JP7841495B2 - Airbag system - Google Patents

Description

This disclosure relates to an airbag system.

Patent Document 1 describes an airbag system comprising a torso support section that deploys from behind the occupant's head to the front of the occupant, and a pair of head support sections that deploy on both the left and right sides of the occupant's head and connect to the torso support section. In this airbag system, when deployed, a relief section is formed between the pair of head support sections, penetrating vertically to avoid the occupant's head.

Furthermore, Patent Document 2 describes an airbag device having a rear inflation section that deploys on the back of the seat, a pair of lateral inflation sections extending forward from both the left and right sides of the rear inflation section, and a pair of forward inflation sections extending from the lateral inflation sections toward the center and connected to each other in the center to cover the front of the occupant. In this airbag device, gas passages are formed in the order of rear inflation section, lateral inflation section, and forward inflation section.

Japanese Patent Publication No. 2019-018593 Japanese Patent Publication No. 2019-218013

As described in Patent Documents 1 and 2, when an airbag is configured to include an airbag body that deploys in front of the occupant during deployment, and a pair of front and rear chambers that extend in the longitudinal direction of the vehicle during deployment, forming a loop space between them and the airbag body, the following problems arise. Specifically, if the occupant's head is not in its normal position—for example, if the occupant is lying down and leaning against the window, or if the occupant is in a hunched-over position—the occupant's head may not enter the loop space at all, or may only partially enter it, during airbag deployment. In this case, there are concerns that the deployed airbag may not provide the intended restraint performance, or that it may adversely affect the occupant.

This disclosure was made in consideration of the above facts, and its purpose is to provide an airbag system that can prevent deployed airbags from failing to provide the intended restraint performance or from adversely affecting occupants.

The first embodiment of the airbag system includes a control unit that determines whether or not a vehicle collision has occurred based on an output value from a collision sensor that detects a vehicle collision, and determines whether or not the occupant's head is located within a predetermined range corresponding to the loop space of the airbag, which includes an airbag body that deploys in front of the vehicle to the occupant when deployed and a pair of left and right front and rear chambers that extend in the front-rear direction of the vehicle when deployed and form a loop space between them and the airbag body, based on an output value from a camera that captures a range including the head of an occupant seated in a seat, and deploys the airbag when it is determined that a vehicle collision has occurred and the occupant's head is located within the predetermined range, and the control unit determines whether or not the occupant's head is located within the predetermined range based on a determination model that has been previously learned from training data including data for cases where the occupant's head is not located within the predetermined range, in addition to the output value from the camera .

In the first embodiment, the presence or absence of a vehicle collision is determined based on the output value from the collision sensor. Furthermore, based on the output value from a camera that captures a range including the occupant's head, it is determined whether the occupant's head is located within a predetermined range corresponding to the airbag's loop space. If a vehicle collision is determined to have occurred and the occupant's head is located within the predetermined range, the airbag is deployed. This prevents the airbag from failing to deploy if the occupant's head is not located within the predetermined range, thus preventing the deployed airbag from failing to provide the intended restraint performance or causing adverse effects on the occupant.

Furthermore, in the first embodiment, the control unit determines whether the occupant's head is located within the predetermined range based on a determination model that has been previously learned from training data, which includes data for cases where the occupant's head is not located within the predetermined range, in addition to the output value from the camera. This makes it possible to accurately determine whether the occupant's head is located within the predetermined range corresponding to the airbag loop space without complicating the configuration, such as by using a stereo camera.

In the second embodiment, the data for cases where the occupant's head is not located within the predetermined range in the first embodiment includes data for cases where the occupant is lying down and leaning against the window, and data for cases where the occupant is leaning forward.

According to the second embodiment, in a judgment based on a judgment model pre-trained from training data, it is possible to accurately determine if the occupant's head is not within a predetermined range when the occupant is sleeping and leaning against the window or when the occupant is leaning forward.

A third embodiment is the first embodiment, in which the learning data includes first data in which the output value of the camera when the occupant's head is not located within the predetermined range is associated with position information representing the position of the occupant's head, and second data in which the output value of the camera when the occupant's head is located within the predetermined range is associated with position information representing the position of the occupant's head.

According to the third embodiment, since the training data includes the first and second data described above, in a determination based on a determination model pre-trained from the training data, position information representing the position of the occupant's head can be obtained from the output value from the camera. Then, it is possible to accurately determine from this position information whether or not the occupant's head is located within a predetermined range.

This disclosure has the effect of preventing deployed airbags from failing to provide the intended restraint performance or from causing adverse effects on occupants.

This is a schematic block diagram of an airbag device according to an embodiment. This is a perspective view showing a vehicle airbag. This is an illustrative diagram showing the placement of the indoor camera and specific examples of head positions when it is determined that the occupant's head is not within a predetermined range. This is an illustrative diagram showing an example of the positional relationship between the occupant and the airbag in both the normal occupant position and when the occupant's head is shifted to the side. This is an illustrative diagram showing an example of the positional relationship between the occupant and the airbag in both the normal occupant position and when the occupant's head is shifted forward.

Hereinafter, an example of an embodiment of this disclosure will be described in detail with reference to the drawings. As shown in Figure 1, the airbag system 10 according to this embodiment is provided for each seat of the vehicle and includes a collision sensor 50, an interior camera 52, a control device 54, an inflator 44, and an airbag 32. Hereinafter, with reference to Figure 2, the configuration of the airbag 32 and its surroundings will be described.

The seat 12 shown in Figure 2 is either a front or rear seat of a vehicle. The seat 12 comprises a seat cushion 14, a seat back 16 erected above the rear end of the seat cushion 14, and a headrest 18 connected to the upper end of the seat back 16. An occupant D seated on the seat cushion 14 of the seat 12 is restrained by a seat belt 22 provided by the seat belt device 20. The seat belt device 20 is a three-point seat belt device, a so-called seat belt device with a retractor, anchor, and buckle (not shown) attached to the seat 12.

The airbag 32 is normally folded and stored in a module case 46 along with a pair of inflators 44. The module case 46 is formed in a hollow rectangular parallelepiped shape. The module case 46 is located at the rear of the upper part of the seat 12 (specifically, behind the upper end of the seat back 16 and the headrest 18) and is fixed to the upper end of the seat back 16 or to the vehicle body (not shown).

The airbag 32 receives gas from a pair of inflators 44 and inflates and deploys (deploys and inflates) from the rear to the front of the upper part of the seat 12. This airbag 32 includes a pair of front and rear chambers 34 and an airbag body 40. When deployed, the airbag body 40 deploys in front of the occupant D. The pair of front and rear chambers 34 extend in the longitudinal direction of the vehicle when deployed, forming a loop space 36 between them and the airbag body 40.

More specifically, the front and rear chambers 34 have a pair of front and rear extending portions 34A that inflate and deploy forward, passing on both sides of the occupant D's head H when the occupant D is in a standard position, and a connecting portion 34B that connects the front ends of the pair of front and rear extending portions 34A in the left-right direction. The airbag body 40 inflates and deploys towards the occupant D side (rear side) after the front and rear chambers 34, at the rear side of the connecting portion 34B.

The front and rear chambers 34 are formed in a long, bag-like shape, and the airbag body 40 is also formed in a bag-like shape. The longitudinal ends of the front and rear chambers 34 are a pair of left and right inflator housing sections that house a pair of inflators 44. The longitudinal center of the front and rear chambers 34 is a connecting section 34B, and the area between the connecting section 34B and the pair of inflator housing sections is the front and rear extension section 34A. The pair of inflators 44 are combustion-type or cold-gas-type cylinder inflators that generate gas when activated. The operation of these inflators 44 is controlled by a control device 54.

On the other hand, the collision sensor 50 shown in Figure 1 consists of an acceleration sensor that detects acceleration in the longitudinal direction of the vehicle. It detects a vehicle collision when it detects an acceleration of a predetermined value or higher, corresponding to a collision with an object, for a predetermined period of time or longer.

The interior camera 52 captures the area within its shooting range, including the head H of the occupant D seated in the seat 12. More specifically, the interior camera 52 includes a monocular interior camera 52F (see Figure 3(A)) for capturing the occupant D seated in the front seat, and a monocular interior camera 52R (see Figure 3(B)) for capturing the occupant D seated in the rear seat.

As shown in Figure 3(A), the interior camera 52F is mounted on the front end of the vehicle's roof and in the center in the vehicle's width direction. The camera's field of view and shooting range are adjusted so that the head H of the occupant D seated in the driver's seat is within the shooting range even when the head H of the occupant D is located at positions 70A, 70B, 70C, etc. in Figure 3(C), or when the head H of the occupant D seated in the passenger seat is located at positions 70D, 70E, 70F, etc. in Figure 3(C).

Positions 70A and 70C indicate the position of occupant D's head H when the occupant D is seated in the driver's seat and is reclined, with their head H tilted to the left or right (for example, leaning against the window). Position 70B indicates the position of occupant D's head H when the occupant D is seated in the driver's seat and is leaning forward.

Furthermore, positions 70D and 70F indicate the position of occupant D's head H when occupant D is reclined in the passenger seat and their head H is tilted to the left or right (for example, leaning against the window). Position 70E indicates the position of occupant D's head H when occupant D is leaning forward in the passenger seat.

As shown in Figure 3(B), the interior camera 52R is mounted on the roof of the vehicle, in the center of both the front-to-rear and width-to-front-to-back directions. Similar to the interior camera 52F, the field of view and shooting range of the interior camera 52R are adjusted so that the head H of the occupant D is within the shooting range even when the occupant D is reclined in the rear seat with their head H tilted to the left or right, or when the occupant D is leaning forward.

The control device 54 includes a collision determination unit 56, an occupant posture determination unit 58, and an integrated determination unit 60. The collision determination unit 56 determines whether a vehicle collision has occurred based on the output value from the collision sensor 50 and outputs the determination result to the integrated determination unit 60. The occupant posture determination unit 58 includes a memory unit that stores pre-trained determination models. Based on the output value from the interior camera 52 and the pre-trained determination model, the occupant posture determination unit 58 determines whether the occupant D's head H is located within a predetermined range corresponding to the airbag 32's loop space 36 and outputs the determination result to the integrated determination unit 60.

The above-mentioned trained judgment model is generated using training data that includes data for cases where occupant D's head H is not located within a predetermined range corresponding to the loop space 36 of the airbag 32. Specifically, the training data includes first and second data sets. The first data set is data where the output value of the interior camera 52 when occupant D's head H is not located within the predetermined range is associated with positional information (an example of training data) representing the position of occupant D's head H. The second data set is data where the output value of the interior camera 52 when occupant D's head H is located within the predetermined range is associated with positional information (an example of training data) representing the position of occupant D's head H. Furthermore, the first data set includes data for cases where occupant D is lying down and leaning against the window, and where occupant D is leaning forward. The positional information mentioned above is, for example, information such as "300 mm to the left of the center (position in standard posture)."

In this embodiment, based on the above-mentioned training data, a model is trained to output the position of occupant D's head H in the situation represented by the output value of the indoor camera 52, thereby pre-generating a trained judgment model. In this embodiment, a neural network as shown in Figure 1 can be used as an example of the model, and deep learning can be used as an example of the learning algorithm.

The occupant posture determination unit 58 obtains the determination result for the position of occupant D's head H by performing signal processing using a multilayer neural network based on the output value from the interior camera 52 and a trained determination model, each time an output value from the interior camera 52 is input. In this embodiment, this determination result is obtained as data in the format of, for example, an 80% probability that occupant D's head H is "400mm to the left of the center," a 10% probability that it is "300mm to the left of the center," a 10% probability that it is "200mm to the left of the center," and a 0% probability that it is "100mm to the left of the center." Then, in the above example, the position of occupant D's head H is determined to be "400mm to the left of the center," and by comparing "400mm to the left" with a predetermined value, it is determined whether or not occupant D's head H is located within a predetermined range.

The integrated determination unit 60 activates the inflator 44 to deploy the airbag 32 when the collision determination unit 56 determines that a vehicle collision has occurred, and the occupant posture determination unit 58 determines that the occupant's head H is within a predetermined range. The control device 54, including the collision determination unit 56, occupant posture determination unit 58, and integrated determination unit 60 described above, is an example of a control unit in this disclosure.

Next, the operation of this embodiment will be explained. As shown in Figures 4(A) and 5(A), when occupant D is seated in the seat 12 in a standard position, or when the seat back 16 is reclined and the occupant D is seated in a comfortable position, occupant D's head H is located within a predetermined range corresponding to the loop space 36. Therefore, if the vehicle collides with an object in this state and the airbag 32 deploys, as indicated by the notation "normal position," occupant D's head H will be located within the loop space 36 of the deployed airbag 32, and occupant D's head H can be appropriately restrained.

However, as shown in Figure 4(B), if occupant D is seated in seat 12 and is lying down, leaning against the window, occupant D's head H is located to the side, outside the predetermined range corresponding to the loop space 36. Therefore, if the vehicle were to collide with an object in this state and the airbag 32 deployed, there is a risk that occupant D's head H might move outside the loop space 36 of the deployed airbag 32, or that the front and rear chambers 34 of the airbag 32 might directly hit the back of occupant D's head, as indicated by the phrase "when the head is shifted to the side."

Furthermore, as shown in Figure 5(B), when occupant D is seated in seat 12 in a hunched-over position, occupant D's head H is located in front of the predetermined range corresponding to the loop space 36. Therefore, if the vehicle were to collide with an object in this state and the airbag 32 deployed, as indicated by the phrase "when the head is shifted forward," occupant D's head H would come into contact with the airbag body 40, potentially causing excessive neck flexion or even disengaging from the loop space 36 of the deployed airbag 32 by restricting only occupant D's head H.

In contrast, in this embodiment, if the occupant posture determination unit 58 determines that the occupant D's head H is not located within a predetermined range corresponding to the loop space 36, the airbag device 10 will not deploy the airbag 32 of the corresponding seat 12, even if the vehicle collides with an object in this state. This prevents the occupant D's head H from coming out of the loop space 36 of the deployed airbag 32, or the front and rear chambers 34 of the airbag 32 directly hitting the back of the occupant D's head, as shown in Figure 4(B) in a seat 12 where the seated occupant D is reclined and leaning against the window. Furthermore, as shown in Figure 5(B) in a seat 12 where the seated occupant D is leaning forward, the occupant D's head H comes into contact with the airbag body 40, restraining only the occupant D's head H, thereby preventing excessive neck flexion or the occupant D's head H from coming out of the loop space 36 of the deployed airbag 32.

In this embodiment, the control device 54 of the airbag system 10 determines whether a vehicle collision has occurred based on the output value from the collision sensor 50 that detects a vehicle collision, and also determines whether the head H of occupant D is located within a predetermined range corresponding to the loop space 36 of the airbag 32, which includes the airbag body 40 that deploys in front of the occupant when deployed and a pair of left and right front and rear chambers 34 that extend in the front-rear direction of the vehicle when deployed and form a loop space 36 between them and the airbag body 40. If a vehicle collision is determined and the head H is located within the predetermined range, the airbag 32 is deployed. This prevents the airbag 32 from not deploying if the occupant D's head H is not located within the predetermined range, thus preventing the deployed airbag 32 from failing to provide the intended restraint performance or adversely affecting occupant D.

Furthermore, in this embodiment, the control device 54 determines whether the occupant D's head H is located within a predetermined range, based on a determination model pre-trained from learning data including data from cases where the occupant D's head H is not located within a predetermined range, in addition to the output value from the interior camera 52. This allows for accurate determination of whether the occupant D's head H is located within a predetermined range corresponding to the airbag 32's loop space 36, without complicating the configuration.

Furthermore, in this embodiment, the data for cases where the occupant D's head H is not located within a predetermined range includes data for cases where the occupant D is lying down and leaning against the window, and data for cases where the occupant D is leaning forward. This allows for accurate determination, based on a pre-trained judgment model using training data, that the occupant D's head H is not located within the predetermined range when the occupant D is lying down and leaning against the window, or when the occupant D is leaning forward.

Furthermore, in this embodiment, the training data includes: first data where the output value of the indoor camera 52 when the occupant D's head H is not located within a predetermined range is associated with position information representing the position of the head H; and second data where the output value of the indoor camera 52 when the occupant D's head H is located within a predetermined range is associated with position information representing the position of the head H. This allows for accurate determination of whether the occupant D's head H is located within a predetermined range based on the position information obtained from the output value of the indoor camera 52 in a determination based on a pre-trained determination model from the training data.

In the above embodiment, a method was described in which a trained determination model is generated that outputs the position of occupant D's head H from the output value of the indoor camera 52 by using data that associates the output value of the indoor camera 52 with position information representing the position of occupant D's head H as training data. However, this disclosure is not limited to this. For example, data that associates the output value of the indoor camera 52 with identification information indicating whether or not occupant D's head H is located within a predetermined range may be used as training data. By using this training data, a trained determination model is generated that outputs whether or not occupant D's head H is located within a predetermined range from the output value of the indoor camera 52. In this case, it is possible to directly determine whether or not occupant D's head H is located within a predetermined range from the output value of the indoor camera 52 based on the generated determination model.

Furthermore, while the above embodiment describes a method of training a neural network model as an example of a model using deep learning, it is not limited to this. For example, other models different from the neural network model may be trained using other learning methods different from deep learning.

Furthermore, in the above embodiment, the determination of whether occupant D's head H is located within a predetermined range was made based on the output value from the indoor camera 52 and a determination model pre-trained from training data including data for cases where occupant D's head H is not located within a predetermined range corresponding to the loop space 36. However, this disclosure is not limited to this. For example, the indoor camera 52 may be a stereo camera to obtain the three-dimensional coordinates of occupant D's head H, and the determination of whether occupant D's head H is within a predetermined range may be made from the obtained three-dimensional coordinates.

10 Airbag system 32 Airbag 34 Front and rear chambers 36 Loop space 40 Airbag body 50 Collision sensor 52 Interior camera 54 Control unit (control unit)
56 Collision determination unit 58 Occupant posture determination unit 60 Overall determination unit

Claims (3)

The control unit determines whether a vehicle collision has occurred based on the output value from a collision sensor that detects a vehicle collision, and determines whether the occupant's head is located within a predetermined range corresponding to the loop space of the airbag, which includes an airbag body that deploys in front of the vehicle when deployed and a pair of front and rear chambers that extend in the front-rear direction of the vehicle when deployed and form a loop space between them and the airbag body, based on the output value from a camera that captures a range including the head of an occupant seated in a seat, and deploys the airbag if it determines that a vehicle collision has occurred and that the occupant's head is located within the predetermined range.
The control unit determines whether or not the occupant's head is located within the predetermined range, based on a determination model that has been previously learned from training data including data for cases where the occupant's head is not located within the predetermined range, in addition to the output value from the camera. The airbag device according to claim 1, wherein the data for cases where the occupant's head is not located within the predetermined range includes data for cases where the occupant is lying down and leaning against the window, and data for cases where the occupant is leaning forward. The airbag device according to claim 1, wherein the learning data includes first data in which the output value of the camera when the occupant's head is not located within the predetermined range is associated with position information representing the position of the occupant's head, and second data in which the output value of the camera when the occupant's head is located within the predetermined range is associated with position information representing the position of the occupant's head.