Method for automatically presetting a seat in a motor vehicle and motor vehicle with a system for automatically presetting a seat

Abstract

The invention relates to a method for automatically presetting a seat (11) in a motor vehicle (13) before an entry process and to a motor vehicle with a system for automatically presetting a seat.In a first step, a sensor system that detects the predetermined environment of the vehicle (13) detects an intention to enter the vehicle (S1) for at least one person located in a predetermined environment outside the vehicle (13). In a second step, an individual seating position, dependent on the body dimensions of the respective person, is determined for each person for whom an intention to enter has been detected (S2). In a third step, an approach position is determined (S3). If an intention to enter has been detected for more than one person, the approach position corresponds to an intermediate position, which is calculated from the individual seating positions using a predefined calculation rule (S3.2). In a final step, at least one seat is adjusted to the determined approach position (S4).

Description

[0001] The invention relates to a method for automatically presetting a seat in a motor vehicle before an entry process and to a motor vehicle with a system for automatically presetting a seat, wherein the system is configured to carry out the method.

[0002] With the transformation of vehicles into intelligent and connected spaces, user expectations regarding comfort, convenience, and personalization are constantly rising. The car has long since transcended its role as a mere means of transportation. It is evolving into an extension of the personal living space, adapting to the individual needs and preferences of its occupants. This development is reflected in a multitude of innovations aimed at making the driving experience more intuitive and enjoyable.

[0003] Current vehicle models already offer a certain degree of personalization in terms of seat settings. Memory functions allow preferred seat positions to be saved, which can then often be recalled at the touch of a button or via a link to the vehicle key. For example, the driver's seat moves to the last saved position as soon as the correct key is detected in the ignition or the door is opened.

[0004] While these features represent progress, they do have limitations. Personalization typically only occurs after opening the vehicle door by pressing a button located on the inside of the door, after physically entering the vehicle, or after explicit interaction with the key. Seat personalization can be done mechanically or electrically, or through interaction with the vehicle's touchscreen. All solutions share the drawback of requiring multiple steps to pre-adjust the seat. This means that the driver or passenger must first get into a potentially uncomfortable or suboptimal seating position before the adjustment begins. This moment of manual intervention or waiting for automatic adjustment disrupts the otherwise desired seamless user experience.In some situations, this delay is perceived as particularly unpleasant, for example when the user of the motor vehicle is standing in the rain, or can even be dangerous, for example when the motor vehicle is parked on a road with high traffic volume.

[0005] Other solutions to facilitate vehicle entry involve identifying the user via a registered mobile device (smartphone or smartwatch). The mobile device can either be generally authorized or used, or a certificate for a specific authorization profile and limited duration can be issued via a link. This allows the seat's entry position to be saved and automatically retrieved for each key or certificate.

[0006] However, in some situations, mobile devices and physical keys are not carried by the respective driver or passenger, or cannot be clearly assigned to a specific user. For example, in a shared vehicle or a rental car, often only one key or a generic access device is used that is not specifically assigned to an individual person, or identification is carried out via an app on the mobile device that is not permanently linked to the vehicle. Furthermore, when a person uses the vehicle for the first time, the system may not yet have learned the necessary association with the stored seat position and preferences.

[0007] In all these cases, it is unclear who the actual driver or passenger will be, and therefore the personalization cannot implement the previously saved optimal seating position and various other preferences in the vehicle. In the worst-case scenario, this can lead to a situation where a taller person, after a shorter driver, may have difficulty getting into the vehicle without prior manual seat adjustment, or may even be unable to do so. Especially in multi-driver households, where the same vehicle is used by people with very different body sizes and seating preferences, quick and precise adjustment is advantageous. Furthermore, the subsequent manual seat adjustment results in a loss of time, which can be perceived as significant and may reduce the desired comfort.In the worst-case scenario, manually adjusting the seat is not only uncomfortable but can also distract the driver just before or during the journey. The seemingly simple solution of moving the seat to its rearmost position if the occupant cannot be clearly identified does solve the problem of anyone being able to get into the vehicle, but it has several drawbacks. Firstly, the time factor plays a role here as well, since in many cases, especially with very short people, a considerable amount of time is wasted moving the seat all the way back and then forward again to find a comfortable driving position. Secondly, it has been found that many vehicle users are irritated when the seat is moved to its maximum rearward position, so this solution is not perceived as comfortable and ultimately has very low acceptance.

[0008] From DE 196 30 189 A1, a vehicle seat with a power-operated adjustment mechanism and an access device for a space located behind the vehicle seat is known. The access device is designed as a control device for actuating the actuators of the vehicle seat's adjustment mechanism, thereby reducing the number of actuating and operating devices. A first signal moves the seat to a forward end position, and a second signal returns it to its original position. A disadvantage of this design is that the seat position to which it moves is not variably adjustable.

[0009] German patent application DE 10 2006 052 087 B4 discloses a vehicle with a device for facilitating entry and exit. The device allows for the combined adjustment of a front seat and at least one of the vehicle components "rear seat" and / or "electrically adjustable side window in the rear area" and / or "sliding side sill in the door opening area" to make access to the vehicle interior, particularly the rear area, more comfortable. A disadvantage is that the seat is not automatically pre-adjusted before entry.

[0010] DE 10 2022 111 362 B3 relates to a method for adjusting at least one interior element of a vehicle and an emergency assistance system for carrying out at least one support measure for the occupant, wherein the target position is chosen in such a way that the movement space limited by the at least one interior element is increased for the occupant's access to and / or exit from the vehicle. A disadvantage is that the occupant's body dimensions are not taken into account when adjusting the interior element.

[0011] The present invention is based on the objective of facilitating entry into a motor vehicle, particularly in situations where more than one person with different body dimensions approaches the motor vehicle.

[0012] The problem is solved by the subject matter of the independent claims. Advantageous embodiments of the invention are described by the dependent claims, the following description, and the figures.

[0013] The invention provides a method for automatically presetting a seat in a motor vehicle before boarding, comprising the following steps: In a first step, a sensor system, which detects the predetermined environment of the motor vehicle, detects an intention to board for at least one person located in a predetermined environment outside the motor vehicle. In a subsequent step, a seating position, dependent on the body dimensions of each person, is determined for each person for whom an intention to board has been detected. In a further step, an approach position is determined, wherein, if an intention to board has been detected for more than one person, the approach position corresponds to an intermediate position, which is calculated from the previously determined individual seating positions using a predetermined calculation formula.In a final step, at least one seat is adjusted to the determined approach position. This procedure can, of course, be carried out for multiple seats in the vehicle by performing the described steps individually for each seat and for each person whose intention to enter is detected.

[0014] In other words, the system automatically adjusts the seat to a suitable position before a person enters the vehicle. When one or more people approach the vehicle, this is detected by the sensor system. This detection can occur in various ways. For example, a person's movements can be recorded by at least one camera. Alternatively, motion data from the accelerometer and / or gyroscope built into a mobile device can be analyzed. A seating position tailored to each person's body dimensions is then determined. Factors such as height, build, and / or leg length are taken into account. The seat is then adjusted to the appropriate position.In the straightforward case where only one person wants to board, this approach position corresponds to the individual seating position determined for that person in the previous step. If an intention to board is detected for more than one person, the seat is adjusted to the intermediate position calculated according to the formula.

[0015] When moving the seat, care can be taken to ensure that no person and / or animal is injured by the movement of the seat by using additional interior sensors, such as cameras in the interior and / or seat sensors and / or seat belt buckle sensors, in addition to the sensors already described. These sensors are designed to detect a person and / or an animal near the moving seat who could be injured by the seat movement.

[0016] The described method according to the invention offers the advantage that a person intending to enter the vehicle can do so quickly and comfortably. By automatically moving the seat into the determined approach position before the entry process, entry is ensured without the person having to manually adjust the seat beforehand, which would represent a considerable loss of time.

[0017] A sensor system capable of detecting the predetermined surroundings of a motor vehicle and identifying individuals intending to enter can be implemented in various ways, for example, by at least one camera capable of detecting visible and / or infrared light, and / or by at least one radar sensor and / or proximity sensor and / or LiDAR sensor (Light Imaging, Detection and Ranging). The latter are devices that use laser light to measure the distance to objects. Radar sensors can measure the distance between a person and the motor vehicle, thereby detecting the intention to enter and avoiding false detections. A sensor system designed as a combination of such cameras and distance-measuring sensors can detect the intention to enter for multiple individuals in the vicinity of the motor vehicle, given a specific vehicle door and the seat behind it.The predetermined environment of the motor vehicle typically corresponds to the maximum detection range of the sensors used.

[0018] The invention also includes embodiments that offer additional advantages. Whenever this application refers to a seat, it naturally also means that the respective embodiment can be carried out for multiple seats by performing the same process steps of the respective embodiment for each seat.

[0019] One embodiment for recognizing the intention to enter is achieved by identifying, from several potential persons in the environment outside the motor vehicle, those persons who have an intention to enter, by analyzing for each person a movement path measured by the sensor system to determine whether the respective person is approaching a door of the motor vehicle behind which the seat is located.

[0020] In this embodiment, if at least two people are near the vehicle, it is possible to determine individually for each person whether they intend to enter the seat or not. This determination of entry intention is made possible by the sensor system measuring the positions and movements of each potential person, i.e., their trajectories, and using these measured positions and movements to identify those individuals who actually have a concrete intention to enter. This identification is achieved, in particular, by checking whether the person is moving purposefully towards the vehicle door leading to the seat.

[0021] For this purpose, the sensor system can, for example, be designed as an ultra-wideband (UWB) system. A UWB system for positioning can be taken from the prior art, such as that known for keyless entry systems. In this system, very short pulses are transmitted across a wide frequency spectrum by a UWB transmitter and received by a UWB receiver. By using the wide frequency band, the position of one or more people can be determined. In the context of this invention, typically at least three UWB transmitters are installed in the motor vehicle, while one UWB receiver is located in a key for the motor vehicle and / or in a mobile device carried by a person in the vicinity of the motor vehicle. By continuously repeating the positioning process, not only the position but also the movement of the respective person can be tracked.

[0022] This embodiment is particularly advantageous because the distinction between persons intending to enter and those without such intentions ensures that the automatic presetting of the seat only occurs when and for the person who actually wants to enter, while simultaneously preventing unintended seat presetting, which ultimately increases the acceptance of the invention.

[0023] One embodiment relates to the process step of determining the individual seating position for each person with a detected intention to enter. This involves determining the respective individual seating position by means of a morphological scan to ascertain the body measurements of the respective person and calculating the individual seating position from these measurements.

[0024] Morphological scanning is a technique that allows for the precise estimation of a person's height and build, and serves to determine the body dimensions relevant for the automatic seat pre-adjustment. The estimation can be performed in various ways, for example, by taking photos or videos of the individual with cameras and estimating the body dimensions from the recordings using photogrammetry or structured light analysis. Photogrammetry is a measurement technique that makes it possible to obtain three-dimensional information about people or objects from two-dimensional photographs. To estimate body dimensions, photos of a person are taken from different angles; for this purpose, the sensor system may have multiple cameras.Software then identifies distinctive points on the body in the camera images and calculates a 3D model of the person, from which various body measurements such as height, shoulder width, torso depth, and / or leg length are precisely derived. This technique is particularly advantageous because it is contactless and the body measurements are taken quickly. The structured light method is an optical 3D measurement technique that projects structured light patterns (typically parallel stripes) onto an object and records their distortion from a different angle using a camera. When applied to people, a known stripe pattern is projected onto the body. The camera, positioned at a different angle than the projector of the stripe pattern, captures the distorted pattern as its image, which results from the irregular surface of the human body.From the way the projected stripes are deformed by the contours of the body, software can calculate the precise topography of the body surface in a known manner. This results in a precise 3D scan of the person, from which the body measurements for adjusting the seat can then be determined. The advantage of the structured light method is its independence from ambient lighting and its feasibility with only a single camera.

[0025] This embodiment is particularly advantageous because, through the use of morphological scanning, especially by means of photogrammetry or structured light scanning, the body dimensions relevant for seat adjustment can be determined automatically and very accurately without the person having to have physical contact with sensors or manual measurements being required. This, in turn, allows the respective individual seating position to be determined precisely.

[0026] Another embodiment concerning the method step for determining the individual seating position for each person with a recognized intention to enter results from the fact that, in the event that at least one person with a recognized intention to enter is assigned a digital user profile in a database, which contains a set of position data defining the individual seating position of this person and / or the body measurements of this person, the individual seating position for this person is determined by retrieving the set of position data from the database according to the user profile of the respective person and / or generating the seating position from the body measurements stored in the database.

[0027] In other words, if a user profile already exists in a database for one or more persons with a recognized intention to enter, which already contains one or more determined seating positions and / or the body measurements of the respective person, these existing entries will be used to determine the individual seating position for these persons, using the most up-to-date set of position data and / or the most up-to-date body measurements.

[0028] Such a database entry, which represents a user profile along with an individual seating position, can be created via a predefined personalization setting, for example, by a person manually entering a personalized seating position on the vehicle display, which should be used for them during the next entry process.

[0029] Any given seating position is uniquely defined by a set of positional data. This set of positional data comprises the specific values ​​of all relevant seat adjustment parameters necessary to achieve the individual seating position adapted to the person's body measurements. Thus, each set of positional data forms a unique signature for that seating position.

[0030] The digital user profile is a collection of data stored in a database, specifically assigned to an identified individual. It contains information relevant for the automatic seat presetting, such as one or more sets of previously determined seat positions and / or the individual's body measurements. Entries in the database can be made manually by a user, for example, by manually entering their body measurements at an input interface in the vehicle or in an app on a mobile device. In this context, the database is a digital storage medium and can be implemented in various ways, such as an in-vehicle database or a cloud database on an external server. For the correct user profile, and thus the appropriate seat position, to be retrieved from the database, the individual must be uniquely identified.This matching process typically works via wireless communication with a digital key or other paired mobile device carried by the individual. Once the vehicle receives the unique signal or certificate from this device, it can directly link the individual to the corresponding user profile in its database. Possible identification methods include a digital key (cryptographic smartphone certificate) directly linked to the user profile; a physical radio key with a unique ID assigned to the user profile within the system; and a paired mobile device, such as a connected smartphone or smartwatch with a unique identifier, identified via a wireless data communication connection, such as Bluetooth or Wi-Fi.In this way, the correct user profile can be identified in the database and the correct individual seating position and / or body measurements of the respective person can be retrieved.

[0031] The advantage of this design is that it enables a quick and precise determination of the individual seating position without requiring the collection of new data or calculations each time. If a digital user profile with saved seating positions or body measurements already exists for a person, the system can retrieve or generate the saved seating position directly and efficiently from the database. This ensures precise adaptation to the individual preferences of the person and avoids potential time losses or inaccuracies that could occur when determining the position again.

[0032] In a further embodiment relating to the method step for determining the individual seating position for each person with a detected intention to board, it is provided that if a person with a detected intention to board does not have a digital user profile assigned to them in a database, which contains at least one set of position data defining that person's individual seating position and / or their body dimensions, a new digital user profile is created in the database, the created user profile is assigned to that person, the seating position adapted for that person is determined by means of a morphological scan to determine that person's body dimensions from which their seating position is determined, and a new set of position data defining the determined seating position is added to the newly created user profile.

[0033] This implementation is particularly relevant for situations where people want to enter the vehicle for the first time, as a user profile usually does not yet exist for these individuals. To determine the appropriate seating position in such a case, the person's body measurements are first taken before they enter the vehicle. Then, the individual seating position is determined from these measurements and added to the newly created user profile in the database.

[0034] This design is particularly advantageous because it allows for the automatic determination of the individual seating position for new or previously unknown individuals, which is then stored in the database along with the user profile and available for future entry procedures. Otherwise, a person without a stored profile would have to adjust the seat manually in order to even be able to enter the vehicle, which can be quite time-consuming.

[0035] In one embodiment, it is provided that, in the event that a person has manually adjusted the seat, which was set to the approach position, to a new seating position before the boarding process, a new set of position data defining the new, manually set seating position is determined and the newly determined set of position data is added to the digital user profile associated with that person.

[0036] If the automatically preset seat position was incorrect and the boarding person had to adjust the seat themselves, this manual adjustment will be recorded in the user profile in the database, so that the appropriate seat position for that person can be selected the next time they want to board.

[0037] The advantage of this design is its ability to detect manual adjustments and take them into account during subsequent entry procedures. This makes determining the seating position more accurate over time.

[0038] Another embodiment regarding the determination of the respective individual seating position results from the fact that, in the event that no body measurements can be determined for at least one person with a recognized intention to enter, the individual seating position for the respective person is calculated on the basis of generic data, which includes statistical average values ​​of a population, ergonomic standard settings or manufacturer-predefined standard values.

[0039] If, for any reason, no body measurements are available, a default setting is used for the seating position. Preferably, this default seat setting is based on anthropological data for relevant percentiles in a specific world region. In addition to average height and / or build, extremely short and tall individuals are also taken into account, for example, the smallest and largest 5% of a Gaussian distribution, to ensure occupancy for the majority of a population. This approach is relevant in situations where no information whatsoever is available about the person boarding. This is typically the case for a newly delivered vehicle or after the seat's stored settings have been reset to the factory defaults.

[0040] The advantage of this design is the guarantee of automatic presetting, even when a person's individual body measurements are unknown. Without this feature, a suitable seating position could not be preset in such cases. By using statistical averages, ergonomic standards, or manufacturer-specified values, this design ensures that a seating position is set that allows at least a large portion of a representative population to enter the vehicle without having to manually adjust the seat beforehand.

[0041] In one embodiment, the calculation rule stipulates that the intermediate position corresponds to the individual seating position identified for the tallest person with a detected intention to enter. If an intention to enter is detected for two or more people, the seat is adjusted to the individual seating position determined in a previous step for the tallest person with a detected intention to enter, or to an average of the previously determined individual seating positions.

[0042] The first option has the advantage of ensuring that, even if the boarding intention is unclear, the seat is preset so that everyone can still board, but at the same time, the seat is not set to its extreme position. The advantage of the middle setting is that neither the shortest nor the tallest person is disproportionately disadvantaged.

[0043] One embodiment of the method provides that, from several potential persons in the environment outside the motor vehicle, precisely the person who has an actual intention to enter is identified by using a machine learning model acting as an identification model, which is trained using situation- and behavior-specific training data that describe past entry situations and the respective behavior of the persons involved and include an approach position observed for the previous entry process, and is trained by this process to identify the person who has an actual intention to enter for the current entry process.

[0044] If more than one person is near the vehicle, the machine learning model, acting as an identification model, decides which of these people will actually get into the vehicle. This means that in a subsequent step, an individual seating position only needs to be determined for that person, to which the seat is then adjusted.

[0045] The identification model is designed as a machine learning model. Generally speaking, a machine learning model receives a given input and calculates an output based on it. It uses internal algorithms and methods to do this. In this implementation, the given input consists of characteristics describing the current entry situation. These characteristics include the location and time of the current entry process, as well as information about the people involved. Based on this input, the identification model decides which of the people involved in the current entry situation will actually enter. This calculated output can, for example, be presented as a probability value.

[0046] The internal algorithms and methods result from the specific design of the machine learning model. Here, the identification model can be implemented, for example, as an artificial neural network or a decision tree model. Furthermore, such a model can be trained using a training algorithm based on training data. Such a training algorithm could, for example, be based on the backpropagation algorithm. The training data for this model consists of situation- and behavior-specific training data that describes past boarding processes and the respective behavior of the people involved, and includes an observed approach position for each boarding process.

[0047] The training data for this model consists of characterizing data describing past vehicle entry situations. In addition to the location and time of each entry, it includes information on how many people were involved, how the individuals behaved throughout the entire entry process—from the moment they were detected by the sensor system until a person actually sat down—and the seat position to which it was automatically adjusted and / or manually adjusted by the entering person. The identity of each individual is also relevant. This means that a specific person can be uniquely assigned to a profile or identifier, allowing for the recognition of recurring entries by the same person. This identity can be established in various ways.This often happens when a mobile device (a smartphone or smartwatch) sends a digital certificate or other identifier to the vehicle. However, a traditional key fob also has a unique ID that can be used to identify a specific person.

[0048] Through training, the identification model can recognize correlations and patterns between various parameters within the training data and the question of which person will actually get into the car in a given entry situation, especially in recurring situations. This will be illustrated with an example. Every morning, a father drives his daughter to school. When both father and daughter approach the car, this is a recurring situation with recognizable individuals, where it is clear which person will get into the driver's seat. The identification model can learn this recurring, repetitive pattern. When this situation occurs—that is, when father and daughter approach the car in the morning—the machine learning model outputs the learned person with the intention to get in, namely the father.

[0049] The advantage of this embodiment is that the entire process becomes faster and more efficient, because firstly, an individual seating position only needs to be determined for one person and not for all persons in the vicinity of the motor vehicle, and secondly, because no compromise position needs to be determined.

[0050] In a preferred embodiment, the calculation procedure includes the use of a machine learning model functioning as a compromise model to calculate the intermediate position, wherein the machine learning model functioning as a compromise model is trained using situation- and behavior-specific training data that describe past boarding situations and the respective behavior of the persons involved and each include an approach position observed for the previous boarding process, and is trained by the training to calculate the intermediate position for the current boarding process from the determined seating positions of the current boarding process.This means that instead of using a rigid, pre-programmed formula to calculate the intermediate position, a machine learning model, acting as a compromise model, is used to calculate a suitable approach position for the current entry process.

[0051] The compromise model, like the identification model, is a machine learning model. It is trained using the same training data (as described above). However, the compromise model differs from the identification model in its input and the output it calculates. Like the identification model, the compromise model receives as input, in addition to the characteristics describing the current boarding situation, particularly the location and time, all individual seating positions determined in a previous step for each person intending to board. The intermediate position is calculated from these seating positions according to the calculation formula. Furthermore, it is used in a different step of the process.While the identification model is used in the first procedural step to determine a clear, actual intention to enter, the compromise model is intended for the third procedural step, in which the intermediate position is calculated if an intention to enter has been detected for more than one person.

[0052] The trained compromise model is able to recognize non-trivial patterns, personal preferences, and correlations in entry situations and take these into account when calculating the appropriate intermediate position. For example, over time, it might become apparent that for a two-person household consisting of a man and a woman, both of whom use the vehicle, in situations where it cannot be clearly decided who will get in the seat, an appropriate intermediate position is not necessarily given by a simple arithmetic mean of the two individual seating positions, but rather by a weighted mean, in which, for example, the man's individual seating position is weighted more heavily than the woman's, or vice versa.The compromise model can also link the location and / or time of the current boarding process with a corresponding seating position. For example, an output of the compromise model can specify that at a particular location and / or time, if a specific group of people is detected, the seat will be adjusted to the individual seating position of the tallest person in that group.

[0053] The advantage of this design is that by using the compromise model to calculate the intermediate position, a suitable approach position can be determined for a wide variety of situations. A rigid calculation method, such as an average of the determined seating positions, does not take into account situation- and behavior-specific context, which in practice leads to a less comfortable and less reliable seat pre-setting.

[0054] In a preferred embodiment, the previously automatically preset seat is adjusted after an exit process to a subsequent approach position for a subsequent entry process, which is calculated by using a machine learning model, here referred to as a "prediction model", on the basis of situation- and behavior-specific data that describe entry processes and the respective behavior of the persons involved, before the intention to enter is recognized.

[0055] In other words, after a person exits the vehicle, the seat is prepared for the next person to enter. In many situations, it's possible to predict who will use the vehicle next. For example, a household of two people who share a vehicle on predetermined days. The prediction model can learn these recurring patterns and, based on this, predict a suitable seating position for the person it forecasts will enter the vehicle next.

[0056] The prediction model is also a machine learning model. It shares similarities with the identification model and the compromise model, but is not the same. Key differences lie in the input to the prediction model and its output. The input for the prediction model is information about the situation after the last passenger exit, specifically the location and / or time and / or the last passenger. From this input, a prediction is made about the approach position for the next expected entry and / or the person who will enter next. The prediction model establishes a predictive baseline setting, i.e., before any intention to enter is even detected, whereas the compromise model is used if an intention to enter has been detected for more than one person, meaning it is not clear which person will get into the car.

[0057] The prediction model is similar to the other two machine learning models in that it is also trained using situation- and / or behavior-specific data. The internal structure of the models can be the same or very similar; in particular, they use algorithms and computational rules to recognize patterns and relationships in the data.

[0058] This embodiment is advantageous for several reasons. It is efficient and resource-saving because the sensor system does not need to monitor the vehicle's surroundings and identify the intention to enter if the next driver can be predicted with a sufficiently high degree of probability. This embodiment naturally applies to entry processes that occur after the entry process described above, which performs the active environmental monitoring described. Furthermore, unnecessary seat adjustment is avoided. If it is highly probable which person will drive next, the seat does not need to be adjusted to an intermediate position. Finally, it increases the acceptance of the invention among people who find it irritating to see the seat automatically adjusted before they enter.This irritation is eliminated if the seat is adjusted before the person approaches the vehicle.

[0059] One embodiment of the identification model and / or the compromise model or the prediction model provides that the situation- and behavior-specific data are related to at least one geofencing event.

[0060] This means that the three machine learning models use location-based knowledge to calculate the approach position. Geofencing describes the automated triggering of an action by crossing a geolocated boundary on the Earth's surface or in the air. A geofence is an invisible, digital fence drawn around a specific geographical area. Whenever a registered device (such as a smartphone or the car itself) enters or leaves this virtual fence, a geofencing event is triggered. This event is a data point containing information about the location, time, and the person or device involved. For example, a geofence can be placed around a parking lot or garage. Here, a precise geographical area is defined (for example, using GPS coordinates and a radius). Entering or leaving the geofence is then registered as a geofence event.The data associated with this event can be used as training data for both models. Geofencing data typically describes an "if-then" relationship based on location.

[0061] This implementation is advantageous because geofencing data provides a direct indication of the vehicle's location and, consequently, the purpose of its stay. Furthermore, human behavior is often tied to specific locations. The models then learn these connections between specific locations and / or behaviors and the behaviors occurring at those locations and / or times. While other data, such as pure smartphone movement data, are also essential, they only gain contextual significance when combined with geofencing data.

[0062] A further embodiment of the method according to the invention with regard to the identification model and / or the compromise model and / or the prediction model results in the identification model and / or the compromise model and / or the prediction model being updated by performing an evaluation of the set approach position by a person who has entered the motor vehicle and / or by recording an actual seating position that was manually set by the person who entered the motor vehicle immediately before entering the motor vehicle.

[0063] Feedback allows machine learning models to be continuously improved. Two methods are provided for this: explicit and implicit feedback. After a person enters the vehicle, they directly evaluate the preset driving position. Explicit evaluation can occur in several ways, such as via voice command, touchscreen input, or button press on the steering wheel or seat. The evaluation can be presented as a yes / no answer, a scale, or a specific verbal description. Feedback on the preset driving position can also be implicit. Any manual adjustment of the seat by the user after the automatic preset position—that is, if the person entering the vehicle could not get in without manually adjusting the seat—is considered a correction and thus feedback.Similarly, an assessment of the approach position is implicitly made if the seat has not been manually adjusted.

[0064] The advantage of this design is that the models are continuously improved through feedback and adapted to the everyday life of the respective driver or drivers, thus allowing individual preferences to be better taken into account.

[0065] One embodiment of the inventive method with regard to calculating the approach position is that, in the case that an intention to enter has been detected for exactly one person, the approach position corresponds exactly to the seating position that was determined for this person in a previous step.

[0066] In this embodiment, it is stipulated that if an intention to enter is detected by exactly one person, the resulting approach position of the seat corresponds directly to the individual seating position previously determined for that specific person. This means that in a situation without ambiguity, the need for a compromise is eliminated.

[0067] This embodiment of the method is advantageous because it simplifies and accelerates the process as soon as a clear situation is identified. If it is clear that exactly one person intends to enter the vehicle, further calculations for the approach position are unnecessary. The previously determined seating position for this person can be used directly as the entry position for the seat without any further intermediate steps. This minimizes setup time and simultaneously ensures that the person can actually enter the vehicle without manually adjusting the seat.

[0068] One embodiment of the method is characterized by the fact that, after a completed entry process for the person who has entered the motor vehicle and sat down on the preset seat, an intention to exit is detected and then the previously set seat is moved to an exit position, wherein the exit position corresponds to the individual seating position determined for the person who entered, or to an actual seating position manually set by the person who entered the motor vehicle immediately before entering the motor vehicle.

[0069] In a manner analogous to how the seat is preset to facilitate entry, in this embodiment the previously set seat is moved to an exit position to make it easier for the person who has entered to get out. For the seat to move to the exit position, the sensor system must detect when the person sitting in the seat intends to exit. This is the same sensor system that previously detected the person's intention to enter. To detect the intention to exit, at least one door lock sensor and / or at least one seatbelt buckle sensor are used. The door lock sensor detects the person's intention to exit by measuring when the person opens the corresponding door. The seatbelt buckle sensor detects the person's intention to exit by measuring when the seatbelt is engaged.Upon detection of the intention to exit, the seat is moved to the approach position. This approach position corresponds either to the individual seating position previously determined for that person or to a position manually set by that person immediately before the previous entry. The approach position can be identical to the seat's driving position or different from it. Because the approach position is determined by the individual seating position, it is automatically ensured that the body dimensions of the exiting person are taken into account.

[0070] This embodiment is advantageous because it provides an additional comfort feature for the person entering the vehicle. By pre-setting the seat to the exit position, the person can exit comfortably without having to adjust the seat themselves, which would otherwise result in a considerable loss of time.

[0071] Also part of the invention is a motor vehicle with a system for automatically presetting at least one seat in the motor vehicle, wherein the system is configured to perform a method according to one of the preceding claims.

[0072] The invention also includes combinations of the features of the described embodiments. The invention therefore also includes realizations that each exhibit a combination of the features of several of the described embodiments, provided that the embodiments have not been described as mutually exclusive.

[0073] The following are exemplary embodiments of the invention described. This is illustrated by: Fig. 1 a schematic representation of an embodiment of the method according to the invention; Fig. 2 schematic representations of a motor vehicle according to the invention and of a further embodiment of the method according to the invention; Fig. 3 schematic representations of a motor vehicle according to the invention and of a further embodiment of the method according to the invention.

[0074] The exemplary embodiments described below are preferred embodiments of the invention. In these exemplary embodiments, the described components each represent individual features of the invention, which can be considered independently of one another and each further develops the invention independently. Therefore, the disclosure is intended to include combinations of features of the embodiments other than those shown. Furthermore, the described embodiments can also be supplemented by further features of the invention already described.

[0075] In the figures, identical reference symbols denote functionally equivalent elements.

[0076] Fig. Figure 1 shows a schematic representation of an embodiment of a method for automatically presetting a seat 11 in a motor vehicle 13 before an entry process. The illustrated method comprises several process steps. In a first step S1, an intention to enter is detected for at least one person in the vicinity of the motor vehicle 13, whereby the respective intention to enter is detected by a sensor system. The sensor system can be implemented, for example, by several cameras 15 for visible light and / or infrared light and / or LiDAR sensors 17.Particularly advantageous for this process step is the use of a UWB system consisting of at least three UWB transmitters 19 and at least one UWB receiver, because this system is capable of detecting a movement path for each potential person in the vicinity of the vehicle 13. This allows the intention to enter to be recognized if, for example, the movement path leads directly to a door of the vehicle 13. In certain situations, there may be several potential persons with an intention to enter, so it cannot be clearly determined for which person the seat 11 should be preset. In certain situations, a machine learning model acting as an identification model can be used to precisely identify the person who will actually enter the seat 11.In a subsequent process step S2, an individual seating position is determined for each person with a previously identified intention to enter the vehicle. This individual seating position depends on the person's body dimensions, particularly their height and build, and can be uniquely defined by a specific set of positional data. Determining the seating position can be done in various ways. If a person wishes to use the vehicle 13 for the first time, their body dimensions are determined through morphological scanning, from which their individual seating position can be calculated. The determined seating positions can also be added to a digital user profile in a database, which is linked to the respective person. In a subsequent step S3, an approach position is calculated.First, a criterion K is used to determine how an approach position is calculated. Criterion K distinguishes between two cases: whether an intention to board has been detected for exactly one person (n=1) or for more than one person (n>1). In the trivial case where an intention to board has been detected for only one person, step S3.1 stipulates that the approach position corresponds exactly to the individual seating position determined for the person with the intention to board in a previous step. Otherwise, the approach position corresponds to an intermediate position calculated in step S3.2 from the previously determined individual seating positions according to a predefined calculation rule.In a preferred embodiment, a compromise model is used to calculate the intermediate position. This model is implemented as a machine learning model, trained using situation- and behavior-specific data, and designed to calculate a suitable intermediate position for the current boarding situation from the determined seating positions. After the approach position has been calculated, seat 11 is adjusted to the calculated approach position in step S4. Steps S3.1 or S3.2 can be updated by the boarding passenger providing a rating B for the adjusted approach position. If only one person intended to board and manually adjusted seat 11 to facilitate boarding, the newly adjusted seat position can be added to that person's user profile in the database and thus taken into account during the next boarding process. If, according to S3.Once an intermediate position has been calculated, the evaluation B can be performed explicitly, by having the person entering the vehicle evaluate the set approach position, or implicitly, by registering whether the person entering the vehicle manually adjusted the seat 11 immediately before entering. Evaluation B allows for the continuous improvement of process step S3 for calculating the approach position. In the final step S5 of this embodiment of the method according to the invention, after an exit process, the seat 11 is set to a subsequent approach position for a subsequent entry process. This subsequent approach position is calculated by a prediction model before any new entry intention is even detected. The prediction model is also designed as a machine learning model and can predict which person will enter next and calculate an approach position accordingly.

[0077] Fig. Figure 2 shows schematic representations of a motor vehicle 13 according to the invention and a further embodiment of the method according to the invention. Two persons are in the vicinity of the motor vehicle 13, with person A 25 approaching the seat 11 of the motor vehicle 13 and intending to enter it, while person B 27 is moving away from the motor vehicle 13. In this example, the sensor system of the motor vehicle 13 comprises two cameras 15 and two LiDAR sensors 17, as well as three UWB transmitters 19. In the first step S1 of the illustrated embodiment of the method according to the invention, the sensor system can detect that person B 27 is moving away from the motor vehicle 13, i.e., does not intend to enter it, and that person A 25 is moving directly towards it. Consequently, an intention to enter is detected for person A 25, and in a subsequent step S2, an individual seating position is determined for person A 25.Since person A 25 will be entering vehicle 13 for the first time, their body dimensions must be determined by morphological scanning. For this purpose, several images of person A 25 are taken by cameras 15, and using photogrammetry and structured light scanning, person A 25's body dimensions, in particular their height and build, as well as the ratio of arm to leg length, can be determined from the images. A new digital user profile is also created for person A 25 in a database, and the determined seating position is added to the user profile so that the determined seating position can be retrieved the next time person A 25 enters the vehicle. The next process step S3.1, calculating an approach position, is trivial in this example, since an intention to enter was only detected for one person (n=1). Thus, the approach position is simply the previously determined seating position.Therefore, in a subsequent step, S4 is adjusted to the determined seating position.

[0078] In Fig.Figure 3 again shows a schematic representation of a motor vehicle 13 according to the invention and a further embodiment of the method according to the invention. Three people are in the immediate vicinity of the motor vehicle 13, with person A 25, person B 27, and person C 29 all moving towards the motor vehicle 13. In this example as well, the sensor system of the motor vehicle 13 comprises two cameras 15 and two LiDAR sensors 17, as well as three UWB transmitters 19. Each of the three people is also carrying a mobile device (smartphone). The mobile devices are each equipped with a UWB receiver. The UWB receivers can receive short, broadband radio pulses transmitted by the UWB transmitters 19 of the motor vehicle 13. They register the respective time at which these pulses arrive.In this way, the respective time interval that the signal took to travel from the transmitter to the receiver, and thus the respective distance between transmitter and receiver, can be calculated. Based on the distances to the three transmitters, the exact position of the mobile device can be calculated using trilateration. By continuously repeating this position measurement, the movement path of all three people can be measured using the UWB method, and it can thus be determined that all three people are moving towards the vehicle 13. In a first step S1 of the procedure, it is therefore recognized that all three people intend to enter the vehicle. It is therefore not yet clear which person will sit in seat 11. In the next step S2 of the procedure, an individual seating position is determined for each person.For this purpose, digital user profiles stored in an in-vehicle database, containing a set of positional data that defines an individual seating position, can be used. Because an identifying cryptographic certificate is installed on each mobile device, the appropriate user profile can be found for each person. In the next step of the exemplary procedure, an approach position is calculated from the three determined seating positions. Since an intention to enter was detected for more than one person (n=3), the approach position corresponds to an intermediate position, which is calculated in step S3.2 according to a formula based on the three determined seating positions. A preferred formula involves using the compromise model to calculate the intermediate position.The compromise model calculates a suitable intermediate position for this situation based on the determined individual seating positions. In a subsequent step, S4, seat 11 is adjusted to this intermediate position, allowing all three people to enter the vehicle. The person who has entered (for example, person C 29) can provide feedback (B) on the adjusted seating position, for example, by entering an input on a display in the vehicle. This allows the compromise model to be improved. After the person who entered has exited the vehicle, in a final step, S5, of the exemplary procedure, seat 11 is moved to a new position, with the new seating position being calculated by the prediction model.

[0079] Overall, the examples show how an automatic presetting of a seat 11 in a motor vehicle 13 can be implemented before an entry process in various situations. QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] DE 196 30 189 A1

[0008] DE 10 2006 052 087 B4

[0009] DE 10 2022 111 362 B3

[0010]

Claims

[1] Method for automatically presetting a seat (11) in a motor vehicle (13) prior to an entry procedure, comprising the following steps: - for at least one person located in a predetermined environment outside the motor vehicle (13), an intention to enter is detected by a sensor system that detects the predetermined environment of the motor vehicle (13) (S1); - For each person for whom an intention to board has been detected, an individual seating position is determined, which depends on the body dimensions of the respective person (S2); - an approach position is determined (S3), whereby in the event that an intention to board has been detected for more than one person, the approach position corresponds to an intermediate position which is calculated from the previously determined individual seating positions using a predefined calculation rule (S3.2); - the seat (11) is adjusted to the determined approach position (S4). [2] Method according to claim 1, wherein from several potential persons in the environment outside the motor vehicle (13) those persons who have an intention to enter are identified by analyzing for each person a movement path measured by the sensor system to determine whether the respective person is approaching a door of the motor vehicle (13) behind which the seat (11) is located. [3] Method according to one of the preceding claims, wherein the respective individual sitting position is determined by determining the body dimensions of the respective person by means of a morphological scanning and calculating the individual sitting position from the determined body dimensions. [4] Method according to one of the preceding claims, wherein, in the event that at least one person with a recognized intention to enter is assigned a digital user profile in a database which contains a set of position data defining the individual seating position of that person and / or the body measurements of that person, the individual seating position for that person is determined by retrieving the set of position data from the database according to the user profile of the respective person and / or by generating the seating position from the body measurements stored in the database. [5] Method according to one of the preceding claims, wherein in the event that a person with a recognized intention to enter is not assigned a digital user profile in a database which contains at least one set of position data defining the individual seating position of that person and / or the body dimensions of that person, • a new digital user profile created in the database, • the created user profile assigned to this person, • the seating position adapted for this person is determined by determining the person's body measurements through a morphological scan, from which the seating position of this person is determined, and • A new set of position data, defining the determined seating position, is added to the newly created user profile. [6] Method according to one of claims 4 or 5, wherein, in the event that a person has manually adjusted the seat set to the approach position to a new seat position before the boarding process, a new set of position data defining the new, manually set seat position is determined and the newly determined set of position data is added to the digital user profile associated with that person. [7] Method according to one of the preceding claims, wherein, in the event that no body measurements can be determined for at least one person with a recognized intention to enter, the individual seating position for the respective person is calculated on the basis of generic data which includes statistical averages of a population, ergonomic standard settings or manufacturer-defined standard values. [8] Method according to one of the preceding claims, wherein the calculation rule provides that the intermediate position corresponds to the individual seating position determined for the tallest person with a recognized intention to board, or to an average of the previously determined individual seating positions. [9] Method according to one of the preceding claims, wherein from several potential persons in the environment outside the motor vehicle (13) exactly the person who has an actual intention to enter is identified by using a machine learning model acting as an identification model, which is trained by means of situation- and behavior-specific training data that describe past entry situations and the respective behavior of the persons involved and each include an approach position observed for the entry process at that time, and is trained by the training to identify the person who has an actual intention to enter for the current entry process. [10] Method according to one of the preceding claims, wherein the calculation procedure comprises the use of a machine learning model functioning as a compromise model to calculate the intermediate position, wherein the machine learning model functioning as a compromise model is trained by means of situation- and behavior-specific training data that describe past boarding situations and the respective behavior of the persons involved and each include an approach position observed for the boarding process at that time, and is trained by the training to calculate the intermediate position for the current boarding process from the determined seating positions of the current boarding process. [11] Method according to one of the preceding claims, wherein after an exit process the at least one seat (11) is set into a subsequent approach position for a subsequent entry process, which is calculated by using a machine learning model acting as a prediction model on the basis of situation- and behavior-specific data observed in the past, which describe entry processes and the respective behavior of the persons involved, before the intention to enter is recognized (S5). [12] Method according to any one of claims 9 to 11, wherein the situation- and behavior-specific data are related to at least one geofencing event. [13] Method according to any one of claims 9 to 12, wherein the identification model and / or the compromise model and / or the prediction model are updated by performing an evaluation of the set approach position by a person who has entered the motor vehicle (13) and / or by capturing an actual seating position that was manually set by the person who entered the motor vehicle (13) immediately before entering the motor vehicle (13). [14] Method according to one of the preceding claims, wherein, in the event that an intention to enter has been detected for exactly one person, the approach position corresponds exactly to the seating position determined for that person in a previous step. [15] Method according to one of the preceding claims, wherein, after a completed entry process for the person who has entered the motor vehicle (13) and sat down on the preset seat (11), an intention to exit is detected and then the previously set seat (11) is moved to an exit position, wherein the exit position corresponds to the individual seat position determined for the person who has entered, or to an actual seat position manually set by the person who has entered the motor vehicle (13) immediately before entering the motor vehicle (13). [16] Motor vehicle (13) with a system for automatically presetting at least one seat (11) in the motor vehicle (13), wherein the system is configured to perform a method according to one of the preceding claims.

Images