Method for adapting a triggering algorithm for a personal restraint device and control device for adapting a triggering algorithm for a personal restraint device

Abstract

The invention discloses a method (100) for adapting a triggering algorithm of a personal restraint device of a vehicle (1) based on a detected vehicle interior state of the vehicle (1). The method (100) comprises: - detecting (110, 120) key points (7) of a vehicle occupant (6) by means of an optical sensor device (3, 4); - determining (112, 122) a vehicle occupant posture of the vehicle occupant (6) based on a connection between the detected key points (7) and a skeleton-like representation of body parts of the vehicle occupant (6), wherein the skeleton-like representation reflects the relative position and orientation of the individual body parts of the vehicle occupant (6); - predicting (114, 124) a future vehicle occupant posture of the vehicle occupant (6) based on a predicted future position of at least one key point (7); and - modifying (116, 126) a triggering algorithm of a personal restraint device based on the predicted future posture of the vehicle occupant (6). The invention further discloses a control device (2) for adapting a triggering algorithm of a personal restraint device of a vehicle (1) based on a detected vehicle interior state of the vehicle (1), wherein the control device (2) is configured or designed to perform a method (100) according to the invention.

Description

Technical Field

[0001] The invention relates to a method for adapting a triggering algorithm of a personal restraint device of a vehicle based on detected vehicle interior states of the vehicle. The invention also relates to a control device for adapting a triggering algorithm of a personal restraint device of a vehicle based on detected vehicle interior states of the vehicle. Background Art

[0002] For many years, it has been known to provide personal protection devices in vehicles, such as impact protection systems designed as personal restraints, which are intended to protect vehicle occupants in the event of a collision or impact, with the goal of preventing injuries to the occupants as far as possible, or at least reducing their severity. Typically, an airbag is used as a personal restraint, which catches the occupant in the event of an impact. The airbag is deployed and expanded between the occupant and the vehicle interior components within a short time range of 10ms to 50ms by the inflow of fluid, forming a cushion, thereby preventing the occupant from colliding with hard components of the vehicle interior, such as the steering wheel.

[0003] Furthermore, currently, vehicle occupant recognition in a vehicle is usually performed using sensors installed in the vehicle seats. These sensors are designed for occupant recognition by means of mass estimation. Furthermore, the so-called seat belt buckle is used to detect whether the vehicle occupant is wearing a seat belt while driving.

[0004] Known restraints are usually designed to provide the vehicle occupant with the greatest possible protection only when the vehicle occupant is in a predetermined position. If the vehicle occupant is no longer in this predetermined position, the protective effect of the restraint may be reduced. During a collision, the vehicle occupant will no longer be optimally protected.

[0005] Furthermore, due to semi-autonomous or autonomous driving, more and more passengers will interact with objects in the vehicle (such as laptops, mobile phones, tablets and / or musical instruments) during their journey and hold these or other mobile objects in their hands.

[0006] For example, if the driver's hands are on the steering wheel blocking the steering air bag, or their hands are placed on top of the air bag area, the driver may be injured by their hands or moving objects held in their hands if the air bag deploys.

[0007] In order to maximize occupant protection and reduce the risk of injury during a collision, it is desirable to change the deployment characteristics of an airbag or other operating characteristics of a personal restraint component based on the detected vehicle interior state of the vehicle. Specifically, it is desirable to control factors such as the inflation profile and deployment time of an airbag based on the position and / or posture of an occupant in a seat. Summary of the invention

[0008] In view of the above, an object of the present invention is to provide a method and a control device for adapting a triggering algorithm of a personal restraint device of a vehicle based on a detected vehicle interior state of the vehicle, the method and the control device ensuring accurate determination of the interior state so as to be able to well protect the occupants when the restraint device is triggered.

[0009] According to one aspect of the invention, there is provided a method for adapting a triggering algorithm of a personal restraint device of a vehicle based on a detected vehicle interior state of the vehicle. The method comprises:

[0010] - Detection of key points of vehicle occupants by means of optical sensor devices;

[0011] - determining a vehicle occupant pose of the vehicle occupant based on a connection of the detected key points with a skeleton-like representation of body parts of the vehicle occupant, wherein the skeleton-like representation reflects the relative position and orientation of the various body parts of the vehicle occupant;

[0012] - predicting a future vehicle occupant posture of a vehicle occupant based on the predicted future position of at least one key point; and

[0013] - Modifying the triggering algorithm of the personal restraint device based on the predicted future posture of the vehicle occupant.

[0014] The invention is based on the consideration that a very precise determination of the vehicle interior state can be achieved by detecting key points of a vehicle occupant by means of an optical sensor device and determining the posture of the vehicle occupant based on a connection of the detected key points with a skeleton-like representation of the body parts of the vehicle occupant, wherein the skeleton-like representation reflects the relative position and orientation of the individual body parts of the vehicle occupant. The invention is further based on the consideration that a maximum protection of the vehicle occupant in the event of a collision or impact can be achieved by estimating the future posture of the vehicle occupant based on the predicted future position of at least one key point within at least the next few microseconds, in particular shortly before or at the time of a collision or impact, and by modifying the triggering algorithm of the personal restraint device based on this predicted future posture of the vehicle occupant.

[0015] According to the invention, an accurate determination of the interior state is ensured so that the vehicle occupants can be protected to the greatest extent possible in the event that the restraint device is triggered.

[0016] According to an advantageous embodiment, the method further comprises:

[0017] - detecting a moving object held by a vehicle occupant by means of an optical sensor device;

[0018] - predict the future state of moving objects; and

[0019] - Modifying a triggering algorithm for a personal restraint device based on a predicted future posture of a vehicle occupant and a predicted future state of moving objects.

[0020] In this way, moving objects held by vehicle occupants are also taken into account, so that a more accurate determination of the vehicle interior state can be achieved, thereby further maximizing the protection of the vehicle occupants in the event of a collision or impact.

[0021] According to another specific embodiment, the predicted future state of the moving object comprises at least one of the future position of the moving object in the interior of the vehicle, the future speed of the moving object or the future orientation of the moving object. Preferably, the position of the moving object in the vehicle is described as the position of the moving object relative to the personal restraint device and / or the steering wheel. The predicted future state of the moving object preferably comprises the future position of the moving object in the interior of the vehicle, the future speed of the moving object and the future orientation of the moving object. Based on these parameters, a very accurate state of the moving object can be described and used to modify the triggering algorithm, which helps to further enhance safety and protection.

[0022] According to another specific embodiment, the method further comprises:

[0023] - correlating the position and / or motion of the moving object with the position and / or motion of a key point representing a wrist of a vehicle occupant;

[0024] - Determining from the association which hand of the vehicle occupant is holding the moving object.

[0025] This determination of the hand holding the object is well suited to more accurately describe the vehicle interior state and thus modify the triggering algorithm more appropriately.

[0026] According to another specific embodiment, the predicted future position of at least one key point is estimated based on at least one of vehicle occupant size, vehicle occupant seat position, vehicle occupant seat back angle, seat belt status, vehicle speed or vehicle acceleration. Preferably, the vehicle occupant size is estimated based on the size of a body part (particularly a limb) of the vehicle occupant.

[0027] According to another specific embodiment, the predicted future state of the moving object is estimated based on at least one of the vehicle occupant size, the vehicle occupant seat position, the vehicle occupant seat back angle, the seat belt state, the vehicle speed or the vehicle acceleration. Preferably, the vehicle occupant size is estimated based on the size of the vehicle occupant's body part, especially the limbs.

[0028] According to another specific embodiment, the predicted future position of at least one key point is estimated based on a calculated future key point velocity vector, wherein the future key point velocity vector is obtained by the vector sum of the product of the estimated future key point velocity vector and the current vehicle velocity vector multiplied by a first scalar parameter derived from the signal of the impact sensor. This will enable dynamic key point tracking and a particularly accurate prediction of future key point positions and thus of future vehicle occupant postures. The impact sensor may preferably be an acceleration sensor, an angular rate sensor or an impact or contact sensor (such as a pressure sensor) or a combination of one or more of these sensors. In particular, the first scalar parameter may also be determined by taking into account vehicle data such as vehicle size and vehicle weight.

[0029] In a preferred further embodiment, the future key point velocity vector is calculated by the following formula:

[0030]

[0031] in is the future key point velocity vector, is the estimated future keypoint velocity vector, (α) is the first scalar parameter derived from the signal of the impact sensor, is the current vehicle velocity vector, (β) is a second scalar parameter derived from the signal of the impact sensor, and is the current collision target velocity vector.

[0032] In this way, the determination of the future keypoint velocity vector further takes into account the current collision target velocity vector multiplied by the second scalar value parameter derived from the signal of the impact sensor. In particular, in the event of a collision, the future vehicle occupant posture can be further accurately predicted in this way.

[0033] Wherein, the collision target is the object with which the vehicle collides, wherein the collision target may be a stationary object or a moving object around the vehicle. Therefore, the collision target may be, for example, another vehicle, a bicycle, a motorcycle, a pedestrian, a crash barrier, a street lamp or a tree. For a stationary object, the current collision target velocity vector is usually zero. The collision target velocity vector may be derived in particular from a signal of an ambient sensor of the vehicle (such as a radar sensor, a lidar sensor, an ultrasonic sensor or an ambient camera). The second scalar value parameter may be derived, for example, from another signal of the same collision sensor from which the first scalar parameter is derived, or from the same signal of the same collision sensor from which the first scalar parameter is derived. However, the second scalar value parameter may also be derived, for example, from another collision sensor, wherein the other collision sensor may preferably be an acceleration sensor, an angular rate sensor or an impact or contact sensor (such as a pressure sensor) or a combination of one or more of these sensors. In particular, the second scalar parameter may also be determined by taking into account vehicle data (such as vehicle size and vehicle weight).

[0034] According to another specific embodiment, the predicted future position of the mobile object is estimated based on the calculated future mobile object velocity vector, wherein the future mobile object velocity vector is obtained by the vector sum of the product of the estimated mobile object velocity vector and the current vehicle velocity vector multiplied by a first scalar parameter derived from the signal of the impact sensor. This will enable dynamic mobile object tracking and a particularly accurate prediction of the future mobile object position. The impact sensor can preferably be an acceleration sensor, an angular rate sensor or an impact or contact sensor (such as a pressure sensor) or a combination of one or more of these sensors. In particular, the first scalar parameter can also be determined by taking into account vehicle data (such as vehicle size and vehicle weight).

[0035] In a further preferred embodiment, the future moving object velocity vector is calculated by the following formula:

[0036]

[0037] in is the future moving object velocity vector, is the estimated future moving object velocity vector, (α) is the first scalar parameter derived from the signal of the impact sensor, is the current vehicle velocity vector, (β) is a second scalar parameter derived from the signal of the impact sensor, and is the current collision target velocity vector.

[0038] In this way, the determination of the future moving object velocity vector in particular further takes into account the current collision target velocity vector multiplied by the second scalar value parameter derived from the signal of the impact sensor. In particular in the case of a collision, the future moving object position can be predicted more accurately in this way.

[0039] Wherein, the collision target is the object with which the vehicle collides, wherein the collision target may be a stationary object or a moving object around the vehicle. Therefore, the collision target may be, for example, another vehicle, a bicycle, a motorcycle, a pedestrian, a crash barrier, a street lamp or a tree. For a stationary object, the current collision target velocity vector is usually zero. The collision target velocity vector may be derived in particular from a signal of an ambient sensor of the vehicle (such as a radar sensor, a lidar sensor, an ultrasonic sensor or an ambient camera). The second scalar value parameter may be derived, for example, from another signal of the same collision sensor from which the first scalar parameter is derived, or from the same signal of the same collision sensor from which the first scalar parameter is derived. However, the second scalar value parameter may also be derived, for example, from another collision sensor, wherein the other collision sensor may preferably be an acceleration sensor, an angular rate sensor or an impact or contact sensor (such as a pressure sensor) or a combination of one or more of these sensors. In particular, the second scalar parameter may also be determined by taking into account vehicle data (such as vehicle size and vehicle weight).

[0040] According to another specific embodiment, the key points of the vehicle occupants are detected by using at least an IR camera and a 3D camera. The IR camera uses an infrared emitter in order to reliably identify and detect the vehicle interior state also at night. The 3D camera preferably operates according to the time-of-flight method and allows the vehicle interior state to be detected based on actual motion vectors in three-dimensional space. Preferably, the 3D camera and the IR camera are interior cameras located in the vehicle, wherein the 3D camera and the IR camera are in particular integrated in the roof lining between the front seats of the vehicle and / or are located in the area close to the rearview mirror. In particular, the 3D camera and the IR camera can form a single-structure camera unit.

[0041] In a preferred further embodiment, the 2D key points of the vehicle occupant detected by the IR camera are converted into 3D key points of the vehicle occupant by fusing information provided by the 3D camera.Preferably, the vehicle occupant pose of the vehicle occupant is determined based on the 3D key points of the vehicle occupant.

[0042] In a preferred further embodiment, only the IR camera is used to detect moving objects held by a vehicle occupant.

[0043] According to another specific embodiment, the key points of the vehicle occupant are the vehicle occupant skeletal joints. Such key points allow a very precise determination of the vehicle occupant posture, since they have a high degree of relevance and a large amount of information about the position, orientation and movement of the various body parts.

[0044] According to another aspect of the invention, a control device for adapting a triggering algorithm of a personal restraint device of a vehicle based on a detected vehicle interior state of the vehicle, wherein the control device is configured to:

[0045] - Detection of key points of vehicle occupants by means of optical sensor devices;

[0046] - determining a vehicle occupant pose of the vehicle occupant based on a connection of the detected key points with a skeleton-like representation of body parts of the vehicle occupant, wherein the skeleton-like representation reflects the relative position and orientation of the various body parts of the vehicle occupant;

[0047] - predicting a future vehicle occupant posture of a vehicle occupant based on the predicted future position of at least one key point; and

[0048] - Modifying the triggering algorithm of the personal restraint device based on the predicted future posture of the vehicle occupant.

[0049] The advantages and specific preferred embodiments described for the method according to the invention also apply correspondingly to the control device according to the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments and descriptions of the present invention are used to explain the present invention and do not constitute improper limitations on the present invention. In the drawings:

[0051] Figure 1 A schematic diagram shows a vehicle comprising a control device for adapting a triggering algorithm for a personal restraint device based on detected vehicle interior states according to a preferred embodiment of the invention;

[0052] Figure 2 Shown by Figure 1 A flowchart of a method for adapting a triggering algorithm for a personal restraint device according to a preferred embodiment of the present invention executed by a control device shown;

[0053] Figure 3 Shown is the Figure 2 A method for adapting a control device of a triggering algorithm of a personal restraint device; and

[0054] Figure 4 A flow chart of a method for adapting a triggering algorithm of a personal restraint device according to another preferred embodiment of the present invention is shown. DETAILED DESCRIPTION

[0055] The present invention is described in detail below with reference to the accompanying drawings and in combination with the embodiments. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict. In all the drawings, the corresponding parts / components always have the same reference numerals.

[0056] Figure 1 A schematic diagram shows a vehicle 1 comprising a control device 2 for adapting a triggering algorithm of a personal restraint device (not shown) of the vehicle 1 based on detected vehicle interior states. The vehicle 1 comprises a 3D camera 3 and an IR camera 4 as optical sensor devices, wherein the 3D camera 3 and the IR camera 4 are integrated in the roof lining between the vehicle front seats 3 and in the area close to the rearview mirrors of the vehicle 1, and wherein these cameras 3, 4 monitor the interior 5 of the vehicle 1. The vehicle 1 further comprises a surroundings sensor (not shown), such as a radar sensor, a lidar sensor, an ultrasonic sensor or a surroundings camera or a combination thereof, and at least one crash sensor (not shown).

[0057] The control device 2 is configured to detect the skeletal joints of the vehicle occupant 6 as the key points 7 of the vehicle occupant 6 through the cameras 3 and 4. Thus, the 2D key points of the vehicle occupant 6 detected by the IR camera 4 are converted into 3D key points 7 of the vehicle occupant 6 by fusing the information provided by the 3D camera 3. Figure 1 In the diagram, only individual key points 7 are provided with reference numerals 7 in order to avoid overloading the diagram. On the basis of these key points 7, a vehicle occupant posture of the vehicle occupant 6 is determined based on a connection of the detected key points 7 with a skeleton-like representation of the body parts of the vehicle occupant 6, wherein the skeleton-like representation reflects the relative position and orientation of the individual body parts of the vehicle occupant 6.

[0058] The control device 2 is further configured to predict a future vehicle occupant posture of the vehicle occupant 6 based on the predicted future position of the at least one key point 7 and to modify a triggering algorithm of the personal restraint device based on the predicted future posture of the vehicle occupant 6 .

[0059] exist Figure 2 The method 100 performed by the control device 2 is shown and described in more detail in FIG.

[0060] Figure 2 A method for adapting the vehicle interior state based on the detected vehicle interior state of the vehicle 1 is shown. Figure 1Flow chart of a method 100 for a triggering algorithm of a personal restraint device of a vehicle 1 is shown, wherein the vehicle 1 has collided or is about to collide with a collision target, and wherein the collision target is a moving target vehicle. In step 110, key points 7 of a vehicle occupant 6 are detected by an optical sensor device, wherein the optical sensor device is a combination of an IR camera 4 and a 3D camera 3, and wherein the 2D key points of the vehicle occupant 6 detected by the IR camera 4 are converted into 3D key points 7 of the vehicle occupant 6 by fusing information provided by the 3D camera 3.

[0061] In step 112 , a vehicle occupant posture of the vehicle occupant 6 is determined based on a connection of the detected key points 7 with a skeleton-like representation of body parts of the vehicle occupant 6 , wherein the skeleton-like representation reflects the relative position and orientation of individual body parts of the vehicle occupant 6 .

[0062] At step 114, a future vehicle occupant posture of the vehicle occupant 6 is predicted based on the predicted future position of the at least one key point 7. The predicted future position of the at least one key point 7 is estimated based on a calculated future key point velocity vector, wherein the future key point velocity vector is derived from the vector sum of the estimated key point velocity vector, the current vehicle velocity vector multiplied by a first scalar parameter derived from the signal of the impact sensor, and the current collision target velocity vector multiplied by a second scalar parameter derived from the signal of the same impact sensor. In this way, the future key point velocity vector is calculated by the following formula:

[0063]

[0064] in is the future key point velocity vector, is the estimated future keypoint velocity vector, (α) is the first scalar parameter derived from the signal of the impact sensor, is the current vehicle velocity vector, (β) is a second scalar parameter derived from the signal of the impact sensor, and is the current collision target velocity vector.

[0065] The first scalar parameter and the second scalar parameter are each further determined by taking into account the vehicle size and the vehicle weight of the vehicle 1. The collision target velocity vector is derived from a signal of a surroundings sensor of the vehicle 1, such as a radar sensor, a lidar sensor, an ultrasonic sensor or a surroundings camera.

[0066] This will enable dynamic keypoint tracking and particularly accurate prediction of future keypoint positions and, therefore, future vehicle occupant poses.

[0067] At step 116 , the triggering algorithm of the personal restraint device is modified based on the predicted future posture of the vehicle occupant 6 .

[0068] Thus, an accurate determination of the interior state is ensured, so that the vehicle occupant 6 can be protected to the greatest extent possible in the event that the restraint device is triggered.

[0069] Figure 3 A control device 2 is shown for adapting a triggering algorithm for a personal restraint device of the vehicle 1 based on a detected vehicle interior state of the vehicle 1. The control device 2 is configured or designed for executing a triggering algorithm based on a detected vehicle interior state of the vehicle 1. Figure 2 Method 100.

[0070] Figure 4 1 shows a flow chart of a method 100 for adapting a triggering algorithm of a personal restraint device of a vehicle 1 based on a detected vehicle interior state of the vehicle 1 according to another embodiment. The method 100 substantially corresponds to Figure 2 The method 100 described in Figure 4 The method 100 includes several other aspects.

[0071] In step 120, key points 7 of the vehicle occupant 6 are detected by means of an optical sensor device, wherein the optical sensor device is a combination of an IR camera 4 and a 3D camera 3, and wherein the 2D key points of the vehicle occupant 6 detected by the IR camera 4 are converted into 3D key points 7 of the vehicle occupant 6 by fusing information provided by the 3D camera 3. Furthermore, a moving object held by the vehicle occupant 6 is detected by means of the IR camera 4.

[0072] In step 122 , a vehicle occupant posture of the vehicle occupant 6 is determined based on a connection of the detected key points 7 with a skeleton-like representation of body parts of the vehicle occupant 6 , wherein the skeleton-like representation reflects the relative position and orientation of individual body parts of the vehicle occupant 6 .

[0073] In step 124, a future vehicle occupant posture of the vehicle occupant 6 is predicted based on the predicted future position of the at least one key point 7. The predicted future position of the at least one key point 7 is estimated based on the calculated future key point velocity vector, wherein the future key point velocity vector is derived from the vector sum of the estimated key point velocity vector, the current vehicle velocity vector multiplied by a first scalar parameter derived from the signal of one impact sensor, and the current collision target velocity vector multiplied by a second scalar parameter derived from the signal of another impact sensor. In this way, the future key point velocity vector is calculated by the following formula:

[0074]

[0075] in is the future key point velocity vector, is the estimated future keypoint velocity vector, (α) is a first scalar parameter derived from a signal of an impact sensor, is the current vehicle velocity vector, (β) is a second scalar parameter derived from the signal of another impact sensor, and is the current collision target velocity vector.

[0076] The first scalar parameter and the second scalar parameter are each further determined by taking into account the vehicle size and the vehicle weight of the vehicle 1. The collision target velocity vector is derived from a signal of a surroundings sensor of the vehicle 1, such as a radar sensor, a lidar sensor, an ultrasonic sensor or a surroundings camera.

[0077] This will enable dynamic keypoint tracking and particularly accurate prediction of future keypoint positions and, therefore, future vehicle occupant poses.

[0078] In addition, the future state of the moving object is predicted, wherein the predicted future state of the moving object includes the future position of the moving object relative to the steering wheel in the vehicle interior 5, the future speed of the moving object, and the future orientation of the moving object. The predicted future position of the moving object is estimated based on the calculated future moving object speed vector, wherein the future moving object speed vector is obtained by the vector sum of the estimated moving object speed vector, the current vehicle speed vector multiplied by the first scalar parameter, and the current collision target speed vector multiplied by the second scalar parameter. In this way, the future moving object speed vector is calculated by the following formula:

[0079]

[0080] in is the future moving object velocity vector, is the estimated future moving object velocity vector, (α) is a first scalar parameter derived from a signal of an impact sensor, is the current vehicle velocity vector, (β) is a second scalar parameter derived from the signal of another impact sensor, and is the current collision target velocity vector.

[0081] The first scalar parameter and the second scalar parameter are each further determined by taking into account the vehicle size and the vehicle weight of the vehicle 1. The collision target velocity vector is derived from a signal of a surroundings sensor of the vehicle 1, such as a radar sensor, a lidar sensor, an ultrasonic sensor or a surroundings camera.

[0082] This will enable dynamic moving object tracking and particularly accurate prediction of future moving object positions.

[0083] In step 126 , the triggering algorithm of the personal restraint device is modified based on the predicted future posture of the vehicle occupant 6 and the predicted future moving object state.

[0084] In this way, moving objects held by the vehicle occupant 6 are also taken into account, so that the vehicle interior state can be determined more accurately, thereby further maximizing the protection of the vehicle occupant 6 in the event of a collision or impact.

Claims

1. A method (100) for adapting a triggering algorithm of a personal restraint device of a vehicle (1) based on a detected vehicle interior state of the vehicle (1), the method (100) include: - detecting (110, 120) key points (7) of a vehicle occupant (6) by means of an optical sensor device (3, 4); - determining (112, 122) a vehicle occupant posture of the vehicle occupant (6) based on a connection of the detected key points (7) with a skeleton-like representation of body parts of the vehicle occupant (6), wherein the skeleton-like representation reflects the relative position and orientation of the various body parts of the vehicle occupant (6); - predicting (114, 124) a future vehicle occupant posture of the vehicle occupant (6) based on the predicted future position of at least one key point (7); and - modifying (116, 126) the triggering algorithm of the personal restraint device based on the predicted future posture of the vehicle occupant (6), By detecting key points (7) of the vehicle occupant (6) using at least an IR camera (4) and a 3D camera (3), It is characterized in that By fusing the information provided by the 3D camera, the 2D key points of the vehicle occupant (6) detected by the IR camera (4) are converted into 3D key points (7) of the vehicle occupant (6).

2. The method (100) according to claim 1, It is characterized in that The method further comprises: - detecting (120) a moving object held by the vehicle occupant (6) by means of the optical sensor device (3, 4); - predicting (124) the future state of the moving object; and - modifying (126) the triggering algorithm of the personal restraint device based on the predicted future posture of the vehicle occupant (6) and the predicted future moving object state.

3. The method (100) according to claim 2, It is characterized in that The predicted future state of the moving object includes at least one of a future position of the moving object in the vehicle interior (5), a future speed of the moving object, or a future orientation of the moving object.

4. The method (100) according to any one of claims 2 to 3, It is characterized in that The method further comprises: - associating the position and / or movement of the mobile object with the position and / or movement of a key point (7) representing the wrist of a vehicle occupant; - determining from the association which hand of the vehicle occupant (6) is holding the moving object.

5. The method (100) according to any one of claims 1 to 3, It is characterized in that A predicted future position of at least one key point (7) is estimated based on at least one of vehicle occupant size, vehicle occupant seat position, vehicle occupant seat back angle, seat belt status, vehicle speed, or vehicle acceleration.

6. The method (100) according to claim 5, It is characterized in that The vehicle occupant size is estimated based on the size of a body part of the vehicle occupant (6).

7. The method (100) according to any one of claims 2 to 3, It is characterized in that A predicted future state of the moving object is estimated based on at least one of a vehicle occupant size, a vehicle occupant seat position, a vehicle occupant seat back angle, a seat belt status, a vehicle speed, or a vehicle acceleration.

8. The method (100) according to claim 7, It is characterized in that The vehicle occupant size is estimated based on the size of a body part of the vehicle occupant (6).

9. The method (100) according to any one of claims 1 to 3, It is characterized in that A predicted future position of at least one key point (7) is estimated based on a calculated future key point velocity vector, wherein the future key point velocity vector is derived from the vector sum of the estimated future key point velocity vector and the product of the current vehicle velocity vector multiplied by a first scalar parameter derived from a signal of an impact sensor.

10. The method (100) according to claim 9, It is characterized in that The future key point velocity vector is calculated by the following formula: in, is the velocity vector of the future key point, is the estimated future keypoint velocity vector, (α) is the first scalar parameter derived from the signal of the impact sensor, is the current vehicle velocity vector, (β) is a second scalar parameter derived from the signal of the impact sensor, and is the current collision target velocity vector.

11. The method (100) according to any one of claims 2 to 3, It is characterized in that A predicted future position of the moving object is estimated based on a calculated future moving object velocity vector resulting from a vector sum of an estimated future moving object velocity vector and a current vehicle velocity vector multiplied by a first scalar parameter derived from a signal of an impact sensor.

12. The method (100) according to claim 11, It is characterized in that The future moving object velocity vector is calculated using the following formula: in, is the velocity vector of the future moving object, is the estimated future moving object velocity vector, (α) is the first scalar parameter derived from the signal of the impact sensor, is the current vehicle velocity vector, (β) is a second scalar parameter derived from the signal of the impact sensor, and is the current collision target velocity vector.

13. The method (100) according to any one of claims 1 to 3, It is characterized in that A vehicle occupant posture of the vehicle occupant (6) is determined based on the 3D key points (7) of the vehicle occupant (6).

14. The method (100) according to any one of claims 2 to 3, It is characterized in that Only the IR camera (4) is used to detect the moving object held by the vehicle occupant (6).

15. The method according to any one of claims 1 to 3, It is characterized in that The key points (7) of the vehicle occupant (6) are the vehicle occupant skeletal joints.

16. A control device (2) for adapting a triggering algorithm of a personal restraint device of a vehicle (1) based on a detected vehicle interior state of the vehicle (1), in, The control device (2) is configured to: - detecting key points (7) of a vehicle occupant (6) by means of optical sensor devices (3, 4); - determining a vehicle occupant posture of the vehicle occupant (6) based on a connection of the detected key points (7) with a skeleton-like representation of body parts of the vehicle occupant (6), wherein the skeleton-like representation reflects the relative position and orientation of the individual body parts of the vehicle occupant (6); - predicting a future vehicle occupant posture of the vehicle occupant (6) based on the predicted future position of at least one key point (7); and - modifying the triggering algorithm of the personal restraint device based on the predicted future posture of the vehicle occupant (6), By detecting key points (7) of the vehicle occupant (6) using at least an IR camera (4) and a 3D camera (3), It is characterized in that By fusing the information provided by the 3D camera, the 2D key points of the vehicle occupant (6) detected by the IR camera (4) are converted into 3D key points (7) of the vehicle occupant (6).

Images