WO2024046623A1

WO2024046623A1 - Transport means, assembly and method for categorising a passenger of a transport means

Info

Publication number: WO2024046623A1
Application number: PCT/EP2023/068088
Authority: WO
Inventors: Thomas Ewender; Johannes Feldmaier; Simone CAMBERG; Florian Ober
Original assignee: Bayerische Motoren Werke Aktiengesellschaft
Priority date: 2022-08-29
Filing date: 2023-06-30
Publication date: 2024-03-07
Also published as: DE102022121814A1

Abstract

The invention relates to an assembly, a transport means (10) and a method for categorising a passenger (1, 2) of a transport means (10). The method comprises the steps: determining an electrical signal of a motor (3, 4) for adjusting a seat (5, 6) on which the passenger (1, 2) sits; detecting the passenger (1, 2) by means of an optical sensor (7) in order to generate an image; comparing the electrical signal and the image with a respective reference; and, depending on a result of the comparison, categorising the passenger (1, 2) as a "child on a child seat (9)" or as an "adult".

Description

Means of transport, arrangement and method for categorizing an occupant of a means of transport

Description

The present invention relates to an arrangement, a means of transport and a method for categorizing an occupant of a means of transport. In particular, the present invention relates to an implementation of an automatic categorization of an occupant that is not very complex in terms of hardware, for example to switch off an airbag.

Under certain circumstances, airbags in vehicles must be switched off or automatically deactivated so as not to become a safety risk in the event of an impact. For example, airbags may not be switched on if a child seat oriented against the direction of travel is mounted and occupied on the seat in question. In some countries, automatic airbag deactivation is a legal requirement for certain usage scenarios. In the prior art, to detect an occupied seat, a sensor mat is used in the seats to determine the occupancy status and the mass/occupant category. While the airbag deployment is prevented when the seat is empty, the airbag is deployed if necessary when the seat is occupied by an adult. If children (possibly with a booster seat) are seated in the seat in question, airbag deployment is usually also prevented. In the case of an unoccupied seat, a rear-facing child seat and/or an occupant who is three to six years old, an airbag located on the seat will not be deployed, otherwise it will normally be deployed.

The sensor mats in the seat are hardware that increases vehicle mass, seating complexity and the cost of the means of transportation. There are also cable harnesses to forward the sensor mat signals to design. If there is no sensor mat concept, which is also referred to as a “Passive Occupant Detection System (PODS)”, the problem with purely camera-based occupant detection is that the purely image-based distinction between a six-year-old child on a booster seat and one The so-called 5% woman is difficult because the height estimate can be incorrect when using the booster seat. In addition, the booster seat may be optically obscured and therefore cannot be captured by the camera (obscuration by clothing, e.g. dress, skirt, blanket or other accessories), so that a correction based on optical signals is not possible.

Based on the aforementioned prior art, it is an object of the present invention to implement correct occupant classification and optional airbag deactivation with the simplest possible hardware, in particular hardware-neutral.

The above-mentioned object is achieved according to the invention by a method having the features according to claim 1. Accordingly, a categorization of an occupant (e.g. driver, front passenger, rear passenger, etc.) of a means of transport (e.g. a car, van, motorcycle, truck, aircraft and/or watercraft) is proposed. The categorization can classify the occupant in particular in such a way that in the event of an impact, a decision is made based on the occupant category as to whether an airbag will be deployed or not. In a first step, an electrical signal from a motor for adjusting a seat on which the occupant sits is determined. The electrical signal can include, for example, a current value, a power consumption, a thermal parameter, a torque, a force (both of the latter, for example through a sensor on a shaft and/or on an electric motor) or similar. The motor can be used in particular to generate the electrical signal. For example, the current value can be the current consumption when adjusting the seat at a point in time at which the occupant is sitting on the seat with his full weight or reduced by the weight of his legs placed in the footwell. For this purpose, a slight, in particular short, seat adjustment can be carried out in the manner of a comfort access, through which the seat is only adjusted for the occupant from a rear position to a position further forward after getting in to make boarding easier and then to optimize the seating position for the upcoming journey. In a further step, the occupant is optically detected using an optical sensor (eg an interior camera of the means of transport) in order to save an image of the occupant (if necessary as a moving image or video). In particular, a digital representation of the image can be created here. In a further step, the electrical signal from the motor and the image are compared with a respective reference. In particular, the reference may have been trained or learned in steps further back. In any case, the reference is suitable for estimating (determining) a mass of the occupant based on the electrical signal of the engine and for classifying the occupant's posture at least as to whether it is suitable for determining the mass based on the electrical signal of the engine. In other words, the reference can provide information about whether the photograph of the occupant suggests that the occupant has assumed a normal sitting position, so that it can be ruled out that the occupant's limbs are being used to transfer his mass to other areas of the Vehicle interior than in the seat and footwell. In this way, it can be ensured that the current occupant on the seat has a mass that can be represented sufficiently precisely by the sensor mat of the seat and/or the electrical signal from the motor. Depending on the result of the comparison, the occupant is then categorized or classified. In other words, it is determined whether it is a child in a child seat and/or a booster seat or an adult. This categorization/classification can also be done automatically. In particular, a signal can be output through which automatic further measures can be initiated depending on the categorization. In this way, a sensor mat for the seat can be omitted, meaning that the means of transport can be designed to be lighter, more cost-effective and less prone to errors.

The subclaims show preferred developments of the invention.

The motor for adjusting the seat can in particular be an electric motor, most preferably a brushless DC motor. Depending on their power consumption, these motors can be used to: Torque and to determine the mass of the occupant via the torque. In particular, a brushless DC motor can also output or reveal other electrical parameters/operating variables in order to determine the mass of the occupant.

A machine-based learning algorithm (ML) can preferably be used to make the method according to the invention more robust and efficient. In a further step, a large number of test subjects can therefore be presented to an optical sensor, which are optically recorded, in particular in a corresponding structure, as will later be present in the vehicle. For example, the optical sensor can be arranged at a position such as the interior camera used in the vehicle to optically detect the occupant. Accordingly, the seating can also be designed accordingly during the training (learning process). The perspective and distance can also be essentially identical to the series vehicle. The mass of the test subjects can be determined in a conventional method using a scale and communicated to the machine-based learning algorithm for the respective test subject. In this way, the machine-based learning algorithm, which, for example, carries out a weight estimate from the electrical signal of the engine and/or the images of the occupants, can determine under which circumstances the estimated mass corresponds sufficiently to the actual mass. The constellations of electrical signal and position of the occupant in the image can then be used to categorize the training data set as “permissible” if the mass of the subject determined by the electric motor and the optical signal corresponds sufficiently to the actual mass. In other words, the training data set can be assessed as permissible if a mass of the subject determined by the electric motor corresponds sufficiently to the training data set. Thus, without the data of a sensor mat, a sufficiently valid mass estimate of an occupant can be made, through which the occupant can be categorized as “child” or “adult”.

In a further step, the training data set that is considered permissible can be used as a predefined reference for a valid categorization based on the electrical signal. In other words, the learned can Training data set represents the predefined reference or can be used as part of the predefined reference, which has already been introduced above.

When comparing the electrical signal with the predefined reference, body posture in particular can be determined using the optical sensor. In other words, it can be determined to what extent the posture has an influence on the mass, which is determined using the electrical signal from the motor. If the posture is a normal posture in which only part of the occupant's weight is registered in the footwell, while the main component can be registered in a seat surface (and backrest) of the seat, an electrical signal from the motor can be considered "representative". the mass of the occupant and used for a valid determination of the mass. However, the reference can also indicate a posture in which the occupant, for example, kneels on the seat, supports himself significantly on the center console of the means of transport or elsewhere, so that the electrical signal that was determined by means of the motor does not reflect the actual weight of the occupant represented. In this case, the determination of the electrical signal can be repeated at a later point in time and the electrical signal now determined can be used to determine the mass of the occupant, provided that the image of the occupant does not now represent an abnormal posture. In this way, the mass of an occupant can be successfully determined using the electrical signal from an electric motor, even without a pressure sensor/sensor mat.

As a final step, the method according to the invention can be completed by switching off an airbag for the seat if the seat is occupied and the category is “child in child seat”. In this case, an airbag will not be deployed if the vehicle is involved in an accident of sufficient severity. Otherwise (the category is “Adult in Seat”) the airbag will activate and deploy as intended in the event of a collision to protect the occupant.

According to a second aspect of the present invention, an arrangement for categorizing an occupant of a means of transport is proposed. The arrangement includes a data input, an evaluation unit and a data output. The aforementioned elements of the arrangement can be designed as an electronic control device. The means of transport can be designed as a car, van, motorcycle, truck, aircraft and/or watercraft. The evaluation unit is set up to use the data input to determine an electrical signal from a motor for adjusting a seat on which the occupant is sitting. In addition, the evaluation unit is set up to detect the occupant by means of the data input (for example in conjunction with an optical sensor) to generate an image or to determine an image of the occupant. The electrical signal and the image can then be compared with a respective reference and, depending on the result of the comparison, the evaluation unit categorizes the occupant as a “child in a child seat” or as an “adult” using the data output. Depending on the category, the evaluation unit can now automatically deactivate or activate an airbag of the means of transport via the data output. The arrangement according to the invention is therefore able to realize the features, combinations of features and the advantages resulting from them in such a way that reference is made to the above statements in order to avoid repetition.

According to a third aspect of the present invention, a means of transportation is proposed which has an arrangement according to the second-mentioned aspect of the invention. In this way, the means of transport can also implement the features, combinations of features and advantages in a corresponding manner.

Further details, features and advantages of the invention result from the following description and the figures. Show it:

Figure 1 shows a schematic representation of an exemplary embodiment of a means of transportation according to the invention with an exemplary embodiment of an arrangement according to the invention when carrying out a method according to the invention; and Figure 2 is a flowchart illustrating steps of an exemplary embodiment of a method according to the invention for categorizing an occupant of a means of transport.

Figure 1 shows a car as a means of transport 10, in the first row of which a passenger has taken a seat as an occupant 1 on a passenger seat and a child has taken a seat as an occupant 2 on a back seat 6. The child 2 sits on a booster or booster seat 9. The seating is adjustable using electric motors 3, 4. Depending on the masses or occupants 1, 2 located on the seats, the currents drawn by the electric motors 3, 4 are of different levels and can therefore be used to determine the mass of the occupants 1, 2. An interior camera 7 is aligned so that both occupants 1, 2 are optically captured in their entirety. Here, the posture of the occupants 1, 2 and the presence of the booster seat 9 can be determined using the data from the interior camera 7. This information or images/images of the occupants 1, 2, as well as the input variables of the electric motors 3, 4, are fed via a data input 11 to an evaluation unit 12 in the form of an electronic control device. An airbag 8 is switched off or reactivated via a data output 13 of the control unit and, if necessary, a message is output on a central information screen 14. The message can, for example, provide information about the automatic activation/deactivation of the airbag 8.

Figure 2 shows steps of an exemplary embodiment of a method according to the invention for categorizing an occupant of a means of transport. In step 100, a machine-based learning algorithm is trained by presenting different subjects to an optical sensor and linking them to their respective actual masses to produce a training data set. In other words, the machine-based learning algorithm is presented with a first test subject on a seat similar to that of the vehicle and the test subject's mass, determined (for example using a scale), is made known. In step 200, a mass of the test subjects is determined using an electric motor in the seat in order to supplement the training data set. Kinematics and motor as well as the spatial relationships between the sensors and the seat/occupant can correspond to those of the series vehicle. In step 300 the Training data set is assessed as permissible if a mass of the subject determined by the electric motor corresponds sufficiently to the training data set. Feedback is provided, so to speak, as to whether the image of the subject and the mass determined using the electrical signal “match” the actual mass of the subject. If this is the case, the training data set is used to define the predefined reference and the series vehicle is equipped with the predefined reference. In step 400, an electrical signal from a motor for adjusting a seat on which the occupant of the means of transport sits is then determined. The electrical signal can be, for example, a current input from the motor. In step 500, the occupant is also detected using an optical sensor in the form of an interior camera of the means of transport to generate an image or an image of the occupant. In step 600, the electrical signal is then compared with a first reference and the image of the occupant with a second reference and, depending on the result of the respective comparison, the occupant is categorized as a “child” or as an “adult”. In particular, based on the child's seat size, it can be concluded that the child is in a child seat or on a booster seat and therefore the airbag of the means of transport must be deactivated automatically. In step 700, the occupant is categorized as a “child in child seat.” Finally, in step 800, an airbag for the seat is deactivated if the seat is occupied and the category is “child in child seat”.

According to a core idea of the present invention, a method is carried out, which can preferably be designed as a machine-based learning method. This can be divided into two basic phases:

- Recording and labeling of training data: Creation of camera recordings with test subjects of a known weight, determination of a basic information base ("ground truth") about measured weight: If the measured weight via electric motors is within a predefined range of the actual weight (e.g. ± 5 kg ), then the data set is a positive class, otherwise it is a negative class. - Implementation: Determination of the class as the output of the machine learning-based method, e.g. the neural network. As soon as the person has been classified into a positive class, the weight measurement is carried out using an electric motor. An even simpler method could be kept as a base/fallback option. This would involve using training data that uses classic misalignments (out-of-position) of sitting positions and permissible (in-position) sitting positions.

This means that there is no need for a sensor mat for seat occupancy detection or occupant classification, which means that vehicles can be designed to be cheaper, lighter and less prone to errors.

List of reference symbols:

1, 2 occupants

3, 4 electric motor

5, 6 seat

7 interior camera

8 airbags

9 booster seat

10 means of transportation

11 Data input

12 evaluation unit

13 data output

14 central information display

100-800 process steps

Claims

Patent claims:

1. Method for categorizing an occupant (1, 2) of a means of transport (10) comprising the steps

• Determining (400) an electrical signal from a motor (3, 4) for adjusting a seat (5, 6) on which the occupant (1, 2) sits,

• Detecting (500) the occupant (1, 2) using an optical sensor (7) to generate an image,

• Comparing (600) the electrical signal and the image with a respective reference and depending on a result of the comparison

• Categorize (700) the occupant (1, 2) as a “child in child seat (9)” or as an “adult”.

2. The method according to claim 1, wherein the motor (3, 4) is a brushless DC motor.

3. The method according to claim 1 or 2, wherein the electrical signal represents a current or a torque.

4. The method according to any one of the preceding claims further comprising

- Training (100) of a machine-based learning algorithm by presenting subjects to an optical sensor and linking them to their actual mass to produce a training data set.

5. The method according to claim 4 further comprising

- Determining (200) a mass of the test subjects using an electric motor of a seat to supplement the training data set and

- Evaluate (300) the training data set as permissible if a mass of the subject determined by the electric motor corresponds sufficiently to the training data set.

6. The method according to claim 5 further comprising

- Using the valid training data set as a predefined reference for valid categorization based on the electrical signal. Method according to one of the preceding claims, wherein when comparing the electrical signal with the reference, a body posture is determined by means of the optical sensor (7), which is predefined for a valid determination of the mass of the occupant (1, 2). Method according to one of the preceding claims, wherein no signal from a pressure sensor is included in the comparison. Method according to one of the preceding claims further comprising the steps

- Switching off (800) an airbag (8) for the seat (5, 6) if the seat (5, 6) is occupied and the category is “Child in child seat (9)” and

- Switching on an airbag (8) for the seat (5, 6) in all other cases. Arrangement for categorizing an occupant (1, 2) of a means of transport (10).

• a data input (11),

• an evaluation unit (12) and

• a data output (13), where

• the evaluation unit (12) is set up to use the data input (11) to determine an electrical signal from a motor (3, 4) for adjusting a seat (5, 6) on which the occupant (5, 6) sits and the To detect occupants (1, 2) using an optical sensor (7) to generate an image, to compare the electrical signal and the image with a respective reference and depending on a result of the comparison

• Using the data output (13) to categorize the occupant as a “child in a child seat (9)” or as an “adult”. Means of transport (10) comprising an arrangement according to claim 10.