Calibration of gesture recognition algorithm for opening region of vehicle closed by movable part by gesture control

Abstract

According to a method for calibrating a gesture recognition algorithm for opening a vehicle area closed by a movable part (3) by gesture control, the gesture recognition algorithm is provided in a computer-readable format, which is adapted to recognize a predetermined gesture as a function of a time-varying sensor signal from an ultrasonic sensor (5) of the vehicle (1). The ultrasonic sensor (5) is used to generate a first sensor signal that varies over time while the first person (8) performs a gestural action in the field of view of the ultrasonic sensor (5), from which first sensor signal a first value of a predetermined parameter of a gesture recognition algorithm is determined and stored.

Description

Technical Field

[0001] The invention relates to a method for calibrating a gesture recognition algorithm for opening a vehicle area closed by a movable part by gesture control. The invention also relates to a method for opening a vehicle area closed by a movable part by gesture control, a corresponding vehicle device and a computer program product. Background Art

[0002] For example, when a user extends his legs into a target area designated for this purpose near the vehicle, devices for automatically opening a vehicle trunk lid or tailgate are known. For this reason, for example, a capacitive sensor is employed to detect the presence of the legs in the target area.

[0003] The disadvantage of this is that the reaction of these devices is relatively unspecific, so there is a risk of false positive detection if, for example, other objects are brought into the target area without the user's corresponding intention to automatically open the trunk lid or tailgate. To counteract this, for example, the device can be configured to make the target area appropriately smaller, thereby reducing the possibility of objects being unintentionally brought into the target area. However, this is also disadvantageous because if the user does want to automatically open the trunk lid or tailgate, they must enter a relatively small target area, which in turn means that the reliability of identifying the user's intention is reduced.

[0004] Furthermore, it is known to use ultrasonic sensors for gesture recognition. Since the distance of an object from the ultrasonic sensor can be evaluated in a time-varying manner by the ultrasonic sensor, the temporal behavior of, for example, a user's legs and gestures made by the user with his legs can also be recognized and characterized by the ultrasonic sensor.

[0005] However, when ultrasonic sensors are used for gesture recognition, it is also possible in principle that a corresponding movement of the user is not recognized as a gesture, since, for example, the movement may differ in detail depending on the individual or user.

[0006] Document DE 10 2016 013 935 A1 describes a method for automatically executing vehicle functions. In this case, the user is identified via the vehicle key and gestures or behavior patterns are subsequently recorded via one or more vehicle cameras. Based on this, the vehicle settings can be personalized, for example, the seat position can be set in a correspondingly personalized manner. Summary of the invention

[0007] The object of the present invention is to provide a possibility for contactless opening of a vehicle area which is closed by a movable part of the vehicle, by means of which the user's intention to open can be reliably detected.

[0008] This object is achieved by the respective subject matter of the independent claims. Advantageous developments and preferred embodiments are subject matter of the dependent claims.

[0009] The invention is based on the idea of ​​calibrating a predetermined gesture recognition algorithm for ultrasound-based gesture recognition by determining and storing values ​​of predetermined parameters of the gesture recognition algorithm based on sensor signals generated by the vehicle's ultrasonic sensor when a person performs a gesture action.

[0010] According to one aspect of the invention, a method for calibrating a gesture recognition algorithm for opening an area of ​​a vehicle, in particular a motor vehicle, which is closed by a movable part, by gesture control is proposed. The gesture recognition algorithm is provided in a computer-readable format. The gesture recognition algorithm is adapted to recognize a predetermined gesture based on a time-varying sensor signal of an ultrasonic sensor of the vehicle. When a first person performs a gesture action in the field of view of the ultrasonic sensor, a time-varying first sensor signal is generated by the ultrasonic sensor. A first value of a predetermined parameter of the gesture recognition algorithm is determined and stored, in particular adapted as a function, in particular adapted by at least one computing unit.

[0011] For example, the ultrasonic sensor of the vehicle and optionally further ultrasonic sensors may be part of an ultrasonic sensor system of the vehicle. The ultrasonic sensor system may also have at least one computing unit which may activate the ultrasonic sensor and may determine and store a first value of a predetermined parameter based on a first sensor signal.

[0012] The ultrasonic sensor system can be operated, for example, in a calibration mode and can generate a first sensor signal in the calibration mode. In addition, the ultrasonic sensor system can be operated in a gesture recognition mode, for example, in particular after the first value of the parameter has been determined and stored. The gesture recognition algorithm is particularly suitable for recognizing gestures in the gesture recognition mode based on the sensor signal of the ultrasonic sensor.

[0013] The at least one computing unit may activate the ultrasonic sensor to emit ultrasonic waves, for example to activate or deactivate a calibration mode and / or a gesture recognition mode.

[0014] For example, the ultrasonic wave can be emitted by the ultrasonic sensor in the form of one or more ultrasonic pulses. The part of the emitted ultrasonic wave that is reflected in the environment and detected by the ultrasonic sensor causes the ultrasonic sensor to generate a sensor signal or a first sensor signal, in particular as an electrical signal, a voltage signal or a current signal. The ultrasonic sensor can also be called an ultrasonic transducer.

[0015] For example, the first sensor signal can correspond to the amplitude or intensity of the ultrasonic wave which is reflected and detected at a certain point in time or at the corresponding envelope in each case.

[0016] Since the first person is located in the field of view of the ultrasonic sensor and performs a gesture action, the first sensor signal reflects the gesture action or is generated according to the gesture action. A gesture action may be understood as a movement of a body part of the first person, in particular a leg or a knee, which represents or is intended to represent or is intended to replicate a predetermined gesture. During the performance of the gesture action, the first person may be located in a predetermined gesture recognition area outside the vehicle, for example, in particular in the immediate vicinity of the ultrasonic sensor, wherein the field of view of the ultrasonic sensor covers the gesture recognition area.

[0017] For example, after determining and storing the first value of the parameter, the calibration mode can be deactivated. For example, the gesture recognition mode can then be activated. The determination and storage of the first value of the parameter of the gesture recognition algorithm, in particular the adaptation, can be understood as calibration of the gesture recognition algorithm. After carrying out the method according to the invention for calibrating the gesture recognition algorithm, gesture recognition can be used to recognize a predetermined gesture based on the first value of the parameter. In a similar manner, other parameters of the gesture recognition algorithm can also be determined and stored based on the first sensor signal and used for further use of the gesture recognition algorithm in the gesture recognition mode.

[0018] Since the first value of the parameter is determined in a person-dependent manner based on the gesture movement of the first person, characteristic movement patterns or properties of the gesture movement of the first person are taken into account, for example for subsequent gesture recognition in a gesture recognition mode. For example, in principle, a gesture recognition algorithm can already detect gestures based on the sensor signals of the ultrasonic sensor before the calibration method according to the invention is performed. However, the method according to the invention takes into account the person-dependent properties of the gesture movement and adjusts the gesture recognition mode accordingly. As a result, the reliability of gesture recognition in the gesture recognition mode can be increased. In other words, the probability that the first person wishes to perform a gesture but is not recognized by the gesture recognition algorithm due to the characteristic properties of the first person's movement can be reduced.

[0019] According to at least one embodiment of the method for calibrating a gesture recognition algorithm, a characteristic attribute of a gesture movement of a first person is determined based on a first sensor signal, in particular by at least one computing unit, and a first value of a parameter is determined based on the characteristic attribute determined about the first person, in particular by at least one computing unit.

[0020] For example, the characteristic attribute may be the initial distance of the body part moved by the first person during the gesture, or the characteristic attribute may be derived therefrom. Thus, advantageously, it may be considered that different persons typically maintain different distances from the vehicle when performing gestures.

[0021] Likewise, the characteristic attributes may correspond to, or be derived from, the final spacing of the body parts after or at the completion of the gesture action.

[0022] A gesture action can correspond to a movement of a body part, such as in particular a leg, from a starting position in the direction of the vehicle or the ultrasonic sensor and back to the starting position or an approximate starting position. Thus, the distance between the body part and the ultrasonic sensor is equal to the initial distance, in particular at the beginning of the gesture action, then decreases and then increases again until the final distance is reached.

[0023] By taking the final distance into account, it can be taken into account that different persons, when performing a gesture action, typically move or take a step towards, for example, the ultrasonic sensor. In other words, the person may not return the leg completely to the starting position and the initial distance, so that the final distance does not match the initial distance. This can be used for the person-specific characterization and the determination of the corresponding person-specific first value of the parameter.

[0024] The feature attribute may also correspond to or be derived from the distance between the final spacing and the initial spacing.

[0025] The characteristic attribute may also correspond to or be derived from the speed of movement of the body part during the gesture action.

[0026] Thus, advantageously, it may be taken into account that different persons typically move body parts at different speeds in order to perform a gesture action.

[0027] The characteristic attribute may correspond to or be derived from the duration of the gesture. Duration may also be used to map and account for person-specific characteristics of gestures, such as different speeds of gestures and person-specific body proportions, such as leg length, etc.

[0028] The characteristic attribute may also correspond to or be derived from the amplitude of the signal pulses of the first sensor signal during the gesture action, in particular the maximum amplitude. The maximum amplitude may be affected by, for example, the body shape, physique or clothing of the first person, which are also other typical variables affecting the first value of the characteristic or parameter of the gesture action.

[0029] The characteristic attributes are not limited to the examples mentioned. The mentioned multiple characteristic attributes or other characteristic attributes can be determined based on the first sensor signal, and the first value of the parameter can be determined based on these characteristic attributes. Two or more characteristic attributes can also be combined or offset with each other to determine the first value of the parameter.

[0030] In general, a parameter may correspond directly to one of the feature attributes or be derived from one or more of the feature attributes.

[0031] By using the characteristic attributes as a basis for determining the first value of the parameter, the gesture recognition algorithm can be influenced in a targeted manner, thereby increasing the reliability of gesture recognition. Thus, the first value of the parameter can be determined in particular in such a way that the gesture recognition algorithm can take into account particularly relevant characteristic attributes during gesture recognition, while characteristic attributes that may be less relevant are not given any weight or are given a smaller weight during gesture recognition. In particular, this can prevent the gesture recognition algorithm from being overspecified and prevent any changes in the gesture movements of the same person from being able to influence the reliability of gesture recognition.

[0032] According to at least one embodiment, when the first person repeatedly performs a gesture action in the field of view of the ultrasonic sensor, the first sensor signal is repeatedly generated. Based on the repeatedly generated first sensor signal, a characteristic attribute about the gesture action of the first person is also repeatedly determined. The first value of the parameter is determined based on the repeatedly determined characteristic attribute, such as by averaging or some other statistical processing.

[0033] In this way, fluctuations in characteristic attributes of the same person during gesture actions can be offset, thereby further improving the reliability of gesture recognition.

[0034] In accordance with at least one embodiment, a gesture recognition algorithm implements a state machine.

[0035] The state machine provides an algorithm which can establish one of two or more predefined states, such as the first sensor signal or a gesture action or a predefined state of the first person, based on the first sensor signal, and in particular can detect the presence of a gesture based on a time sequence of states determined in this way.

[0036] For example, different states of the state machine may correspond to a state where a body part is stationary within a certain spacing range of the ultrasonic sensor and / or a state where a body part is moving in a certain direction. In a non-limiting example, for example, if a state where a body part is stationary in a gesture recognition area is followed by a state where the body part is moving toward the ultrasonic sensor, which is followed by a state where the body part is moving away from the ultrasonic sensor, which is followed by a state where the body part is stationary in the gesture recognition area again, a gesture may be detected.

[0037] For example, the various states may be identified using predetermined parameters and other parameters. For example, one or more characteristic attributes may be directly used or processed as parameters of a state machine for identifying the various states. In this way, by specifying a first value for a parameter on an individual-specific basis, state identification, and thus gesture recognition, may be directly influenced by the state machine.

[0038] According to at least one embodiment, when a second person (in particular a person different from the first person) performs a gesture action in the field of view of the ultrasonic sensor, a second sensor signal that varies over time is generated by the ultrasonic sensor. Based on the second sensor signal, in particular by at least one computing unit, a characteristic attribute of the gesture action of the second person is determined. Based on the determined characteristic attribute of the second person, in particular by at least one computing unit, a second value of the parameter is determined and stored.

[0039] Therefore, different parameter values ​​assigned to different people or different variants of gesture recognition algorithms can be stored and retained. For example, corresponding user profiles can be created for different people, which store respective parameter values ​​or correspondingly adapted gesture recognition algorithms. This can further improve the reliability of gesture recognition.

[0040] According to at least one embodiment, the gesture recognition algorithm includes a trained recurrent neural network (RNN), such as a long short term memory (LSTM), and the parameters are weighting factors or bias parameters of the RNN.

[0041] In such an embodiment, the gesture recognition algorithm can directly recognize the gesture, for example, based on the input data generated from the sensor signal of the ultrasonic sensor, without performing the intermediate step of determining the characteristic attributes. This has the advantage that it is not necessary to know the individual-specific differences in gesture movements and their impact on gesture detection.

[0042] In such an embodiment, the method for calibrating the gesture recognition algorithm may be understood as the training of the RNN based on the first sensor signal being improved or refined.

[0043] For example, if the ultrasonic sensor system is in calibration mode, it can be assumed that the gesture action performed actually corresponds to a gesture. Therefore, the parameterization of the RNN can be adjusted according to the first sensor signal as needed, for example to increase the confidence of detecting a gesture based on this.

[0044] According to at least one embodiment, a calibration mode of the ultrasonic sensor system is activated, for example by at least one computing unit, and the first sensor signal is generated in the calibration mode. For example, the calibration mode can be initiated by a first person.

[0045] For example, at least one radio signal can be transmitted by an electronic device external to the vehicle to the at least one computing unit, and the calibration mode can be activated as a function of the at least one transmitted radio signal.

[0046] For example, the electronic device outside the vehicle, in particular the electronic device that is therefore not part of the vehicle, may be equivalent to a mobile electronic terminal, a smart phone, a smart watch or some other so-called wearable device, a vehicle key or a key chain, etc. Thus, in particular, the calibration mode may be initiated when the first person notices that the reliability of the gesture detection needs to be improved.

[0047] According to at least one embodiment, identification information of the first person is transmitted to at least one computing unit using at least one radio signal. Based on the identification information, a first value of the parameter is stored in a user profile of the first person. For example, in a corresponding embodiment, the same applies to the second person.

[0048] In various embodiments, the calibration mode may also be activated based on user input recorded via a user input device of the vehicle.

[0049] According to another aspect of the invention, a method for opening a vehicle area closed by a movable part by gesture control is specified. The method according to the invention for calibrating a gesture recognition algorithm is performed here. After the method for calibrating a gesture recognition algorithm is performed, a further first sensor signal varying over time is generated by the ultrasonic sensor while the first person again performs a gesture action within the field of view of the ultrasonic sensor, for example during a gesture recognition mode. By applying the gesture recognition algorithm to the input data based on the further sensor signal, a predetermined gesture is detected according to the stored first value, in particular by at least one computing unit. In response to the detection of the gesture, the movable part is automatically moved to open the closed area.

[0050] Responsive to the detection of a gesture may be understood as automatically moving the movable component only if a gesture has been detected. In other words, input data is generated from the further first sensor signal, and a gesture recognition algorithm is applied to the input data according to the first values ​​of the stored parameters in order to detect a gesture.

[0051] For example, if the first person is located in the gesture recognition area, the gesture recognition mode may be activated. However, other conditions for activating the gesture recognition mode may also be envisaged.

[0052] Since the further first sensor signal reflects the distance between the first person or a body part of the first person and the ultrasonic sensor, the manifestation of the gesture movement can be recognized as a gesture and detected accordingly.

[0053] Depending on the embodiment of the method and the vehicle, the movable part can be a door, for example, in particular a side door, a driver's door, a passenger door, a rear door, a back door, a sliding door, a tailgate or a luggage compartment lid. Preferably, it is a tailgate or a luggage compartment lid. In this case, the closed area of ​​the vehicle is, for example, a luggage compartment, or an open or closed cargo area of ​​the vehicle. The open cargo area can in particular be open at the top, but this does not exclude that the open cargo area is closed in the rear direction by a tailgate, for example in the case of a pickup truck or a minivan with an open cargo surface.

[0054] For application cases or application scenarios that may occur in the method but are not explicitly described here, it is conceivable according to the present invention to output error messages and / or user feedback input requests, and / or set default settings and / or preset initial states.

[0055] According to another aspect of the invention, a device for a vehicle is specified for opening a vehicle area closed by a movable part by gesture control. The device has an ultrasonic sensor system with an ultrasonic sensor, wherein the ultrasonic sensor system is designed to generate a first sensor signal that varies over time by the ultrasonic sensor when a first person performs a gesture action in the field of view of the ultrasonic sensor. The ultrasonic sensor system has at least one computing unit that is designed to determine and store a first value of a predetermined parameter of a gesture recognition algorithm based on the first sensor signal.

[0056] A computing unit may be understood in particular as a data processing device comprising a processing circuit. Thus, a computing unit may process data, in particular for performing computing operations. Optionally, these also include operations for performing indexed accesses to data structures, such as look-up tables (LUTs).

[0057] The computing unit may specifically include one or more computers, one or more microcontrollers and / or one or more integrated circuits, such as one or more application-specific integrated circuits (ASICs), one or more field-programmable gate arrays (FPGAs) and / or one or more systems-on-a-chip (SoCs). The computing unit may also include one or more processors, such as one or more microprocessors, one or more central processing units (CPUs), one or more graphics processing units (GPUs) and / or one or more signal processors, in particular one or more digital signal processors (DSPs). The computing unit may also include physical or virtual computer groups or other types of the above-mentioned units.

[0058] In various exemplary embodiments, a computing unit includes one or more hardware and / or software interfaces and / or one or more storage units.

[0059] The memory cell may be configured as a volatile data memory, such as a dynamic random access memory (DRAM) or a static random access memory (SRAM), or as a non-volatile data memory, such as a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory or a flash EEPROM, a ferroelectric random access memory (FRAM), a magnetoresistive random-access memory (MRAM), or a phase-change random-access memory (PCRAM).

[0060] According to at least one embodiment, the device has a storage medium that stores a gesture recognition algorithm in a computer-readable format. The ultrasonic sensor system is designed to generate, by the ultrasonic sensor, a further first sensor signal that varies over time when the first person again performs a gesture action in the field of view of the ultrasonic sensor after determining and storing the first value of the parameter. The at least one computing unit is designed to apply the gesture recognition algorithm to input data based on the further first sensor signal in accordance with the stored first value of the parameter in order to detect a predetermined gesture. The at least one computing unit is designed to generate at least one control signal in response to the detection of the gesture to automatically move the movable component.

[0061] According to at least one embodiment, the device has at least one actuator which is designed to automatically move the movable part according to a control signal in order to open the closed area.

[0062] The at least one computing unit may transmit the control signal directly to the at least one actuator or to the control unit of the at least one actuator, and the control unit of the at least one actuator generates other control signals and transmits them to the at least one actuator so as to cause the actuator to move the movable component according to the other control signals.

[0063] If the present disclosure relates to a component of a device according to the present invention, in particular at least one computing unit of the device, and the component is designed, formed, configured, etc. to perform or realize a specific function, achieve a specific effect, or serve a specific purpose, this can be understood as that the component is not only basically or theoretically used or suitable for the function, effect, or purpose, but also specifically and actually capable of performing or realizing the function, achieving the effect, or serving the purpose through appropriate adaptation, programming, physical configuration, etc.

[0064] Further embodiments of the device according to the invention follow directly from the various configurations of the method according to the invention and vice versa. In particular, the individual features and the corresponding explanations and advantages of the various embodiments relating to the method according to the invention can be transferred in an analogous manner to the corresponding embodiments of the device according to the invention. In particular, the device according to the invention can be formed or programmed to perform the method according to the invention. In particular, the device according to the invention performs the method according to the invention.

[0065] According to another aspect of the invention, a computer program is specified which contains instructions which, when executed by a device according to the invention, in particular by at least one computing unit, cause the device to perform a method according to the invention for calibrating a gesture recognition algorithm and / or a method according to the invention for opening a closed area of ​​a vehicle by gesture control.

[0066] According to another aspect of the invention, a computer-readable storage medium is specified, which stores a computer program according to the invention.

[0067] Computer program and computer-readable storage medium may be understood as corresponding computer program products containing instructions.

[0068] Further features of the invention can be found in the claims, the drawings and the description of the drawings. The features and feature combinations mentioned in the above description as well as the features and feature combinations mentioned in the description of the drawings and / or shown in the drawings may be included in the invention not only in the combination specified in each case but also in other combinations. In particular, embodiments and feature combinations which do not have all the features of the original claims may also be included in the invention. Furthermore, embodiments and feature combinations which exceed or differ from the feature combinations set out in the back-references of the claims may also be included in the invention.

[0069] The present invention is explained in more detail below based on specific exemplary embodiments and with reference to relevant schematic diagrams. In the accompanying drawings, identical or functionally identical elements may have the same reference numerals. For identical or functionally identical elements in different drawings, their descriptions are not necessarily repeated. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In the attached picture:

[0071] Figure 1 a schematic diagram showing a vehicle according to the invention, comprising an exemplary embodiment of a device for automatically opening a vehicle area closed by a movable part; and

[0072] Figure 2 A schematic flow chart of an exemplary embodiment of a method according to the invention for automatically opening a vehicle area closed by a movable component is shown. DETAILED DESCRIPTION

[0073] Figure 1 A vehicle 1 is schematically depicted with an exemplary embodiment of a device 2 according to the invention for opening a region of the vehicle 1 closed by a movable part 3 by gesture control.

[0074] The movable component 3 is here exemplarily depicted as a trunk lid of the vehicle 1, so that the area closed thereby is the trunk of the vehicle 1. In other embodiments, the movable component may alternatively also be a tailgate or a door of the vehicle 1.

[0075] The vehicle (eg, the device 2) has in particular one or more (in Figure 1In a non-limiting example, there are two) actuators 4a, 4b and a computing unit 6. The computing unit 6 can be regarded as representing one or more control, computing and / or evaluation units, which are functionally combined in the description to simplify the description. The computing unit 6 can specifically activate the actuators 4a, 4b so that these actuators, for example, automatically start the movement of the movable part 3 and optionally release the lock of the movable part 3. Therefore, the vehicle area closed by the movable part 3 can be opened automatically, that is to say without the user manually opening the movable part 3. In known systems, the movable part can be released by the vehicle key of the vehicle 1. This is also possible in the present device 2, but it is not absolutely necessary. Instead, the device 2 can perform a method according to the present invention for opening a closed area by gesture control.

[0076] For this purpose, the device 2 has an ultrasonic sensor 5, which has a field of view in which objects such as persons 8 can be detected. In particular, the ultrasonic sensor 5 can emit ultrasonic pulses and detect the ultrasonic pulse portions reflected by objects in the field of view of the ultrasonic sensor 5, and can generate a corresponding sensor signal that varies over time, which reproduces the intensity of the detected ultrasonic pulse portions. The sensor signal then usually contains signal pulses, also called peaks or echoes, each signal pulse being based on an emitted ultrasonic pulse. The time position of the signal pulses enables the distance of the reflecting object from the ultrasonic sensor 5 to be determined by the calculation unit 6.

[0077] The ultrasonic sensor 5 and the computing unit 6 or parts of the computing unit 6 and possibly other ultrasonic sensors (not shown) can be understood as part of an ultrasonic sensor system of the device 2. The ultrasonic sensor system, in particular the ultrasonic sensor 5, can be used not only in the sense of the present invention but can also optionally be used for other functions, such as parking assistance for the driver of the vehicle 1.

[0078] The method according to the invention for opening an area of ​​a vehicle 1 closed by a movable part 3 by gesture control envisages carrying out the method according to the invention for calibrating a gesture recognition algorithm which is in particular stored on a storage medium of the computing unit 6 in a computer-readable format.

[0079] In a calibration mode (which can be initiated, for example, by a person 8 via a user input interface of the vehicle 1 or an electronic device external to the vehicle (such as a smartphone)), the person 8 can perform gestures in the field of view of the ultrasonic sensor 5, for example, in a predetermined gesture recognition area 7 near the ultrasonic sensor 5, such as moving a leg from a standing position toward the ultrasonic sensor 5 and back. When the person 8 performs a gesture, the ultrasonic sensor 5 generates a corresponding sensor signal that varies over time. Based on this, the calculation unit 6 determines the values ​​of the predetermined parameters of the gesture recognition algorithm, in particular the values ​​of the predetermined parameters for the person 8, and stores them.

[0080] For this purpose, for example, the computing unit 6 can determine characteristic attributes of the gesture of the person 8 from the sensor signal and determine the value of the parameter from the characteristic attributes. Optionally, the person 8 can also repeatedly perform gestures to compensate for fluctuations in the characteristic attributes.

[0081] After the method for calibrating the gesture recognition algorithm has been completed, the same person 8 can then appear again in the gesture recognition area 7 at a later point in time in the gesture recognition mode and perform a gesture action. The calibrated gesture recognition algorithm can now generate input data for the gesture recognition algorithm based on the further sensor signals correspondingly generated by the ultrasonic sensor and apply the gesture recognition algorithm (in particular using or based on the stored parameter values) to the input data. The gesture recognition algorithm can thus detect the predetermined gesture.

[0082] If a gesture is detected, the computing unit 6 can generate at least one control signal for automatically moving the movable part 3 and transmit it to the actuators 4a, 4b, thereby initiating the opening of the closed area.

[0083] For the gesture recognition itself, the computing unit 6 can use methods known per se. For example, the gesture recognition algorithm can implement a state machine. By applying the gesture recognition algorithm to the input data, the computing unit 6 can thus recognize when a gesture action performed by the person 8 corresponds to a predetermined and predefined sequence of two or more states and detect a gesture based thereon.

[0084] However, other algorithms for gesture recognition are also contemplated and can be used. For example, the gesture recognition algorithm may include a trained recurrent neural network RNN.

[0085] Figure 2 A schematic flow chart of an exemplary embodiment of a method according to the invention for opening an area of ​​a vehicle 1 closed by a movable part 3 by gesture control is shown. An embodiment of the method according to the invention for calibrating a gesture recognition algorithm is performed according to steps S1, S2 and S3 which are optionally repeated for a plurality of persons.

[0086] In particular, in step S1, a gesture recognition algorithm is provided in a computer-readable format, and in step S2, a time-varying sensor signal is generated when a person 8 is located in the field of view of the ultrasonic sensor 5 and performs a gesture action. In step S3, the value of a parameter is determined and stored based on the sensor signal, for example by determining a characteristic attribute about the gesture action of the person 8. After performing steps S1 to S3, the calibration mode ends and the gesture recognition mode is activated, for example, at a later point in time. In step S4, when the person 8 performs a gesture action again in the field of view of the ultrasonic sensor 5, a further sensor signal is generated by the ultrasonic sensor 5, and in step S5, the gesture recognition algorithm is applied to the input data based on the further sensor signal according to the stored parameter value, and a predetermined gesture is detected. In step S6, in response to the detection of the gesture, the movable part 3 is automatically moved to open the closed area.

[0087] Thus, a particularly robust gesture recognition can be achieved, which can recognize gesture actions of various types or expressions. For example, the sensor signal can be fed directly to the RNN, which can make a binary decision as to whether the gesture action corresponds to a gesture.

[0088] One advantage of an RNN is that it features an internal state that acts like a memory of previous measurements, making it possible to identify changes between measurements. This also makes external memory for measurement results redundant, saving storage space. For example, the RNN used can be an LSTM.

[0089] The RNN can be trained using the captured data in order to learn internal parameters and recognize when a gesture occurs. It may be advantageous to tailor the pre-trained RNN to the specific model of vehicle 1, as sensor locations and other properties may change depending on the vehicle model. This can be achieved by capturing a dataset specifically using that vehicle model and further training the RNN accordingly.

Claims

1. A method for calibrating a gesture recognition algorithm for opening an area of ​​a vehicle (1) closed by a movable part (3) by gesture control, It is characterized in that - the gesture recognition algorithm is provided in a computer-readable format, the algorithm being adapted to recognize a predetermined gesture based on a time-varying sensor signal of an ultrasonic sensor (5) of the vehicle (1); - generating a first sensor signal varying over time by the ultrasonic sensor (5) when the first person (8) performs a gesture in the field of view of the ultrasonic sensor (5); as well as - Determining and storing a first value of a predetermined parameter of the gesture recognition algorithm based on the first sensor signal.

2. The method according to claim 1, It is characterized in that - determining characteristic attributes of the gesture action of the first person (8) based on the first sensor signal; as well as - determining said first value of said parameter based on said characteristic attribute determined about said first person (8).

3. The method according to claim 2, It is characterized in that The characteristic attributes - is related to the initial distance of the body part moved during the gesture relative to the ultrasonic sensor (5); and / or - is related to the final spacing of the body parts after or at the completion of the gesture; and / or - is related to the speed of movement of the body part during the gesture; and / or - is related to the duration of the gesture; and / or - is related to a maximum amplitude of signal pulses of said first sensor signal during said gesture.

4. The method according to claim 2 or 3, It is characterized in that - repeatedly generating the first sensor signal when the first person (8) repeatedly performs the gesture in the field of view of the ultrasonic sensor (5); - based on the repeatedly generated first sensor signal, repeatedly determining the characteristic attribute about the gesture action of the first person (8); - determining said first value of said parameter based on the repeatedly determined characteristic attribute.

5. The method according to any one of claims 2 to 4, It is characterized in that The gesture recognition algorithm implements a state machine.

6. The method according to any one of claims 2 to 5, It is characterized in that - when a second person performs the gesture in the field of view of the ultrasonic sensor (5), generating a second sensor signal that varies over time by the ultrasonic sensor (5); - determining characteristic attributes of the gesture action of the second person based on the second sensor signal; - Determining and storing a second value of said parameter based on said characteristic attribute determined about said second person.

7. The method according to claim 1, It is characterized in that The gesture recognition algorithm includes a recurrent neural network, and the parameters are weighting factors or bias parameters of the recurrent neural network.

8. A method as claimed in any one of the preceding claims, It is characterized in that A calibration mode of an ultrasonic sensor system (5, 6) comprising the ultrasonic sensor (5) is activated, in which the first sensor signal is generated.

9. The method according to claim 8, It is characterized in that - transmitting at least one radio signal to at least one computing unit (6) of the ultrasonic sensor system (5, 6) via electronic means external to the vehicle; as well as - activating the calibration mode in dependence on the at least one transmitted radio signal.

10. The method according to claim 8 or 9, It is characterized in that - using said at least one radio signal to transmit identification information of said first person (8) to said at least one computing unit (6); and - storing said first value of said parameter in a user profile of said first person (8) based on said identification information.

11. The method according to claim 8, It is characterized in that The calibration mode is activated based on user input recorded via a user input device of the vehicle (1).

12. A method for opening an area of ​​a vehicle (1) closed by a movable part (3) by gesture control, It is characterized in that - performing a method for calibrating the gesture recognition algorithm according to any of the preceding claims; - after carrying out the method for calibrating the gesture recognition algorithm, generating a further first sensor signal varying over time by means of the ultrasonic sensor (5) while the first person (8) again performs a gesture action within the field of view of the ultrasonic sensor (5); - detecting a predetermined gesture according to the stored first value of the parameter by applying the gesture recognition algorithm to input data based on the further first sensor signal; - In response to said detection of said gesture, automatically moving said movable part (3) to open said closed area.

13. A device (2) for a vehicle (1) for opening an area of ​​the vehicle (1) closed by a movable part (3) by gesture control, It is characterized in that - the device (2) has a storage medium storing a gesture recognition algorithm in a computer-readable format, the gesture recognition algorithm being suitable for recognizing a predetermined gesture based on a time-varying sensor signal of an ultrasonic sensor (5) of the vehicle (1); the device (2) having an ultrasonic sensor system (5, 6) with an ultrasonic sensor (5), wherein the ultrasonic sensor system (5, 6) is designed to generate a first sensor signal which varies over time by means of the ultrasonic sensor (5) when a first person (8) performs a gesture in the field of view of the ultrasonic sensor (5); The ultrasonic sensor system (5, 6) has at least one computing unit (6) which is designed to determine and store a first value of a predetermined parameter of a gesture recognition algorithm as a function of the first sensor signal.

14. The device (2) as claimed in claim 13, It is characterized in that the ultrasonic sensor system (5, 6) is designed to generate, after determining and storing the first value of the parameter, a further first sensor signal which varies over time by means of the ultrasonic sensor (5) when the first person (8) again performs the gesture in the field of view of the ultrasonic sensor (5); - the at least one computing unit (6) is designed to apply the gesture recognition algorithm to the input data based on the further first sensor signal in order to detect a predetermined gesture, depending on the stored first value of the parameter; as well as - said at least one computing unit (6) is designed to generate at least one control signal in response to the detection of said gesture in order to automatically move said movable part (3).

15. A computer program product comprising instructions which, when executed by a device (2) as claimed in claim 13, cause the device (2) to perform a method as claimed in any one of claims 1 to 11, and / or, when executed by a device (2) as claimed in claim 14, cause the device (2) to perform a method as claimed in claim 12.

Images