DE102021211739A1 - Determination of the position of the sternum - Google Patents

Abstract

The invention relates to a method for determining a sternum position (36) of an occupant (24) of a motor vehicle (42), a corresponding computer program, and a corresponding device (10) and a system (18).
It is provided that a position of the sternum (36) is determined based on anatomical relationships. This is based on data (44) such as vehicle settings of a driver (24) and/or sensor data with information about the driver (24). This data provides information about a body relation of the driver (24) based on which the position of the sternum (36) is determined by means of a cl and/or an overdetermined system of equations.

Description

The invention relates to a method for determining a sternum position of an occupant of a motor vehicle, a corresponding computer program and a corresponding device and a system.

Every year, quite a number of drivers of vehicles suffer a heart attack/cardiac arrest while driving. As a result, serious accidents often occur. In the future, cardiac arrest can be reliably detected by using interior sensors in the vehicle. After cardiac arrest has been detected, resuscitation can be initiated in the vehicle. The usual cardiac massage should preferably be done using the seat belt; If necessary, a pressure body can also be attached to the belt. Proper positioning relative to the occupant's sternum is very important for the sling, possibly with the compression hull, to perform effective resuscitation; According to medical guidelines, cardiac massage should be applied in the lower half of the sternum, i.e. the breastbone.

From the disclosure document US 2021/0069 521 A1 For example, a defibrillation system is known that includes a cardiac arrest detector that detects whether cardiac arrest has occurred in a passenger while he is seated in an automotive vehicle with a seat belt of the vehicle fastened. Further, the defibrillation system includes a passenger posture detection device that detects a posture of the passenger; a seat drive device that changes the posture of the passenger to a lying posture of the passenger based on a detection signal of the passenger posture detection device when cardiac arrest occurs in the passenger. Finally, the defibrillation system includes a heart position detector configured to search for a position of a heart of the passenger and a defibrillation robot for performing a CPR method or a method using an AED on the passenger's heart. Finally, the defibrillation system includes a controller that controls the operation of the seat drive device, the heart position detector, and the defibrillation robot based on detection signals from the cardiac arrest detector and the passenger posture detection device. The disadvantage here is that the passenger is first transferred to a lying position and only then is a heart position determined. In particular, the heart position is determined using an X-ray camera. The disclosed defibrillation system is therefore expensive and technically complex.

From the disclosure document U.S. 2020 010 4571 A1 an occupant modeling device is known. The occupant modeling device includes a capturing section that captures an image of an area where a face of an occupant is likely to exist. The occupant modeling device includes a model matching section that creates a model of the face based on a first image captured by the capturing section. Further, the occupant modeling device includes a tracking section that fits the model to a second image captured after the first image, and a determination section that determines correctness of a face portion position included in the second model-matched image using learned information certainly. The information is obtained through learning based on correct information and incorrect information about the face portion position. Finally, the occupant modeling device includes a processing section that determines whether to continuously execute a process in the tracking section or to re-execute a process in the model adjustment section according to a determination result in the determination section. The disadvantage here is that only one position of a face of the occupant is determined. From this, a position of the sternum cannot be determined or can only be determined very imprecisely.

From the disclosure document KR 102 096 617 B1 a multi-modal deep learning-based driver drowsiness detection system using an image and PPG data is known. The driver drowsiness detection system includes a pre-processing unit that extracts information about the position of eyes and mouth from a face image of a driver and at the same time collects photoplethysmogram (PPG) data of the driver. Further, the driver drowsiness detection system includes a multi-modal network that integrally fuses the eye and mouth position information with the PPG data, and a classification network that uses the generated fusion data as the eye and mouth position information to determine the drowsiness of the driver recognize the driver. Thus, taking into account all of the driver's image data such as the eyes and mouth and the photoplethysmogram (PPG) data, the drowsiness of the driver is to be recognized with high accuracy. The disadvantage here is that only drowsiness can be detected, but no cardiac arrest. In addition, the published application is silent regarding a determination of a position of the sternum.

The invention is now based on the object of creating a possibility of enabling improved detection of a position of a driver's sternum. In particular, a possibility is to be specified for determining a sternum position based on sensors that are already present.

• - Empfangen von Sensordaten mit Informationen zu dem Insassen;
• - Analysieren der Sensordaten und Abschätzen eines Körpermerkmals des Insassen basierend auf den Sensordaten; und
• - Ermitteln einer Sternumsposition des Insassen basierend auf dem abgeschätzten Körpermerkmal des Insassen und einer anatomischen Regel.

This object is achieved by a method for determining a sternum position of an occupant of a motor vehicle, comprising the steps:

• - receiving sensor data with information about the occupant;
• - analyzing the sensor data and estimating a body feature of the occupant based on the sensor data; and
• - determining a sternum position of the occupant based on the estimated body feature of the occupant and an anatomical rule.

The above object is also achieved by a computer program with program code means to carry out all the steps of a method as defined above when the computer program is executed on a computer, a motor vehicle control device or a corresponding computing unit.

• - einer Eingangsschnittstelle zum Empfangen von Sensordaten mit Informationen zu dem Insassen;
• - einer Analyseeinheit zum Analysieren der Sensordaten und Abschätzen eines Körpermerkmals des Insassen basierend auf den Sensordaten und Ermitteln einer Sternumsposition basierend auf dem abgeschätzten Körpermerkmal des Insassen und einer anatomischen Regel; und
• - einer Ausgangsschnittstelle zum Übermitteln der ermittelten Sternumsposition an eine Fahrzeugeinheit.

In addition, the above object is achieved by a device for determining a sternum position of an occupant of a motor vehicle, with:

• - An input interface for receiving sensor data with information about the occupant;
• - an analysis unit for analyzing the sensor data and estimating a body feature of the occupant based on the sensor data and determining a sternum position based on the estimated body feature of the occupant and an anatomical rule; and
• - an output interface for transmitting the determined sternum position to a vehicle unit.

• - einer Vorrichtung wie zuvor definiert, und
• - wenigstens einer Insassenkamera, einem Sensor zum Erfassen einer Sitzeinstellung des Insassen, einem Sensor zum Erfassen einer Lenkradeinstellung des Insassen, einem Sensor zum Erfassen einer Rückspiegeleinstellung des Insassen, einem Sensor zum Erfassen einer Außenspiegeleinstellung des Insassen, einem Sensor zum Erfassen einer Mittelarmlehneneinstellung des Insassen, einem Sensor zum Erfassen eines Gewichts des Insassen und/oder einem Sensor zum Erfassen eines Gurtbandauszugs.

Finally, the above object is achieved by a system for determining a sternum position of an occupant of a motor vehicle, with:

• - a device as defined above, and
• - at least one occupant camera, a sensor for detecting a seat adjustment of the occupant, a sensor for detecting a steering wheel adjustment of the occupant, a sensor for detecting a rearview mirror adjustment of the occupant, a sensor for detecting an exterior mirror adjustment of the occupant, a sensor for detecting a center armrest adjustment of the occupant, a sensor for detecting a weight of the occupant and/or a sensor for detecting a belt webbing extension.

A cost-effective device can be created by an input interface, which can preferably be used with existing systems, modules and units installed in the motor vehicle. An input interface can be wired and/or preferably wireless and preferably support one or more communication protocols. The analysis unit is preferably designed to carry out the steps of the method. By determining a sternal position of the occupant based on an estimated body feature of the occupant and an anatomical rule, the sternal position can be reliably determined. In particular, body features can be reliably and accurately estimated by various measurements and combinations of the measurements. Deriving a sternum position based on body features and an anatomical rule allows for an accurate determination of the sternum position without having to record it directly with a measurement. It goes without saying that the device can also be created purely by software. In particular, it is conceivable to create the device in the form of a corresponding reprogramming of a motor vehicle control unit.

In a preferred embodiment of the invention, it is provided that the sensor data includes information relating to one or more of the following cockpit settings: position of a seat, position of a steering wheel, position of a headrest, setting of exterior mirrors, setting of a rear-view mirror, setting of a center armrest; and/or one or more of the following occupant-specific measurements: occupant weight, length of a webbing extension from a retractor; and estimating a body characteristic of the occupant comprises estimating at least one or more of the following body characteristics of the occupant: a height, a leg length, proportions, an arm length, a shoulder-to-crown distance, a gender. The settings in the cock pit can be recorded quickly and easily from a technical point of view. In particular, these settings remain constant most of the time, so analyzing these settings once may be sufficient to estimate at least one physical feature of the occupant. Furthermore, there are preferably a large number of different linear dimensions from which the position of the sternum can be determined. It cannot necessarily be assumed that every occupant or driver configures the settings that are optimal for them. A most probable sternum position can be determined from the large number of recorded cock pit settings and/or occupant-specific measurements. In particular, the position of the seat surface in the longitudinal direction, in height and in the inclination can be detected, especially if there is an electric seat adjustment. Furthermore, the position of the steering wheel, in particular how far it is extended and at what angle the steering wheel is tilted to the driver in relation to the plane of the vehicle, can be detected. Furthermore, the position of the headrest in the longitudinal direction, in height and in inclination can be recorded. The adjustment of the two exterior mirrors so that perpendiculars from the center of the mirror surfaces should intersect at the level of the driver's eyes can be detected. The adjustment of the rear view mirror can be detected in relation to the head position.

In a further preferred embodiment of the invention, it is provided that determining a sternum position includes taking into account a large number of pieces of information in an overdetermined system of equations; and determining a most probable sternum position by interpolation based on the overdetermined equation system. In this way, a large amount of information, which is ideally redundant, can be processed in a technically reliable manner and deviations in information or a measurement error can be compensated for. A highly reliable determination of the sternal position can take place.

In a further preferred embodiment of the invention, it is provided that the sensor data include data from an occupant camera, which is preferably arranged in an instrument cluster and/or a headliner. As a result, an interior camera, for example in the instrument cluster or roof liner, can be used, which can already be present for driver observation for autonomous driving functions. An inexpensive method for determining the position of the sternum can be created.

In a further preferred embodiment of the invention it is provided that a body feature of the occupant, in particular a position of the face, a position of the neck, a position of the shoulders, particularly preferably a position of the clavicle, a position of the upper arms and / or the elbows, by means of a Image recognition algorithm, preferably by means of edge detection, area detection and / or color recognition is estimated. Using this information and knowing the anatomical relationships, the position of the sternum can then be estimated. In contrast to a deep learning approach, more assumptions are necessary here, for example about the anatomical relationships. In addition, a higher development effort is required in order to identify and precisely measure the body parts using edge, surface detection and color recognition. On the other hand, it is advantageous that significantly fewer images are required for the development and the estimation of the sternum position can also be carried out better, since the influence of image features is known much more precisely. Preferably, no further sensors, for example in the seats, are necessary. It goes without saying that, alternatively or additionally, an infrared camera can also be used in order to estimate the position of the sternum based on the temperatures in the thoracic area.

In a further preferred embodiment of the invention, it is provided that the physical feature of the occupant is estimated and a position of the sternum is determined by means of an AI, in particular by means of a deep neural network. It is particularly advantageous here that few assumptions to be defined have to be made. An AI learning method uses only image data to determine the position of the sternum. The exact anatomical relationships between different body regions do not have to be known, but a large number of different photos do, so that the network can learn the connections automatically. As a result, the network learns the anatomical rules without having to specify or know the anatomical rules precisely. The anatomical rule is inherently mapped by the structure/weights of the AI.

In a further preferred embodiment of the invention, it is provided that the class can use images of different occupants, in particular occupants of all genders, occupants in different clothing, occupants in different seating positions, particularly preferably in different lighting situations, from the perspective of the occupant camera and information about the Sternum position of the respective occupant is taught. For example, a deep neural network (deep learning) can be trained as part of vehicle development, with network architectures for image recognition systems preferably being used. As part of the training, a large number of images of different drivers from the perspective of the inside camera and information about the position of the sternum, for example as xy coordinates with radii or a bounding box on the image, are fed into the network. Using typical learning algorithms, this network now automatically learns the connection between image data and the position of the sternum. Images are particularly preferred used by many different drivers of all genders, in different clothes, seating positions and lighting situations in order to train a network that is as general as possible. It goes without saying that the trained neural network can be mapped on a control device for use in the vehicle, with the control device receiving images from the occupant camera and estimating the sternum position from them.

Further preferred configurations of the invention result from the remaining features mentioned in the dependent claims.

Unless stated otherwise in the individual case, the various embodiments of the invention mentioned in this application can advantageously be combined with one another. In particular, the combination can allow a more accurate assessment and, for example, compensate for inaccurate measurements of individual sensor values. Thus, a disadvantage of a typical mono inmate camera is that it has no depth information. Without additional information, it is therefore difficult for them to estimate how tall the driver is or whether he is just relatively close to the camera. One approach to finding the correct body size could be to measure known, seat-variable components. For example, the height in pixels of the headrest and the known height of the headrest in centimeters can be used to calculate how far it is from the camera. This enables a more precise estimation of the body size and thus also the position of the sternum.

This depth information can also be determined via the seat sensors, if these are available. If this depth information is available, images from the camera could be scaled appropriately to enable a deep learning approach or feature extraction to make more accurate estimates.

Previous investigations, in particular within the framework of simulations carried out, have shown that the effectiveness of the cardiac massage using the belt system can be increased by using a pressure body. A greater chest compression can preferably be achieved with the same belt force. It goes without saying that the system for determining a sternum position of an occupant of a motor vehicle and/or the corresponding device uses an actuator to adjust a pressure body based on the determined sternum position and/or to initiate a cardiac massage can. The pressure body can either be integrated into the belt system, for example the belt strap or belt lock, or it can be designed as a separate component which is attached or guided to the belt strap or belt lock. The belt system can either be the conventional three-point belt or an additional belt system. So that the belt or the pressure body can achieve an advantageous effect in the case of a cardiac massage, the belt/pressure body should be positioned on the lower half of the sternum relative to the occupant. The position of the webbing route/pressure hull should automatically adjust to the position of the occupant's sternum. There is an adjustment depending on size, current occupant posture on the seat, etc. To do this, the position of the sternum must be determined. The position of the sternum can be determined continuously while the vehicle is being used, or it can only be determined when cardiac arrest has been detected, ie preferably before the start of resuscitation. During resuscitation, the position of the sternum should also be determined continuously so that the belt/pressure body can be repositioned if the position changes.

AI, machine learning or machine learning is a generic term for the "artificial" or machine generation of knowledge from experience: An artificial system learns from examples and can generalize them after the learning phase has ended. To do this, machine learning algorithms build a statistical model based on training data. This means that the examples are not simply learned by heart, but that patterns and regularities are recognized in the learning data. The system can also assess unknown data, a so-called learning transfer, or fail to learn unknown data, for example due to overfitting. In particular, in so-called unsupervised learning, a statistical model can be generated for a given set of inputs, which describes the inputs and contains recognized categories and relationships, thus enabling predictions. Deep learning is a method of machine learning that uses artificial neural networks with numerous intermediate layers between the input layer and the output layer, thereby forming a comprehensive inner structure.

The sternum, or breastbone, is a flat, sword-shaped bone in the center front of the chest where the ribs or their cartilaginous extensions attach. The sternum of the man is slimmer than that of the woman.

Under anatomical rule is in particular a connection between different bodies permas mean among themselves. By applying an anatomical rule, a position of the sternum, among other things, can be determined from various body characteristics, such as arm length or a ratio of arm length to shoulder width.

Body features are to be understood in particular as meaning measurable and/or estimable properties of individual body parts of a human body and/or human body parts. In particular, a body feature may include arm length, weight, neck length, head size, or the like.

• 1 eine schematische Darstellung einer Vorrichtung zur Bestimmung einer Sternumsposition;
• 2a bis 2c eine schematische Darstellungen eines Systems zur Bestimmung einer Sternumsposition und zur Reanimation;
• 3 eine schematische Seitenansicht einer Fahrzeuginnenraums mit einer Insassenkamera;
• 4 eine schematische Darstellung einer Methode zur Bestimmung einer Sternumsposition; und
• 5a und 5b schematische Darstellungen einer weiteren Methode zur Bestimmung einer Sternumsposition;
• 6 eine schematische Darstellung eines Kraftfahrzeugs mit einem System zur Bestimmung einer Sternumsposition; und
• 7 schematisch die Schritte eines erfindungsgemäßen Verfahrens.

The invention is explained below in exemplary embodiments with reference to the associated drawings. Show it:

• 1 a schematic representation of a device for determining a sternum position;
• 2a until 2c a schematic representation of a system for determining a sternum position and for resuscitation;
• 3 a schematic side view of a vehicle interior with an occupant camera;
• 4 a schematic representation of a method for determining a sternal position; and
• 5a and 5b schematic representations of another method for determining a sternal position;
• 6 a schematic representation of a motor vehicle with a system for determining a sternum position; and
• 7 schematically shows the steps of a method according to the invention.

1 shows a device 10 according to the invention for determining a sternum position of an occupant, in particular a driver of a motor vehicle, with an input interface 12, an analysis unit 14 and an output interface 16. It is understood that the input interface 12 and the output interface 16 are also designed together in the manner of a transceiver could be. The separate representation chosen here is for a better overview.

The input interface 12 is designed to receive sensor data with information about the occupant. The sensor data can in particular include at least one of the following data: a position of a seat surface, a position of a steering wheel, a position of a headrest, a setting of exterior mirrors, a setting of a rear-view mirror, a setting of a center armrest; a weight of the occupant, a length of a belt extension from a belt tensioner or retractor.

Analysis unit 14 is designed to analyze the sensor data and to estimate at least one body feature of the occupant based on the sensor data. Further, the analysis unit 14 may determine a sternal position based on the estimated body feature of the occupant and an anatomical rule.

The output interface 16 is designed to transmit the determined sternum position to a vehicle unit.

In the 2a until 2c a system 18 for determining a sternal position and for resuscitation is shown schematically. The system 18 comprises a device 10 as in 1 described, as well as at least one sensor 20 and one vehicle unit 22.

The sensor 20 can in particular include at least one of the following sensors 20: a sensor 20 for detecting a seat adjustment of the occupant, a sensor 20 for detecting a steering wheel adjustment of the occupant, a sensor 20 for detecting a rearview mirror adjustment of the occupant, a sensor 20 for detecting an exterior mirror adjustment of the occupant, a sensor 20 for detecting a center armrest adjustment of the occupant, a sensor 20 for detecting a weight of the occupant and/or a sensor 20 for detecting a belt webbing extension.

The vehicle unit 22 can in particular include a unit for carrying out a resuscitation of the occupant 24 with a first retractor 26 and a second retractor 28 . A buckle 30 can serve as a pressure body 32 for resuscitation or heart compression massage.

In 2 B an initial situation is shown which corresponds to a normal driving situation. The occupant 24 is fixed by a belt, with the belt buckle 30 being located approximately at hip level of the occupant 24 and the belt being fixed on the opposite side by means of an end fitting 34 .

In 2c a resuscitation or cardiac massage is shown schematically. By means of the first retractor 26 and the second retractor 28, the pressure body 32 for resuscitation or cardiac massage can be moved to the transmitted sternal position in order to use the retractors 26, 28 to carry out resuscitation or perform chest compressions. It goes without saying that the position of the pressure hull 32 can be monitored by the system 18 and, if necessary, can be transmitted to the unit for carrying out a resuscitation. The unit for performing a resuscitation can then correct the position of the pressure hull 32 .

In 3 a vehicle interior with a sensor 20 in the form of an occupant camera is schematically illustrated in simplified form in a side view. The occupant camera detects the buckled occupant 24 and/or also a position of the pressure body 32. Based on the sensor data of the occupant camera, a sternum position of the occupant 24 can be determined, preferably continuously. In particular, a position of the pressure body 32 and/or the course of a belt strap relative to the sternum can be checked and repositioned if there are deviations from the desired position.

In 4 a method for determining a sternal position 36 is shown schematically. In the example shown, the sternum position 36 is determined by means of an artificial intelligence, Kl.

The AI includes a deep neural network 38. In this method, an indoor camera, e.g. as in 3 shown, used. The deep neural network 38 is connected to a variety of different drivers from the perspective of the inside camera as well as information about the position of the sternum. The trained neural network 38 estimates a sternum position 36 from occupant camera images, thereby inherently applying an anatomical rule.

The multitude of preferably includes images of different drivers of all genders, in different clothing, seating positions and lighting situations.

In the 5a and 5b another method for determining a sternum position 36 is shown schematically. The position of the sternum follows basic anatomical rules and is therefore within certain distance ranges in relation to other body features such as the shoulders or the position of the head. In this case, image recognition algorithms known in principle from the prior art are preferably used in order to recognize relevant body features, for example by means of edge, surface detection and color recognition.

In particular, a position of the face, for example in the form of a bounding box, a position of the neck, a position of the shoulders, possibly a position of the clavicle, a position of the upper arms and/or a position of the elbows can be recognized.

In the 5a and 5b the recognition of the body features is indicated by crosses. The sternum position 36 of the occupant 24 can then be estimated on the basis of this information and with knowledge of anatomical relationships or application of an anatomical rule.

In 6 a motor vehicle 42 with a system 18 for determining a sternum position 36 is shown schematically.

The system 18 is based on the use of one or more sensors 20 for detecting vehicle elements adjusted by the occupant 24 in order to estimate relevant body features or body proportions. For example, one or more of the settings of the following cockpit elements can be used: seat, steering wheel, headrest, exterior mirror, rear-view mirror, center armrest. Furthermore, a weight of the occupant 24 and/or a length of the seat belt extension from the retractor 26, 28 can be detected. At least one body feature can be estimated on the basis of this data 44 or a subset thereof and a sternum position 36 can be determined therefrom.

Preferably, several of the above items of information are recorded in the form of data 44 and the most probable sternal position 36 is calculated using an overdetermined system of equations, preferably by interpolation. In this way, an incorrect adjustment of a vehicle element by the occupant 24 can be compensated.

In 7 the steps of a method according to the invention for determining a sternum position 36 of an occupant 24 of a motor vehicle 42 are shown schematically. The method can preferably be carried out with an apparatus 10 and/or a system 18 as previously described. In particular, the method can be carried out on a control device for a motor vehicle 42 .

In a first step S1, sensor data with information about the occupant 24 is received.

Then, in a second step S2, the sensor data is analyzed and a physical feature of the occupant 24 is estimated based on the sensor data.

Finally, in a third step S3, a sternum position 36 of the occupant 24 is determined based on the estimated body feature of the occupant 24 and an anatomical rule.

Reference List

10

contraption

12

input interface

14

analysis unit

16

output interface

18

system

20

sensor

22

vehicle unit

24

inmate

26

first retractor

28

second retractor

30

belt buckle

32

pressure hull

34

end fitting

36

sternal position

38

neural network

40

variety of illustrations

42

motor vehicle

44

Data

S1-S3

process steps

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• US 2021/0069521 A1 [0003]
• US20200104571A1 [0004]
• KR 102096617 B1 [0005]

Claims (10)

Method for determining a sternum position (36) of an occupant (24) of a motor vehicle (42), comprising the steps of: - Receiving (S1) of sensor data with information about the occupant (24); - Analyzing (S2) the sensor data and estimating a body feature of the occupant (24) based on the sensor data; and - determining (S3) a sternum position (36) of the occupant (24) based on the estimated body feature of the occupant (24) and an anatomical rule. The method of the preceding claim, wherein the sensor data includes information related to one or more of the following cock pit settings: - position of a seat base, position of a steering wheel, position of a headrest, adjustment of exterior mirrors, adjustment of a rear-view mirror, adjustment of a center armrest; - and/or include one or more of the following occupant-specific measurements: weight of the occupant (24), length of a belt extension from a retractor (26, 28); and estimating a physical characteristic of the occupant (24) comprises estimating at least one or more of the following physical characteristics of the occupant (24): a height, a leg length, proportions, an arm length, a shoulder-to-crown distance, a gender. Method according to one of the preceding claims, wherein determining a sternum position (36) comprises considering a plurality of information in an overdetermined system of equations; and a most probable sternal position (36) is determined by interpolation based on the overdetermined equation system. Method according to one of the preceding claims, wherein the sensor data comprises data (44) from an occupant camera, which is preferably arranged in an instrument cluster and/or a headliner. procedure after claim 4 wherein a body feature of the occupant (24), in particular a position of the face, a position of the neck, a position of the shoulders, particularly preferably a position of the clavicle; a position of the upper arms and/or the elbows is estimated using an image recognition algorithm, preferably using edge detection, area detection and/or color recognition. procedure after claim 4 or 5 , wherein the physical feature of the occupant (24) is estimated and a sternum position (36) is determined by means of an AI, in particular by means of a deep neural network (38). procedure after claim 6 , the class using images of different occupants (24), in particular occupants (24) of all genders, occupants (24) in different clothing, occupants (24) in different seating positions, particularly preferably in different lighting situations, from the perspective of the occupant camera and respectively information about the sternum position (36) of the respective occupant (24) is taught. Computer program with program code means for carrying out all the steps of a method according to one of the preceding claims when the computer program is executed on a computer, a motor vehicle control unit or a corresponding computing unit. Device (10) for determining a sternum position (36) of an occupant (24) of a motor vehicle (42), with: - An input interface (12) for receiving sensor data with information about the occupant (24); - an analysis unit (14) for analyzing the sensor data and estimating a body feature of the occupant (24) based on the sensor data and determining a sternum position (36) based on the estimated body feature of the occupant (24) and an anatomical rule; and - An output interface (16) for transmitting the determined sternum position (36) to a vehicle unit (22). A system (18) for determining a sternum position (36) of an occupant (24) of a motor vehicle (42), comprising: - a device (10) according to the preceding claim, and at least one occupant camera, a sensor (20) for detecting a seat adjustment of the occupant (24), a sensor (20) for detecting a steering wheel adjustment of the occupant (24), a sensor (20 ) for detecting a rearview mirror setting of the occupant (24), a sensor (20) for detecting an outside mirror setting of the occupant (24), a sensor (20) for detecting a center armrest setting of the occupant (24), a sensor (20) for detecting a weight of the occupant (24) and/or a sensor (20) for detecting a seat belt extension.

Images