteratively update the anomaly score when the sequence of detected actuations is appended with one or more newly detected actuations. Then, based on the updated anomaly score, the sequence of appended actuations may be classified as a normal sequence or an abnormal sequence.

[0027] The control unit may be configured to use, for example, the iterative formula s n+1 =s n +s to update the anomaly score, where s n is the anomaly score of n actuations of the sequence of detected actuations, where s is the score of newly detected actuations, where the score s depends on the probability model, and where s n+1 is an updated anomaly score for the sequence of appended actuations, which includes n actuations and the newly detected actuation. The score s of the newly detected actuation can be determined based on a probability value of a transition from (the last actuation of) the sequence of detected actuations to the newly detected actuation. In addition, the score s of the newly detected actuation can be determined based on an anomaly index of an exponential number of the newly detected actuations within the sequence of appended actuations.

[0028] The control unit can be configured to compare the (updated) anomaly score with a predetermined anomaly threshold (which may have been determined based on the training data). According to the comparison result, the attached and / or detected actuated sequence can then be classified as a normal sequence or an abnormal sequence. In particular, if the (updated) anomaly score is less than the abnormal threshold, the attached and / or detected actuated sequence can be classified as a normal sequence. On the other hand, if the (updated) anomaly score is greater than the abnormal threshold, the attached and / or detected actuated sequence can be classified as an abnormal sequence.

[0029] The above-described iterative scheme for updating the anomaly score enables detecting anomalous sequences in a particularly fast manner, thereby improving the quality of the automatic locking function.

[0030] According to another aspect, a household appliance is described, comprising a control unit as described herein. Furthermore, the appliance comprises a user interface having a set of control elements (eg, 2 or more, or 3 or more, or 4 or more than four control elements).

[0031] According to another aspect, a method for controlling a user interface of a household appliance (e.g., a dishwasher, an oven, a refrigerator, a washing machine, a dryer, etc.) is described, wherein the user interface includes a set of control elements. The method includes detecting a sequence of actuations of the set of control elements (wherein each actuation generally corresponds to an actuation of a single control element).

[0032] In addition, the method includes classifying the detected sequence of actuations as a normal sequence or an abnormal sequence (e.g., using a machine learning classification unit). In addition, the method includes locking the user interface if the detected sequence of actuations is classified as an abnormal sequence.

[0033] According to another aspect, a software program is described. The software program may be adapted to be executed on a processor and, when executed on the processor, is used to perform the method steps outlined herein.

[0034] According to another aspect, a storage medium is described. The storage medium may comprise a software program adapted to be executed on a processor and, when executed on the processor, for performing the method steps outlined herein.

[0035] According to another aspect, a computer program product is described. The computer program may comprise executable instructions for performing the method steps outlined herein when executed on a computer.

[0036] It should be noted that the methods and systems outlined herein, including preferred embodiments thereof, may be used alone or in combination with other methods and systems disclosed herein. Furthermore, features outlined in the context of the systems also apply to the corresponding methods. Furthermore, all aspects of the methods and systems outlined herein may be combined arbitrarily. In particular, the features of the claims may be combined with each other in any manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The present invention is explained below in an exemplary manner with reference to the accompanying drawings, in which:

[0038] Figure 1 An example of a household appliance is shown;

[0039] Figure 2a An example of a sequence of actuation of a set of control elements of a household appliance is shown;

[0040] Figure 2b shows the evolution of the anomaly score for one example of a sequence of actuations; and

[0041] Figure 3 A flow chart illustrating an example method for controlling a user interface of a home appliance. DETAILED DESCRIPTION

[0042] As mentioned above, the present invention aims to improve the efficiency, reliability and / or comfort of the locking function of the user interface of the household appliance. In this article, Figure 1An exemplary household appliance 100, in particular a washing machine or a dryer, is shown, which comprises a chamber 103, such as a drum, for treating items, such as clothes or food. The treatment can be performed by one or more actuators 102, such as a heating unit or a motor, wherein the one or more actuators 102 are controlled by a control unit 101 of the appliance 100.

[0043] The appliance 100 includes a user interface 110, wherein the user interface 110 may include a display 112 for providing visual feedback to a user of the appliance 100, and / or one or more control elements 111, such as one or more buttons, one or more virtual buttons on a touch screen, knobs, sliders, etc.

[0044] The control unit 101 can be configured to detect one or more actuations of the one or more control elements 111 of the user interface 110. Furthermore, the control unit 101 can be configured to operate the appliance 100 according to the one or more detected actuations. For example, parameter values ​​of the appliance 100 and / or an operating program of the appliance 100 (e.g., a washing program) can be set according to the one or more detected actuations. Alternatively or additionally, the operating program of the household appliance 100 can be started or stopped according to the one or more detected actuations.

[0045] The user interface 110 of the appliance 100 may include a dedicated control element 111 for activating or deactivating a lock function of the user interface 110. If the lock function of the user interface 110 is activated, actuation of one or more control elements 111 of the user interface 110 and corresponding user input may be ignored. When the lock function is deactivated, actuation of the one or more control elements 111 is considered as user input. The lock function may protect the appliance 100 from user input caused by children or animals.

[0046] It may happen that the user forgets to activate the lock function, thereby failing to protect the appliance 100 from user input caused by a child. In addition, the user may find it uncomfortable to explicitly activate the lock function. In addition, providing a dedicated control element 111 for activating and deactivating the lock function may increase the cost of the user interface 110.

[0047] The control unit 101 may be configured to be able to detect a sequence of actuations of one or more control elements 111 from a set of control elements 111 of the user interface 110 . Figure 2a An example sequence 200 of actuations 201 is shown. Different actuations 201 may exhibit specific characteristics, such as

[0048] Duration of actuation; and / or

[0049] Strength of actuation.

[0050] The sequence 200 of actuations 201 may further indicate a transition 202 between the individual actuations 201. A transition 202 between two actuations 201 may exhibit a certain duration or time interval.

[0051] The control unit 101 may be configured to analyze the sequence 200 of detected actuations 201, in particular with respect to

[0052] One or more control elements 111 that have been actuated;

[0053] the order in which the one or more control elements 111 are actuated;

[0054] the duration of each actuation 201; and / or

[0055] • The time interval between two directional consecutive actuations 201 .

[0056] In particular, the control unit 101 may be configured to be able to determine, in particular classify, whether the sequence 200 of detected actuations 201 is a "normal sequence" or an "abnormal sequence", in particular based on an analysis of the sequence 200 of detected actuations 201. A "normal sequence" may be a sequence that has a relatively high probability of being caused by an adult interacting (in a determined manner) with the appliance 100. On the other hand, an "abnormal sequence" may be a sequence that has a relatively high probability of being caused in a random and / or accidental manner (in particular by a child).

[0057] The control unit 101 may be configured to automatically activate a locking function of the user interface 110 if the detected sequence 200 of actuations 201 is classified as an "abnormal sequence". On the other hand, if the detected sequence 200 is classified as a "normal sequence", the appliance 100 may be operated according to the detected sequence 200 of actuations 201. By doing so, the locking function may be activated in a comfortable, reliable and efficient manner.

[0058] As described above, a touch screen (with one or more virtual control elements 111) can be used to control the household appliance 100. In some appliances 100, changes to the default option settings can be caused by the user via the user interface 110 of the appliance 100. The appliance 100 often has a parental lock feature, so that undesired option setting changes can be prevented. In some appliances 100, this child lock function can be used to prevent the start of an operating cycle.

[0059] However, it may happen that the user forgets to open the lock. Furthermore, it may happen that the touch screen of the appliance 110 is touched by a child or an animal, since the image of one or more control elements 111 is displayed on the touch screen when the lock is open. This may lead to an undesired activation of an operating program of the household appliance 100. Furthermore, the touch screen may be subjected to excessive use, which may shorten the service life of the touch screen. In addition, the sound of keystrokes may be annoying.

[0060] This article describes a device for converting a touch screen to a "ghost" mode (i.e., a black screen and a silent screen) when a child lock is turned on. The activation of the child lock can be performed in an automatic manner, or the user's approval may be required. In both cases, the ghost lock can be triggered by a real-time artificial intelligence algorithm. The unlocking of the appliance 100 can be performed, for example, by a specific button combination and / or a specific button of the appliance 100, or in a SW application on a remote user device (e.g., a smartphone). In this way, a device for improving the safety of a household appliance 100, extending the service life of a touch screen, and / or improving the user experience of a household appliance 100 is described.

[0061] In this article, a locking function is described that works in an automatic manner to reduce the burden on the user by applying an artificial intelligence algorithm. The algorithm can focus on pattern mining for discovering anomalies in the use of the household appliance 100 based on the sequence 200 of buttons 111 that are detected to be pressed. The algorithm can be household appliance 100 and / or model specific.

[0062] Data about the actuation 201 of the control element 111 of the household appliance 100 (particularly a specific type of household appliance 100 and / or a specific model of household appliance 100) can be collected (e.g., in a backend server) as training data. The training data may include characteristic information of each actuation 201 and / or the settings that have been changed.

[0063] The collected data can be converted into a flat file (e.g., a table) so that each event (e.g., each selected program, each temperature value) has its own row. This is the so-called long table. After further transformation and application of some logic, specific cycles can be distinguished. The resulting table can be relatively wide, where one row can be for an entire cycle of the operating program of the appliance 100. These files can be stored in a central data lake.

[0064] The collected training data can be used to enable a "ghost lock" function. To this end, data for each actuation 201 of the control element 111 can be taken into account (regardless of whether the actuation 201 occurs before or after the start of an operating cycle of the appliance 100). In particular, a set of reference sequences of actuations 201 can be defined based on the collected training data, wherein each reference sequence corresponds to a "normal" user interaction (i.e., corresponds to a normal sequence). The set of reference sequences can be defined by a human expert. Furthermore, the set of reference sequences can include sequences 200 of different lengths, i.e., sequences 200 of different numbers of actuations 201. However, the reference sequence can have a maximum length.

[0065] During operation of the appliance 100, a sequence 200 of actuations 201 detected by a user may be compared to the set of reference sequences. As actuations 201 are caused (i.e., as the detected sequence 200 becomes longer), the comparison may be performed repeatedly. In order to increase computational speed, the set of reference sequences may be grouped into lists and / or arrays and / or subsets according to their total length. An iterator may be used, which is a variable of integer type that stores the number of actuations 201 within a sequence 200 of actuations 201 detected by a user. A reference sequence having a total length equal to the iterator value may be identified and may be compared to the detected sequence 200. Based on a comparison with the set of reference sequences, an "abnormal" sequence or pattern may be detected as a sequence 200 that is not classified as a "normal" sequence. An "abnormal" sequence may be, for example, a sequence 200 that includes too many presses 201, multiple presses 201 on the same control element 111, and the like.

[0066] When abnormal use of the appliance 100 is detected, the smart ghost lock may be automatically activated.

[0067] Alternatively or additionally, an accumulated damage model (ADM) can be used to detect abnormal sequences. To this end, a two-dimensional matrix (more generally, a probability model) of the probabilities of different subsequent actuations 201 (after zero, one or more previous actuations 201 have been caused) can be determined (based on the collected training data). After zero, one or more previous actuations 201, each subsequent actuation 201 has a certain probability of being selected (for example, if the temperature has been set in a previous step, the probability that the user sets the temperature again in a subsequent step is relatively low).

[0068] Thus, based on the collected training data, a probability model may be generated that indicates the probability of a subsequent actuation 201 following a sequence 200 of zero, one or more already performed (ie, previous) actuations 201 .

[0069] During the sequence 200 of input actuations 201, an anomaly score may be updated after each actuation 201 using a predetermined probability model. In addition, the anomaly score may be compared to an anomaly threshold. If the anomaly score is less than the anomaly threshold, the sequence 200 of actuations 201 is considered a normal sequence. On the other hand, if the anomaly score is greater than the anomaly threshold, the sequence 200 of actuations 201 is considered an anomaly sequence.

[0070] The anomaly score may be iteratively updated after each newly detected actuation 201, for example, using the following formula, s n+1 =s n +s, where s n is the anomaly score of the previous n actuations 201, s is the score of the newly detected actuation 201, s n+1 is the anomaly score of the n+1 actuations 201 (including the newly detected actuation 201).

[0071] The score s of the newly detected actuation 201 may be determined based on a predetermined probability model. p may be the probability of the newly detected actuation 201 (given one or more previous actuations 201). Then, the score s of the detected actuation 201 may be determined based on (1-p) such that the score is relatively high for relatively unlikely actuations 201 and relatively low for relatively likely actuations 201.

[0072] In addition, the anomaly index "a n+1 ”, which depends on the value of the iterator n+1 and / or the index number of the newly detected actuation 201. The anomaly index may increase as the number of actuations 201 and / or the index number within the detected sequence 200 increases. The anomaly index “a n+1 ". Based on the abnormal index "a n+1 ", for example, based on (1-p)*a n+1 , especially s=(1-p)*a n+1 , to determine the fraction s of detected actuations 201 .

[0073] The formula for determining the anomaly score may utilize a scaling factor. Based on the collected training data, an anomaly threshold and / or a formula for determining the anomaly score may be determined in the context of a training process. Training may be performed in an appliance-specific and / or user-specific manner. Automatic detection of anomaly usage may be activated only after sufficient training data has been collected to perform the training process.

[0074] Once the detected sequence 200 is detected as an abnormal sequence, the ghost lock of the user interface 100 may be activated.

[0075] Figure 2bAn example evolution of the value of the anomaly score 210 as the total length of the sequence 200 of detected actuations 201 increases is shown. When the first actuation 201 is detected (index number 211n=1), a relatively low anomaly score 210(s1) is determined and compared to the anomaly threshold 212, thereby classifying the sequence 200 including only the first actuation 201 as a normal sequence. When the next (second) actuation 201 (n=2) is detected, the anomaly score 210(s2) is updated and compared to the anomaly threshold 212. In the example shown, when the fourth actuation 201 (n=4) is detected, the anomaly score 210(s4) exceeds the anomaly threshold 212, thereby classifying the sequence 200 of actuations 201 as abnormal and thereby causing the user interface 110 to be locked.

[0076] Figure 3 A flow chart of an example method 300 for controlling a user interface 110 of a household appliance 100 is shown, wherein the user interface 110 includes a set of control elements 111 (eg, one or more buttons, and / or one or more knobs, and / or one or more sliders).

[0077] The method 300 comprises detecting 301 a sequence 202 of actuations 201 of a set of control elements 111. The sequence 200 may indicate, for each actuation 201, a control element 111 that has been actuated. Thus, the sequence 200 may indicate one or more control elements 111 that have been actuated. Furthermore, the order in which the one or more control elements 111 have been actuated may be indicated. The user interface 110 may, for example, include control elements 111 "A", "B", and "C". The sequence 202 of the detected actuations 201 may be "A", "A", "C", "A", thereby indicating that control element "A" has been actuated first, followed by another actuation 201 of control element "A", followed by an actuation 201 of control element "C", followed by another actuation 201 of control element "A".

[0078] Thus, the sequence 200 of detected actuations 201 may indicate how (particularly in what order) the control elements 111 of the user interface 110 are actuated (eg touched in case of virtual control elements 111 on a touch screen).

[0079] The method 300 also includes classifying 302 the detected sequence 200 of actuations 201 as a normal sequence or an abnormal sequence. The classification may be performed using a trained classification unit that has been trained using collected training data about actual actuations 201 of control elements 111 of the user interface 110. The classification unit may be trained using machine learning. Thus, artificial intelligence may be used to classify the detected sequence 200 as normal or abnormal.

[0080] Furthermore, the method 300 comprises locking 303 the user interface 110 if the detected sequence 200 of actuations 201 is classified as an abnormal sequence. If an abnormal sequence 200 of actuations 201 is detected, the user interface 110 may be automatically locked. This may involve blackening and / or turning off the touch screen of the user interface 110.

[0081] The measures outlined in this article can improve the safety of use of the household appliance 100. In addition, the service life of the touch screen can be increased. In addition, the user experience of the appliance 100 can be improved.

[0082] It should be noted that the description and the accompanying drawings illustrate only the principles of the proposed method and system. Those skilled in the art will be able to implement various arrangements that, although not explicitly described or illustrated herein, embody the principles of the present invention and are included within its spirit and scope. In addition, all examples and embodiments outlined herein are primarily and explicitly intended to be used only for explanatory purposes to help the reader understand the principles of the proposed method and system. In addition, all statements of the principles, aspects, and embodiments of the present invention and their specific examples provided herein are intended to encompass their equivalents.

Claims

1. A control unit (101) for controlling a user interface (110) of a household appliance (100), the user interface (110) comprising a set of control elements (111); wherein: The control unit (101) is configured to - detecting a sequence (200) of actuations (201) of said set of control elements (111); - classifying a sequence (200) of detected actuations (201) as a normal sequence or an abnormal sequence; as well as - causing the user interface (110) to be locked if the detected sequence (200) of actuations (201) is classified as an abnormal sequence.

2. The control unit (101) according to claim 1, wherein: - the control unit (101) is configured to be able to classify the sequence (200) of detected actuations (201) using a classification unit; as well as - said classification unit has been pre-trained using training data indicating sequences of normal and / or abnormal actuations (201), - wherein the training data has preferably been collected in a backend server from a plurality of different household appliances of the same type and / or model as the household appliance.

3. A control unit (101) according to any one of the preceding claims, wherein: The control unit (101) is configured to - comparing the sequence (200) of detected actuations (201) with a set of reference sequences; wherein the set of reference sequences comprises sequences of actuations (201) each of which is considered to be a normal sequence; and - Classifying the sequence (200) of detected actuations (201) as a normal sequence or an abnormal sequence based on a comparison with said set of reference sequences.

4. The control unit (101) according to claim 3, wherein: - sorting the set of reference sequences according to the total number of actuations (201) contained in the respective reference sequence; - the control unit (101) is configured to be able to determine the total number of actuations (201) contained in the sequence (200) of detected actuations (201); as well as - identifying an actuation (201) having the same sequence (200) as the detected actuation (201) A subset of the total number of reference sequences; and - comparing the sequence (200) of detected actuations (201) with one or more reference sequences comprised in the subset of reference sequences in order to classify the sequence (200) of detected actuations (201) as a normal sequence or an abnormal sequence.

5. The control unit (101) according to claim 3 or 4, wherein: The control unit (101) is configured to - classifying said detected sequence (200) of actuations (201) as a normal sequence if the detected sequence (200) of actuations (201) corresponds to a reference sequence comprised in said set of reference sequences; and / or - Classifying said detected sequence (200) of actuations (201) as an anomalous sequence if said detected sequence (200) of actuations (201) does not correspond to any one of the reference sequences comprised in said set of reference sequences.

6. A control unit (101) according to any one of the preceding claims, wherein: The control unit (101) is configured to - determining an anomaly score (210) of the sequence (200) of detected actuations (201) based on a predetermined probabilistic model of actuations (201) of the set of control elements (111) of the user interface (110); as well as - Classifying the sequence (200) of detected actuations (201) as a normal sequence or an abnormal sequence according to an anomaly score (210).

7. The control unit (101) according to claim 6, wherein: - the predetermined probability model indicates, for each of a set of previous actuations (201), a probability value for each of a plurality of different subsequent actuations (201) following the respective previous actuation (201); and - the control unit (101) is configured to be able to - selecting an actuation (201) from the sequence (200) of detected actuations (201); - determining a subsequent actuation (201) following the selected actuation (201) from the sequence (200) of detected actuations (201); - using a probability model to determine a probability value for the determined subsequent actuation (201); as well as - determining an anomaly score (210) for the sequence (200) of detected actuations (201) based on the determined probability value.

8. The control unit (101) according to claim 6 or 7, wherein: - the control unit (101) is configured to be able to determine an anomaly score (210) of the detected sequence (200) of actuations (201) based on a predetermined anomaly index; as well as - the anomaly index indicates, for each of a plurality of index numbers (211) of actuations (201) in the sequence of actuations (201), a probability that the sequence of actuations (201) including the actuation (201) having the corresponding index number (211) is an anomaly sequence.

9. The control unit (101) according to any one of claims 6 to 8, wherein: The control unit (101) is configured to iteratively, - updating the anomaly score (210) when the sequence (200) of detected actuations (201) is appended with one or more new detected actuations (201); as well as - Classifying the sequence (200) of the appended actuations (201) as a normal sequence or an abnormal sequence according to the updated abnormality score (210).

10. The control unit (101) according to claim 9, wherein: The control unit (101) is configured to use the iterative formula s n+1 =s n +s to update the anomaly score (210); -s n is the anomaly score (210) of n actuations (201) of the sequence (200) of detected actuations (201); - s is the score of newly detected actuations (201), which depends on the probability model; and -s n+1 is an updated anomaly score (210) for the sequence (200) of appended actuations (201), which includes the n actuations (201) and the newly detected actuation (201).

11. The control unit (101) according to any one of the preceding claims, wherein: The control unit (101) is configured to repeatedly, - appending the sequence (200) of detected actuations (201) with the newly detected actuation (201); - classifying the sequence (200) of the additional actuations (201) as a normal sequence or an abnormal sequence; as well as - causing the user interface (110) to be locked if the sequence (200) of additional actuations (201) is classified as an abnormal sequence.

12. The control unit (101) according to any one of the preceding claims, wherein: The control unit (101) is configured to be able to, - determining one or more characteristics of a transition between two consecutive actuations (201) of one or more actuations (201) and / or a sequence of detected actuations (201); wherein the one or more characteristics include one or more of the following: - duration of the actuation (201); - the intensity of the actuation (201); and / or - a time interval of transition between two consecutive actuations (201); and - Classifying the sequence (200) of detected actuations (201) as a normal sequence or an abnormal sequence according to the one or more characteristics.

13. The control unit (101) according to any one of the preceding claims, wherein: The control unit (101) is configured to be able to, - automatically locking the user interface (110) if the detected sequence (200) of actuations (201) is classified as an abnormal sequence; and / or - When the user interface (110) is locked, the touch screen of the user interface (110) is darkened and / or turned off.

14. A household appliance (100), comprising - a user interface (110) having a set of control elements (111); and - A control unit (101) according to any of the preceding claims, configured to enable the user interface (110) to be locked.

15. A method (300) for controlling a user interface (110) of a household appliance (100), the user interface (110) comprising a set of control elements (111); wherein: The method (300) comprises - detecting (301) a sequence (200) of actuations (201) of said set of control elements (111); - classifying (302) the sequence (200) of detected actuations (201) as a normal sequence or an abnormal sequence; as well as - If the detected sequence ( 200 ) of actuations ( 201 ) is classified as an abnormal sequence, locking ( 303 ) the user interface ( 110 ).

Images