Method and device for adjusting a vehicle seat

Abstract

In a method for adjusting a vehicle seat (102), sensor data is detected by at least one pressure sensor (106) positioned in the seat surface or the backrest of the vehicle seat (102), an image of at least the upper body of an occupant (108) sitting in the vehicle seat (102) is captured, and image data corresponding to the image is generated. The pose of the occupant (108) is determined on the basis of the sensor data and the image data, and at least one actuator (114) of the vehicle seat (102) is adjusted on the basis of the determined pose.

Description

[0001] Method and device for adjusting a vehicle seat

[0002] The invention relates to a method for adjusting a vehicle seat. The invention further relates to a device for adjusting a vehicle seat.

[0003] Methods are known from the prior art that can determine a person's pose from images. These methods use machine learning and machine vision algorithms to determine the configuration and orientation of human limbs in space. It is also known to arrange pressure sensors in vehicle seats, for example, to determine whether an occupant is sitting in the vehicle seat or to determine the occupant's weight.

[0004] US 2010 / 080 416 A1 describes a system for determining the position of a driver's eyes and hands. The system combines images from a camera with the current seat position.

[0005] CN 103057503 A discloses an airbag control unit. A detection unit of the airbag control unit includes a seat position sensor that assists a camera in determining the pose of an occupant.

[0006] GB 2548198 A discloses a system for monitoring the seating position and posture of a vehicle occupant. The system comprises a camera and various sensors arranged in a vehicle seat.

[0007] Furthermore, KR 10 2023 064 297 A discloses a system for monitoring a driver's posture in real time. The system comprises a camera and sensors in the driver's seat. The object of the invention is to provide a method and a device for adjusting a vehicle seat that allow the vehicle seat to be adjusted based on a highly accurate pose of an occupant.

[0008] This object is achieved by a method having the features of claim 1 and by a device having the features of the independent device claim. Advantageous further developments are specified in the dependent claims.

[0009] In the proposed method for adjusting a vehicle seat, sensor data is acquired from at least one pressure sensor arranged in the seat surface or backrest of the vehicle seat. An image of at least the upper body of an occupant sitting in the vehicle seat is captured. Image data corresponding to the image is generated. Based on the sensor data and the image data, a pose of the occupant is determined. At least one actuator of the vehicle seat is controlled depending on the determined pose.

[0010] The proposed method can be used for adjusting vehicle seats of a motor vehicle, in particular a passenger car. In the method, the image recorded, for example, by an interior camera of the vehicle is combined with the sensor data recorded by the pressure sensor in order to determine the pose of the occupant. The pressure sensors record the weight force exerted by the occupant on the vehicle seat, i.e. the pressure on a surface of the vehicle seat. In this document, pose is understood to mean the position of key points of the human body in space. In addition, pose is also understood to mean variables derived from the position of key points, such as body height and body proportions.

[0011] Although the pose of the occupant can be determined from the image alone, it is difficult to determine the necessary depth information with high accuracy, particularly along the optical axis of a camera. This is especially true when using a mono camera, i.e. a camera that uses only one lens and is therefore not able to capture spatial depth in the same way as a stereo camera. With the proposed method, the pose determined from the image data can be verified and / or made more precise by taking the sensor data into account. Conversely, the pose determined on the basis of the sensor data can also be verified and / or made more precise by taking the image data into account. Through this verification and precision, for example, the height of the occupant can be determined with high accuracy.Based on the sensor data, it can also be determined whether the occupant is correctly seated in the vehicle seat or, for example, whether the occupant is not leaning against the backrest. Based on this information, it can then be determined whether the pose determined from the image is trustworthy. By combining the sensor data and the image data, a highly accurate pose of the occupant can be determined, which then serves as the basis for adjusting at least one actuator and thus the vehicle seat.

[0012] In one embodiment, at least one actuator of a seat adjustment of the vehicle seat is controlled depending on the determined pose. The seat adjustment allows the occupant to adjust the position and inclination of the vehicle seat. The at least one actuator is in particular a motor that moves the vehicle seat as a whole or parts thereof. For example, the vehicle seat can be moved closer to or further away from the steering wheel, the vehicle seat can be positioned higher or lower, or the inclination of the backrest can be adjusted. Adapting the vehicle seat to the anatomy and pose of the occupant not only increases comfort and ergonomics, but also safety. It is therefore particularly important to perform a seat adjustment based on the occupant's pose as accurately as possible.In a further embodiment, a seat contour of the vehicle seat is controlled depending on the determined pose using the at least one actuator. The seat contour can be achieved in particular by filling or emptying air cushions arranged in the seat surface and / or the backrest. In this embodiment, the at least one actuator is a pump or a valve via which the air flow into or out of one of the air cushions can be regulated. Adjusting the seat contour of the vehicle seat enables optimal distribution of the occupant's body weight, which contributes to reducing fatigue during long journeys and improving ergonomic support. As with seat adjustment, it is essential to adjust the seat contour based on the occupant's pose as accurately as possible.

[0013] The vehicle seat and / or seat contour can be adjusted, particularly during onboarding. An optimal setting determined during onboarding can be saved for each occupant so that this optimal setting can be reproduced during subsequent operation. However, adjustment can also be made subsequently during vehicle operation, particularly at the occupant's request.

[0014] In a further embodiment, the sensor data and / or the image are re-captured at a predetermined interval. The pose of the occupant is re-determined based on the newly acquired sensor data and / or the image data generated from the newly acquired image. The at least one actuator of the vehicle seat is controlled depending on the newly determined pose. The predetermined interval is in particular adapted to the frame rate of a camera, for example an interior camera, used to capture the image and is, for example, 1 / 10, 1 / 25, 1 / 30, or 1 / 60 of a second long, so that the pose of the occupant can be re-determined almost in real time. This allows the setting of the vehicle seat to be dynamically adapted to the current situation.

[0015] For example, the seat contour can be dynamically adjusted to the occupant's current posture. Since the occupant's posture is determined with high precision using the proposed method, feedback during the dynamic adjustment of the vehicle seat is also prevented, which could result from the accumulation of small measurement errors. For example, if the seat contour air cushions were controlled solely via the pressure sensors, this could lead to a situation in which the occupant leans forward only briefly, causing the air cushions in the occupant's back to fill, creating an unnatural and uncomfortable seat contour. This embodiment thus enables particularly reliable dynamic seat adjustment.

[0016] In a further embodiment, a pressure profile along the seat surface and / or the backrest of the vehicle seat is determined based on the sensor data. The occupant's pose is determined taking the pressure profile into account. The pressure profile can preferably be determined along one of the occupant's body axes, in particular the sagittal axis and / or the longitudinal axis. Based on the pressure profile, a highly precise correction of the pose determined from the image can be performed. For example, the pressure profile can be used to determine whether the occupant is leaning forward or whether their lower back is hitting the backrest.

[0017] In a further embodiment, seat adjustment data is determined that corresponds to the current seat adjustment of the vehicle seat. The pose of the occupant is determined taking into account the seat adjustment data. In this document, seat adjustment, also seat configuration, refers to the position of the vehicle seat in the vehicle interior and the relative positions of the individual elements of the vehicle seat to one another. For example, the seat adjustment includes the inclination of the backrest, the height of the headrest, or the setting of a seat length adjustment. The seat adjustment can define very precise reference points, particularly for an image captured with a camera permanently installed in the vehicle. The seat adjustment also determines the position of the pressure sensor in the vehicle interior. The seat adjustment thus provides the reference frame within which the pose of the occupant can be determined very precisely.

[0018] In another embodiment, the occupant's hip point is determined based at least on the sensor data. The hip point, also called the H-point, is part of the pose and an important reference point for optimizing seat adjustment, comfort, and the occupants' sight lines. For example, the location of the hip point can be determined using a standard vector of outer skin coordinates, which in turn can be easily determined from the sensor data and / or the image data.

[0019] In a further embodiment, it is determined at least based on the sensor data whether the occupant is currently leaning against the backrest of the vehicle seat. If the occupant is leaning against the backrest of the vehicle seat, the seat setting can be used particularly reliably as a reference frame, for example to determine the distance of points on the occupant's body to a camera used to capture the occupant's image. This makes it possible to determine the pose even more precisely based on the image data. Furthermore, the information as to whether the occupant is currently leaning against the backrest of the vehicle seat is important for adjusting the vehicle seat, because only then does it make sense to adjust, for example, the seat contour or the backrest.

[0020] In a further embodiment, the sensor data is processed using a machine learning method to determine the pose of the occupant. Alternatively or additionally, the image data is processed using a further machine learning method, in particular a computer vision method, to determine the pose of the occupant. The machine learning method used to process the sensor data is, for example, a neural network or a k-nearest neighbor method. This machine learning method has preferably been trained with a training data set comprising various sensor data, each labeled with specific information related to the pose of the occupant. For example, certain sensor data in the training data set can be labeled with the information that the occupant is currently leaning against the backrest.The additional machine learning method used to process the image data is, for example, a neural network, in particular a convolutional neural network, or a pose estimation algorithm. This network is preferably trained with a training dataset comprising a series of exemplary images of people sitting in a vehicle seat, each labeled with specific information related to the occupant's pose, for example, the spatial position of certain key points relative to a camera used to capture the respective image. Machine learning methods have the ability to learn from large amounts of data and recognize patterns. This makes machine learning methods ideally suited to determining the occupant's pose from the sensor data and / or the image data.

[0021] The invention further relates to a device for adjusting a vehicle seat. The device comprises at least one pressure sensor arranged in the seat surface or the backrest of the vehicle seat, and an image acquisition unit configured to capture an image of at least the upper body of an occupant sitting in the vehicle seat and to generate image data corresponding to the image. The device further comprises a control unit configured to determine a pose of the occupant based on the sensor data and the image data and to control at least one actuator of the vehicle seat depending on the determined pose.

[0022] The device has the same advantages as the claimed method. In particular, the device can be further developed with the features of the dependent claims directed to the method. Furthermore, the method described above can be further developed with the features described in this document in connection with the device.

[0023] In one embodiment, the at least one pressure sensor is configured to output a two-value signal as the sensor data. A first value of the two-value signal corresponds to a force below a predetermined value, and a second value of the two-value signal corresponds to a force above a predetermined value. In this embodiment, the at least one pressure sensor is, for example, a pressure mat arranged on the seating surface of the vehicle seat and configured to determine whether a person of a predetermined minimum weight is sitting in the vehicle seat. Such pressure sensors are already installed as standard in many vehicles and are particularly simple and cost-effective.

[0024] In a further embodiment, the device comprises at least two pressure sensors, each arranged either in the seat surface or the backrest of the vehicle seat. Providing multiple pressure sensors makes it possible, for example, to create a pressure profile in a particularly simple and cost-effective manner, which allows for a significantly more accurate determination of the occupant's weight distribution and thus the occupant's posture.

[0025] Embodiments of the invention are explained in more detail below with reference to the figures, in which:

[0026] Figure 1 is a schematic representation of a device for adjusting a vehicle seat according to an embodiment; and

[0027] Figure 2 shows a flowchart of a method for adjusting a vehicle seat according to an exemplary embodiment. Figure 1 shows a schematic representation of a device 100 for adjusting a vehicle seat 102 according to an exemplary embodiment. The vehicle seat 102 is designed purely by way of example as the driver's seat of a passenger car 104.

[0028] In the illustrated embodiment, the device 100 comprises five pressure sensors 106, two of which are arranged in the seat surface of the vehicle seat 102 and three in the backrest of the vehicle seat 102. The pressure sensors 106 each measure the force with which an occupant 108 sitting in the vehicle seat 102 presses against the surface of the vehicle seat 102, ie, in particular, the weight force of the occupant 108 on the vehicle seat 102. The pressure sensors 106 each generate sensor data depending on the measured force.

[0029] The sensor data can be a two-value signal. A first value can correspond to the situation in which the corresponding pressure sensor 106 is unloaded, i.e., the value of the force measured by the pressure sensor 106 is below a predetermined limit. A second value of the two-value signal can correspond to the situation in which the pressure sensor 106 is loaded, i.e., the value of the force measured by the pressure sensor 106 is above a predetermined limit. The predetermined limit is preferably such that the force exerted on the pressure sensor 106 by an average adolescent human exceeds the predetermined limit. In contrast, the force exerted by a small animal or an average child should be below the predetermined limit.The limit value can, for example, be set such that it is exceeded when the occupant weighs more than 21 kg, the weight of an average six-year-old, or more than 49 kg, the weight of a 5% percentile dummy. Alternatively, the sensor data can also be a multi-value signal whose value corresponds to the exerted force. In the exemplary embodiment shown, the pressure sensors 106 are further arranged such that a pressure profile can be created based on the sensor data. The pressure profile corresponds to the distribution of the force exerted by the occupant 108 on the driver's seat along a line that extends from the seat surface across the backrest to below the headrest.

[0030] The device 100 also includes an image capture unit 110, which is arranged and oriented such that the image capture unit 110 can capture an image of at least the upper body of the occupant 108 when the occupant is seated in the vehicle seat 102. From the captured image, the image capture unit 110 generates image data that can be further processed.

[0031] In the exemplary embodiment shown, the image capture unit 110 is a camera permanently installed in the vehicle interior, which is designed to capture the image as a two-dimensional image. In order to capture the image, the image capture unit 110 is designed, in particular, to capture light in the optical spectrum and / or in the infrared spectrum. The capture of light in the infrared spectrum has the additional advantage that the image capture unit 110 can capture the image even in poor lighting conditions, for example, at night. Alternatively, the image capture unit 110 can also be a time-of-flight camera, a RADAR system, or a LIDAR system in order to generate a three-dimensional image, in particular in the form of a point cloud.

[0032] The device 100 further comprises a control unit 112 configured to receive and process the sensor data from the pressure sensors 106 and the image data from the image acquisition unit 110. The control unit 112 is further configured to control the remaining elements of the device 100 to perform a method for adjusting a vehicle seat 102. In the method, the control unit 112 determines a pose of the occupant 108 based on the sensor data and the image data and controls an actuator 114 of the vehicle seat 102 depending on the determined pose. The method is described in more detail below with reference to Figure 2.

[0033] The controlled actuator 114 is shown only schematically in Figure 1. The actuator 114 can, for example, be a motor of a seat adjustment system, which can, for example, move the vehicle seat 102 within the vehicle interior, adjust the inclination of the backrest, or adjust the height of the headrest. The seat setting adjusted using the seat adjustment system can be provided to the control unit 112 in the form of seat setting data. The actuator 114 can also be a pump or a valve, with the aid of which the air flow into one or more air cushions arranged in the seat surface and / or the backrest can be controlled. By filling and emptying these air cushions, a seat contour of the vehicle seat 102 can be adapted to the occupant 108.

[0034] Figure 2 shows a flowchart of a method for adjusting a vehicle seat 102 according to one exemplary embodiment. The method can be carried out in particular by the device 100 described with reference to Figure 1 and is described below purely by way of example with reference to this device 100.

[0035] The method is started in step S200. In step S202, the pressure sensors 106 each detect the weight force exerted by the occupant 108 on the respective pressure sensor 106. The pressure sensors 106 generate sensor data corresponding to the respectively detected weight force and transmit this to the control unit 112. In step S204, the image acquisition unit 110 captures the image of the occupant 108 sitting in the vehicle seat 102. The image includes at least the upper body of the occupant 108, but can also include the head, arms, and other limbs. The image acquisition unit 110 generates image data corresponding to the image and transmits this to the control unit 112. In the optional step S206, the current seat setting of the vehicle seat 102 is detected, i.e.the position and orientation of the vehicle seat 102 in the vehicle interior and the position and orientation of the individual elements of the vehicle seat 102, such as the backrest and the headrest. Seat adjustment data corresponding to the seat adjustment are generated and transmitted to the control unit 112. The seat adjustment can be detected, in particular, by the vehicle seat 102 itself. Steps S202 to S206 can be performed simultaneously or in any desired order. Furthermore, two of the steps S202 to S206 can be performed simultaneously, while the third step is performed before or after.

[0036] In optional step S208, control unit 112 determines the pressure profile based on the sensor data. In optional step S210, control unit 112 determines whether occupant 108 is currently leaning against the backrest based on the sensor data. Additionally, control unit 112 can also determine whether occupant 108 is currently leaning against the backrest based on the image data. The two steps S208 and S210 can be performed simultaneously or in any order.

[0037] In step S212, the control unit 112 determines the pose of the occupant 108 based on the sensor data and the image data. If the pressure profile was determined in step S208, the control unit 112 determines the pose of the occupant 108 taking the pressure profile into account. If it was determined in step S210 whether the occupant 108 is currently leaning against the backrest, the control unit 112 determines the pose of the occupant 108 taking this information into account. To determine the pose, the control unit 112 determines, in particular, the position of key points of the body of the occupant 108 in the vehicle interior. Features on the body surface, such as the position of the eyes or mouth, can serve as key points. However, key points can also be joint points, such as the positions of the wrists or hip joints, the so-called hip point.Key points located within the body of the occupant 108 can be determined, in particular, using so-called standard vectors from key points on the body surface. To determine the position of the key points and / or the pose of the occupant 108, the control unit 112 can be configured to perform a machine learning method, with the aid of which the sensor data and / or the image data are processed.

[0038] In step S214, the control unit 112 controls the actuator 114 of the vehicle seat 102 depending on the determined pose in order to adjust the vehicle seat 102. For example, the control unit 112 can move the backrest and / or the headrest in order to adapt the seat setting of the vehicle seat 102 to the determined pose, in particular to a determined body size and / or to determined body proportions. The control unit 112 can further adapt the seat contour to the determined pose by controlling the pump and / or the valves with which the air cushions in the seat surface and / or the backrest can be filled and deflated. To dynamically adapt the vehicle seat 102 to the current pose of the occupant 108, steps S202 to S214 can be repeated at a predetermined interval. The method then ends in step S216.

[0039] In the exemplary embodiment described with reference to Figures 1 and 2, at least one of the pressure sensors 106, the image generation unit 110, and the control unit 112 form the device 100 for adjusting a vehicle seat. Further elements and features shown in Figures 1 and 2 and mentioned in the preceding description may be part of the device 100. Likewise, method steps described with reference to the device 100 may be part of the claimed method. List of Reference Symbols

[0040] 100 device

[0041] 102 Vehicle seat

[0042] 104 passenger cars

[0043] 106 Pressure sensor

[0044] 108 inmates

[0045] 110 Image acquisition unit

[0046] 112 Control unit

[0047] 114 Actuator

Claims

Claims 1. A method for adjusting a vehicle seat (102), in which Sensor data from at least one pressure sensor (106) arranged in the seat surface or the backrest of the vehicle seat (102) is acquired; an image of at least the upper body of an occupant (108) sitting in the vehicle seat (102) is acquired, and image data corresponding to the image is generated; a pose of the occupant (108) is determined on the basis of the sensor data and the image data; and at least one actuator (114) of the vehicle seat (102) is controlled depending on the determined pose.

2. The method according to claim 1, wherein at least one actuator (114) of a seat adjustment of the vehicle seat (102) is controlled as a function of the determined pose.

3. The method according to claim 1 or 2, wherein using the at least one actuator (114) a seat contour of the vehicle seat (102) is controlled as a function of the determined pose.

4. Method according to one of the preceding claims, wherein the sensor data and / or the image are newly acquired at a predetermined interval, the pose of the occupant (108) is newly determined on the basis of the newly acquired sensor data and / or the image data generated from the newly acquired image and the at least one actuator (114) of the vehicle seat (102) is controlled as a function of the newly determined pose.

5. Method according to one of the preceding claims, wherein a pressure profile along the seat surface and / or the backrest of the vehicle seat (102) is determined on the basis of the sensor data and the pose of the occupant (108) is determined taking into account the pressure profile.

6. Method according to one of the preceding claims, wherein seat adjustment data are determined which correspond to the current seat adjustment of the vehicle seat (102), and the pose of the occupant (108) is determined taking into account the seat adjustment data.

7. Method according to one of the preceding claims, wherein the hip point of the occupant (108) is determined at least on the basis of the sensor data.

8. Method according to one of the preceding claims, wherein it is determined at least on the basis of the sensor data whether the occupant (108) is currently leaning against the backrest of the vehicle seat (102).

9. The method according to any one of the preceding claims, wherein the sensor data is processed using a machine learning method to determine the pose of the occupant (108), and / or wherein the image data is processed using a further machine learning method, in particular a computer vision method, to determine the pose of the occupant (108).

10. Device (100) for adjusting a vehicle seat (102), comprising at least one pressure sensor (106) arranged in the seat surface or the backrest of the vehicle seat (102); an image acquisition unit (110) designed to capture an image of at least the upper body of an occupant (108) sitting in the vehicle seat (102) and to generate image data corresponding to the image; a control unit (112) designed to determine a pose of the occupant (108) on the basis of the sensor data and the image data and to control at least one actuator (114) of the vehicle seat (102) depending on the determined pose.

11. The device (100) of claim 10, wherein the at least one pressure sensor (106) is configured to output a two-value signal as the sensor data, and wherein a first value of the two-value signal corresponds to a force below a predetermined value and a second value of the two-value signal corresponds to a force above a predetermined value.

12. Device (100) according to claim 10 or 11, wherein the device (100) comprises at least two pressure sensors (106), each arranged either in the seat surface or the backrest of the vehicle seat (102).

Images