A passenger protection method, device, vehicle and storage medium

Abstract

The application provides a passenger protection method and device, a vehicle and a storage medium, and relates to the technical field of vehicles.The method comprises the following steps: determining whether a target vehicle has a potential collision danger according to acquired road traffic environment information; when the target vehicle has a potential collision danger, performing injury probability analysis on a passenger on a target seat in the target vehicle according to the acquired road traffic environment information, off-vehicle information and in-vehicle information of the target vehicle, and obtaining an injury probability result; and when the injury probability result is greater than a first preset probability, adjusting a seat position parameter of the target seat until the injury probability result is less than a second preset probability, wherein the first preset probability is greater than or equal to the second preset probability, and the passenger protection is combined with a passenger sitting posture.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicles, in particular to a passenger protection method and device, a vehicle and a storage medium. BACKGROUND

[0002] With the continuous development of intelligent and networked technologies for vehicles, the cabin presents various use scenarios, such as office, entertainment, rest, etc. In these use scenarios, the passenger sitting posture presents diversity, and the seatback angle is mostly in a large-angle form. The sitting posture has a certain correlation with the injury of the passenger in a traffic accident. Under a large-angle sitting posture, the passenger is more likely to be injured.

[0003] In the related art, the passenger protection method has technical limitations in passenger protection. It mainly considers the passenger protection method and vehicle protection after an accident occurs, such as passive safety measures such as airbag deployment and seatbelt pretensioning to reduce the injury caused by the collision. This method is usually designed and optimized based on a standard sitting posture, or the passive protection is performed after the non-standard sitting posture is instantaneously returned to the standard sitting posture. However, the instantaneous return is likely to cause injury to the passenger when a collision occurs, and the passive protection has a lag when a collision occurs. Therefore, the related art is difficult to adapt to the diversified passenger sitting posture scenarios in the current intelligent cabin, and the protection method has limitations. SUMMARY

[0004] The problem solved by the present application is how to protect the passenger in combination with the passenger sitting posture.

[0005] To solve the above problems, the present application provides a passenger protection method, device, vehicle and storage medium.

[0006] In a first aspect, the present application provides a passenger protection method, comprising:

[0007] determining whether a target vehicle has a potential collision danger according to the acquired road traffic environment information;

[0008] when the target vehicle has a potential collision danger, performing injury probability analysis on a passenger located on a target seat in the target vehicle according to the acquired road traffic environment information, off-vehicle information and in-vehicle information of the target vehicle, and obtaining an injury probability result;

[0009] when the injury probability result is greater than a first preset probability, adjusting the seat position parameter of the target seat until the injury probability result is less than a second preset probability, wherein the first preset probability is greater than or equal to the second preset probability.

[0010] Optionally, the passenger protection method further comprises:

[0011] acquiring a seat target parameter of adjusting a seat position parameter of the target seat to a situation where the injury probability result is less than a second preset probability;

[0012] determining a target restraint parameter of a passenger restraint system matched with the target seat according to the in-vehicle information, the seat target parameter, the road traffic environment information, and the out-of-vehicle information.

[0013] Optionally, after the determining of the target restraint parameter of the passenger restraint system matched with the target seat according to the in-vehicle information, the seat target parameter, the road traffic environment information, and the out-of-vehicle information, the method further comprises:

[0014] controlling the passenger restraint system to perform passenger restraint according to the target restraint parameter in response to a collision signal, and performing passenger injury probability prediction based on the collision signal to obtain a prediction probability result.

[0015] triggering an alarm and rescue when the prediction probability result meets a preset condition.

[0016] Optionally, the road traffic environment information comprises at least one of road information, road condition information, or weather information; and the determining whether the target vehicle has a potential collision risk according to the acquired road traffic environment information comprises:

[0017] determining whether at least one of the road information, the road condition information, or the weather information is abnormal;

[0018] if yes, determining that the target vehicle has a potential collision risk; and if no, determining that the target vehicle does not have a potential collision risk.

[0019] Optionally, the passenger protection method further comprises:

[0020] when the target vehicle has a potential collision risk, triggering a driving warning; and / or,

[0021] determining passenger sitting posture information of a passenger located on the target seat according to the in-vehicle information, and determining whether the passenger sitting posture information meets a safe sitting posture condition; when the passenger sitting posture information does not meet the safe sitting posture condition, triggering a safety warning.

[0022] Optionally, the out-of-vehicle information is acquired based on an out-of-vehicle sensing device.

[0023] the injury probability analysis of the passenger located on the target seat in the target vehicle according to the acquired road traffic environment information, out-of-vehicle information, and in-vehicle information of the target vehicle to obtain an injury probability result comprises:

[0024] input the road traffic environment information, the out-of-vehicle information and the in-vehicle information into a pre-trained occupant injury prediction model to determine injury probabilities of each part of the occupant located on the target seat as the injury probability result.

[0025] Optionally, the seat position parameters include a seat back angle, a seat cushion angle and a footrest angle.

[0026] The adjusting of the seat position parameters of the target seat until the injury probability result is less than a second preset probability when the injury probability result is greater than a first preset probability includes:

[0027] When the injury probability of at least one part is greater than the first preset probability, at least one of the seat back angle, the seat cushion angle and the footrest angle of the target seat is adjusted step by step.

[0028] After each adjustment, the injury probabilities of each part of the occupant located on the target seat are determined again through the occupant injury prediction model, and the seat position parameters are iteratively adjusted based on the re-determined injury probabilities until the injury probabilities of each part are all less than the second preset probability, and the seat target parameters are obtained.

[0029] Optionally, before the input of the road traffic environment information, the out-of-vehicle information and the in-vehicle information into the pre-trained occupant injury prediction model to determine the injury probabilities of each part of the occupant located on the target seat as the injury probability result, the method further includes:

[0030] The obtained first input data set and the real injury probability of each part of the occupant corresponding to the first input data set are taken as a first training sample, and each first input data set includes a set of out-of-vehicle information samples, in-vehicle information samples and road traffic environment samples.

[0031] A first machine learning model is trained through the first training sample until the first machine learning model meets a first preset convergence condition, and the occupant injury prediction model is obtained.

[0032] Optionally, the determination of the target restraint parameters of the occupant restraint system matched with the target seat according to the in-vehicle information, the seat target parameters, the road traffic environment information and the out-of-vehicle information includes:

[0033] The occupant signs of the occupant located on the target seat are obtained according to the in-vehicle information.

[0034] The collision working condition information is determined according to the road traffic environment information and the out-of-vehicle information.

[0035] inputting the occupant sign, the seat target parameter and the crash condition information into a pre-trained occupant restraint parameter optimization model to determine the target restraint parameter.

[0036] Optionally, the crash condition information includes at least one of a crash type, a speed, a relative lateral displacement and an angle.

[0037] Optionally, the target restraint parameter includes at least one of an airbag ignition time, a seat belt pretensioning time, a seat belt force limiting value and a seat belt waistband pretensioning time.

[0038] Optionally, before inputting the occupant sign, the seat target parameter and the crash condition information into a pre-trained occupant restraint parameter optimization model to determine the target restraint parameter, the method further comprises:

[0039] inputting the obtained second input data set and the real restraint parameter of the occupant restraint system corresponding to the second input data set as a second training sample, each of the second input data set including a set of occupant sign training data, seat parameter training data and crash condition training data;

[0040] based on the second training sample, optimizing and training a second machine learning model through a pre-set optimization algorithm until the second machine learning model meets a second pre-set convergence condition, to obtain the occupant restraint parameter optimization model.

[0041] Optionally, the method further comprises:

[0042] obtaining a prediction probability result by predicting an occupant injury probability based on the crash signal and the crash condition information through an occupant injury prediction model, wherein the occupant injury prediction model is trained by traffic accident data samples including data of the target vehicle.

[0043] In a second aspect, the present application further provides an occupant protection device, comprising:

[0044] a driving safety warning control module configured to determine whether the target vehicle is in potential collision danger according to the obtained road traffic environment information;

[0045] The seating posture adjustment control module is used to perform injury probability analysis on the occupant in the target seat inside the target vehicle based on the acquired road traffic environment information, external information and internal information of the target vehicle when there is a potential collision hazard in the target vehicle, and obtain an injury probability result; it is also used to adjust the seat position parameters of the target seat when the injury probability result is greater than a first preset probability, until the injury probability result is less than a second preset probability, wherein the first preset probability is greater than or equal to the second preset probability.

[0046] Thirdly, the present invention also provides a vehicle, including a memory and a processor;

[0047] The memory is used to store computer programs;

[0048] The processor is configured to implement the occupant protection method as described above when executing the computer program.

[0049] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the occupant protection method as described above.

[0050] The beneficial effects of the occupant protection method of the present invention are:

[0051] By acquiring road traffic environment information during vehicle movement, potential collision hazards can be identified in advance. When a potential collision hazard exists, road traffic environment information, external vehicle information, and internal vehicle information are further integrated to analyze the probability of injury to occupants in the target seats. Based on this, when the probability of injury is greater than a first preset probability, the seat position parameters are proactively adjusted until the probability of injury drops below a second preset probability. Proactive seat adjustment only occurs when there is a potential collision hazard and the probability of occupant injury is high. This reduces false triggering, improves the accuracy of occupant protection, and balances occupant comfort and reliability. Adjusting the seat in advance based on pre-collision predictions improves the effectiveness of occupant protection. Furthermore, adjusting the seat at normal speed before a collision avoids secondary injuries that might result from forcibly straightening the seat at the moment of impact. This achieves proactive optimization of seat position based on the occupant's actual sitting posture during the collision avoidance phase, enhancing the effectiveness and safety of occupant protection in non-standard sitting positions. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating the occupant protection method according to an embodiment of the present invention;

[0053] Figure 2 This is another schematic flowchart of the occupant protection method according to an embodiment of the present invention;

[0054] Figure 3 Flow chart of the running phase of the occupant protection method of the embodiment of the present application;

[0055] Figure 4 Flow chart of the seat position step adjustment of the occupant protection method of the embodiment of the present application;

[0056] Figure 5 Flow chart of the collision avoidance phase of the occupant protection method of the embodiment of the present application;

[0057] Figure 6 System block diagram of the occupant protection system of the embodiment of the present application;

[0058] Figure 7 Example diagram of the vehicle of the embodiment of the present application. DETAILED DESCRIPTION

[0059] In order to make the above objectives, features and advantages of the present application more apparent, specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Although some embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms, and should not be interpreted as being limited to the embodiments set forth herein, but rather, these embodiments are provided so as to more completely and thoroughly understand the present application. It should be understood that the drawings and embodiments of the present application are for exemplary purposes only, and are not intended to limit the scope of protection of the present application.

[0060] It should be understood that each of the steps recited in the method embodiments of the present application can be executed in different orders, and / or in parallel. In addition, the method embodiments can include additional steps and / or omit the execution of the steps shown. The scope of the present application is not limited in this respect.

[0061] The term "comprising" and variations thereof as used herein are open-ended, that is "including, but not limited to"; the term "based on" is "based, at least in part, on"; the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments"; the term "optional" means "optional in at least some embodiments". Related definitions are given throughout the detailed description. It should be noted that the concepts mentioned in the present application are merely used to distinguish different devices, modules or units, and are not intended to limit the functions performed by these devices, modules or units, or the sequence or interdependence of these functions.

[0062] It should be noted that the modification of "one" or "multiple" mentioned in the present application is illustrative rather than limiting, and those skilled in the art should understand that, unless otherwise explicitly indicated in the context, it should be understood as "one or more".

[0063] The names of the messages or information exchanged between the multiple devices in the embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0064] like Figure 1 As shown, an embodiment of the present invention provides an occupant protection method, comprising:

[0065] Step S100: Determine whether the target vehicle has a potential collision hazard based on the acquired road traffic environment information.

[0066] The occupant protection method in this invention aims to provide end-to-end protection for occupants by determining whether there is a potential collision hazard in the road traffic environment before a collision. For example, road traffic environment information includes information generated by the road traffic environment obtained through the Internet, such as speed limits, congestion near the vehicle, accident situations, or weather information received through the network. When at least one of the above information is abnormal, it indicates that the probability of an accident is higher than in normal traffic conditions, triggering a pre-collision driving safety warning to remind the driver to pay attention and remind the occupants to maintain proper seating posture.

[0067] Step S200: When the target vehicle has a potential collision hazard, the occupant in the target seat inside the target vehicle is subjected to damage probability analysis based on the obtained road traffic environment information, external information and internal information of the target vehicle, and damage probability results are obtained.

[0068] Occupant injury probability analysis is a process of quantitatively assessing the degree of injury that occupants may suffer under specific collision conditions by integrating data on the vehicle's external environment and the occupants' internal state.

[0069] When a vehicle presents a potential hazard, proactive predictions are made based on both external and internal vehicle information. By identifying potential risks in advance and making corresponding adjustments, the actual severity of occupant injury in a collision can be effectively reduced. External information includes data acquired by onboard sensors, such as images and point clouds obtained through radar and cameras. Internal information includes data acquired by onboard sensors, such as two-dimensional and three-dimensional images or point clouds of the vehicle interior obtained through cameras, 3D cameras, and radar. Damage probability analysis is performed using both external and internal information, incorporating occupants into the assessment system. This ensures that safety protection measures are more closely aligned with actual occupant characteristics and provides an objective basis for the dynamic adjustment of safety system parameters.

[0070] Step S300: When the damage probability result is greater than the first preset probability, the seat position parameters of the target seat are adjusted until the damage probability result is less than the second preset probability, wherein the first preset probability is greater than or equal to the second preset probability.

[0071] The first preset probability and the second preset probability represent preset injury risk thresholds, wherein the first preset probability represents a critical value of a higher risk, and the second preset probability represents a critical value of a lower risk. When the target vehicle has a potential collision danger, an occupant on a target seat is subjected to injury probability analysis according to the acquired road traffic environment information, the off-vehicle information and the in-vehicle information. If the injury probability result is greater than the first preset probability, the seat position parameter is adjusted until the obtained injury probability result is less than the second preset probability, and the seat position parameter corresponding to this time is determined as a seat target parameter.

[0072] After triggering the driving safety warning before the collision, the system enters the collision avoidance stage. In this stage, since there is a certain probability of collision, the system evaluates the injury risk of the occupant based on the current sitting posture of the occupant and the environmental information. When the injury probability is higher than the first preset probability, it indicates that the current seat position parameter has a higher safety risk, and the seat position needs to be optimized in advance to reduce the harm caused by the potential collision. By directly adjusting the seat position to a state where the injury probability is lower than the second preset probability, the failure of the restraint system or the abnormal force on the occupant caused by the seat being in an unsafe posture when the collision occurs can be effectively avoided.

[0073] The adjustment of the seat position is directly associated with the injury risk of the occupant, so that the protection strategy is based on the objective injury probability evaluation result, rather than relying on fixed rules or experience settings, thereby achieving more accurate and effective occupant protection in diversified sitting posture scenarios.

[0074] In the embodiments of the present application, by acquiring the road traffic environment information during the driving stage of the vehicle, it can be identified in advance whether the target vehicle has a potential collision danger. When there is a potential collision danger, the injury probability of the occupant on the target seat is analyzed by further fusing the road traffic environment information, the off-vehicle information and the in-vehicle information. On this basis, when the injury probability result is greater than the first preset probability, the seat position parameter is actively adjusted until the injury probability is reduced to be lower than the second preset probability, so as to obtain the seat target parameter adapted to the current sitting posture of the occupant and the risk scenario. The secondary harm caused by forcibly returning the seat at the moment of collision is avoided, the forward-looking optimization of the seat position in combination with the actual sitting posture of the occupant in the collision avoidance stage is realized, and the effectiveness and safety of the occupant protection under the non-standard sitting posture are improved.

[0075] Optionally, as shown in Figure 2 the occupant protection method further comprises:

[0076] Step S400, acquiring a seat target parameter of adjusting the seat position parameter of the target seat to a state where the injury probability result is less than the second preset probability;

[0077] Step S500, determining a target restraint parameter of an occupant restraint system matched with the target seat according to the in-vehicle information, the seat target parameter, the road traffic environment information and the out-of-vehicle information.

[0078] The target restraint parameter is an occupant restraint system parameter determined to meet the occupant safety requirement, including airbag and seatbelt active protection parameters.

[0079] By the in-vehicle information, the sitting posture and state of the in-vehicle occupant are considered, the out-of-vehicle driving environment is perceived by the road traffic environment information obtained by networking, and the real-time state outside the vehicle is perceived by the out-of-vehicle information. The above three kinds of information are combined, the determination of the restraint system parameter is based on comprehensive environment and occupant state data, and the restraint system parameter is more scientific and reasonable. By integrating multi-source information, the pertinence and effectiveness of the safety protection measures are improved. The human judgment error is reduced, and the occupant safety protection level is improved. By combining the in-vehicle information, the problem that the restraint system cannot be self-adaptively adjusted due to the change of the sitting posture of the occupant is solved. The target restraint parameter can be used to control the occupant restraint system to perform occupant restraint in advance before the collision or to control the occupant restraint system to perform occupant restraint when the collision occurs.

[0080] Optionally, as shown in Figure 2 The occupant protection method further comprises:

[0081] Step S600, in response to a collision signal, controlling the occupant restraint system to perform occupant restraint according to the target restraint parameter, and predicting the occupant injury probability based on the collision signal to obtain a prediction probability result.

[0082] In an embodiment, the collision signal is triggered by the acceleration of the vehicle body or the pressure received by the vehicle body, and can also be triggered by sensors such as radars or cameras provided by the vehicle, or triggered by networking signals.

[0083] In the collision stage, the occupant restraint system is triggered according to the target restraint parameter determined in advance, so as to ensure that the restraint parameter is the optimal parameter of the restraint system under the current situation. At the same time of triggering the restraint system, the occupant injury probability is predicted based on the collision signal, so as to ensure that the alarm and rescue are triggered at the first time.

[0084] Step S700, when the prediction probability result meets a preset condition, triggering an alarm and rescue.

[0085] When it is determined that the occupant may be injured, i.e., the prediction probability result meets the preset condition, an alarm and rescue are triggered.

[0086] For example, the prediction probability result includes prediction results of the injury probability of each part of the occupant's body. When any occupant has a part with an injury probability exceeding the injury probability, an accident alarm and rescue are triggered.

[0087] In embodiments of the present invention, during the driving phase, potential hazards are identified before a collision using road traffic environment information, and a collision avoidance phase is initiated after the potential hazard is identified. Occupant injury probability analysis is performed by combining external and internal vehicle information. Seat position parameters are adjusted incrementally based on the occupant injury probability until the occupant injury probability is less than a second preset probability, avoiding additional damage caused by the seat instantly returning to its original position during a collision. Before a collision, seat target parameters and appropriate target constraint parameters for the current vehicle driving state are predicted based on the target parameters of the seat inside the vehicle, internal vehicle information, and external vehicle information. Occupant constraints are applied based on these target constraint parameters during a collision, resolving the adaptive adjustment problem between occupant posture and the constraint system. After a collision, occupant injury probability is predicted, and when the prediction result meets a preset condition (i.e., a certain probability of occupant injury is predicted), an alarm is triggered, achieving end-to-end protection from the driving phase, collision avoidance phase, collision phase, and accident rescue.

[0088] Optionally, the road traffic environment information includes at least one of road information, road condition information, or weather information; the step of determining whether the target vehicle has a potential collision hazard based on the acquired road traffic environment information includes:

[0089] Determine whether at least one of the road information, traffic condition information, or weather information is abnormal;

[0090] If yes, then the target vehicle is determined to have a potential collision risk; if no, then the target vehicle is determined not to have a potential collision risk.

[0091] In one embodiment, road traffic environment information refers to external environmental data such as road infrastructure conditions, traffic flow status, and weather conditions obtained through the Internet, including basic road information such as road type and speed limit requirements, real-time road condition information such as traffic congestion and road construction, and meteorological elements such as visibility and road surface slipperiness.

[0092] A potential collision hazard indicates that at least one of the above information items is in an abnormal state, such as a sudden change in the road speed limit from high to low, traffic congestion in road traffic environment information, or weather conditions such as rain, snow, or fog. By determining whether potential collision hazards exist in the road traffic environment information, an initial benchmark for environmental risk assessment is established. Using external environmental anomalies as trigger conditions for safety warnings allows for timely responses based on changes in environmental risks, enabling early identification of potential risk environments and avoiding the limitations of relying solely on the vehicle's own condition for judgment.

[0093] Optionally, such as Figure 3 As shown, the occupant protection method further includes:

[0094] When the target vehicle presents a potential collision hazard, a driving warning is triggered; and / or,

[0095] The occupant sitting posture information of the occupant located on the target seat is determined according to the in-vehicle information, and it is judged whether the occupant sitting posture information meets the safe sitting posture condition; when the occupant sitting posture information does not meet the safe sitting posture condition, a safety warning is triggered.

[0096] In an embodiment, when it is determined that the road traffic environment has potential danger, a driving warning is triggered, for example, a text warning on the vehicle screen or a vehicle speed reminder warning is triggered to remind the driver to reduce speed.

[0097] The in-vehicle information represents the occupant sitting posture related data obtained by the in-vehicle sensing device such as 3D camera, radar, etc., including body posture angle, limb position and contact state with the seat and other parameters.

[0098] The occupant injury probability analysis represents the evaluation based on the in-vehicle information and the out-of-vehicle information, which is used to determine the injury risk level of the occupant in potential collision. The safe sitting posture condition represents the criterion for the occupant sitting posture not to meet the safety requirement, including the standards such as feet off the footboard, upper body not straight, shoulders not naturally relaxed, etc. The safety warning includes the prompt such as text warning on the vehicle screen or safety belt vibration reminder, etc.

[0099] In an embodiment, the 3D camera is used to detect whether the body part meets the safe sitting posture condition, for example, to detect whether the feet are placed on the footboard, whether the upper body is straight, whether the shoulders are naturally relaxed, whether the arms are naturally bent, whether the headrest is flush with the head, etc.; or the image recognition model such as YOLO, OpenPose, etc. is used to detect whether the body part meets the safe sitting posture condition.

[0100] In an embodiment, the injury probability result represents the injury probability result of different parts of the occupant in the current sitting posture, for example, the injury probability of each part of the occupant such as head, chest, abdomen, lumbar spine, pelvis, lower limbs, etc. is taken as the injury probability result.

[0101] In an embodiment, the in-vehicle sensing device is used to obtain the occupant sitting posture information, and the machine learning model is used to convert the input data such as occupant sitting posture image or point cloud into the occupant sitting posture information. For example, the machine learning model includes the model such as YOLO, OpenPose, etc. The sitting posture information includes the image information, point cloud information or other information obtained by the in-vehicle sensing device.

[0102] By monitoring the occupant sitting posture state in real time, the occupant sitting posture state is included in the safety warning system, so that the safety warning is more in line with the actual occupant situation; after the environmental risk is triggered, the occupant's own state is further evaluated, so that the accuracy and necessity of the warning are improved; through the double judgment of the sitting posture and the environmental risk, the false positive rate is reduced, and the effectiveness of the occupant safety protection is improved.

[0103] Optionally, the out-of-vehicle information is acquired based on an out-of-vehicle sensing device;

[0104] The damage probability analysis on the occupant on the target seat in the target vehicle based on the acquired road traffic environment information, out-of-vehicle information and in-vehicle information of the target vehicle includes:

[0105] The road traffic environment information, the out-of-vehicle information and the in-vehicle information are input into a pre-trained occupant damage prediction model to determine the damage probability of each part of the occupant on the target seat as the damage probability result.

[0106] The out-of-vehicle information represents the environmental data outside the vehicle acquired in real time by the out-of-vehicle sensing device, including speed, acceleration, position of other vehicles and other data. The in-vehicle information represents the parameters related to the state of the occupant acquired by the in-vehicle sensing device, including seat position parameters and occupant posture data. The occupant damage probability analysis represents the process of calculating the damage risk of each part of the occupant based on the out-of-vehicle information and the in-vehicle information through the seat parameter optimization model.

[0107] The seat position parameters are dynamically associated with the collision environment, and the damage of the occupant in the current posture is evaluated and predicted based on the actual scene data to provide a quantitative basis for seat parameter optimization. Through quantitative analysis, scientific optimization of seat parameters is achieved; the uncertainty caused by relying only on experience adjustment is avoided; the damage evaluation is more in line with the actual collision scene, and the effectiveness of the safety protection measures is improved.

[0108] Optionally, as shown in Figure 4 The seat position parameters include a seat back angle, a seat cushion angle and a foot support angle.

[0109] The adjustment of the seat position parameters of the target seat until the damage probability result is less than a second preset probability when the damage probability result is greater than a first preset probability includes:

[0110] When the damage probability of at least one part is greater than the first preset probability, at least one of the seat back angle, the seat cushion angle and the foot support angle of the target seat is adjusted step by step.

[0111] The damage probability of each part of the occupant on the target seat is determined again through the occupant damage prediction model after each adjustment, and the seat position parameters are iteratively adjusted based on the re-determined damage probability until the damage probability of each part is less than the second preset probability, and the seat target parameters are obtained.

[0112] Step adjustment means that the seat position parameters, including seat back angle, cushion angle and foot support angle, are adjusted in small steps, which directly affects the force distribution of the occupant in the collision.

[0113] In an embodiment, the seat back angle, cushion angle and foot support angle have different iteration steps and iteration ranges, for example, the iteration step of the back angle of a certain type of seat is 10°, the iteration step of the cushion angle is ±5°, and the iteration step of the leg support angle is ±15°; the iteration range of the seat back angle is 20-70°, the iteration range of the cushion angle is 0-30°, and the iteration range of the leg support angle is 0-90°.

[0114] By continuously iterating the injury probability, the seat parameters are gradually optimized to be within the safe range, ensuring that the final seat position meets the safety requirements. Avoiding the discomfort caused by one-time large adjustment of the seat makes the adjustment process smoother and more natural; through iterative evaluation, the final seat parameters meet the safety requirements; based on the quantitative indicators of injury probability, the adjustment is more reasonable.

[0115] Optionally, before the road traffic environment information, the vehicle exterior information and the vehicle interior information are input into the pre-trained occupant injury prediction model to determine the injury probability of each part of the occupant seated on the target seat as the injury probability result, the method further comprises:

[0116] The obtained first input data set and the real injury probability of each part of the occupant corresponding to the first input data set are used as first training samples, and each first input data set includes a set of vehicle exterior information samples, vehicle interior information samples and road traffic environment samples.

[0117] The first machine learning model is trained by the first training sample until the first machine learning model meets the first preset convergence condition, and the occupant injury prediction model is obtained.

[0118] The vehicle exterior information sample includes environmental parameters such as collision type, collision speed, relative lateral displacement and angle; the vehicle interior information sample includes seat structure parameters such as back angle, cushion angle and foot support angle; the real injury probability represents the injury probability data of each part of the occupant obtained in the actual collision experiment or the actual accident record, which is used as the reference standard for model training.

[0119] The first training sample is composed of an out-of-vehicle information sample, an in-vehicle information sample, and a true injury probability, and constitutes a data set required for model training. The first machine learning model refers to an initial prediction model used to learn the relationship between the out-of-vehicle information, the in-vehicle information, and the injury probability. The first preset convergence condition represents an evaluation standard for completion of model training, for example, the first preset convergence condition can be set as: the model prediction error is lower than a certain threshold or the accuracy rate reaches a certain requirement. The occupant injury prediction model represents a final model obtained by training, which is used to predict the injury probability according to the out-of-vehicle information and the in-vehicle information. The prediction model is constructed by historical data, which improves the objectivity and reliability of injury probability prediction; avoids the limitation of relying only on subjective experience for seat adjustment; and through the preset training completion condition, the model is ensured to meet the requirements.

[0120] In an embodiment, a back propagation artificial neural network (BP-ANN) model is used as the first machine learning model to effectively handle the nonlinear relationship between the out-of-vehicle information, the in-vehicle information, and the injury probability, and to accurately predict the injury probability by learning the patterns in the historical data.

[0121] Optionally, as shown in Figure 5 determining a target restraint parameter of an occupant restraint system matched with the target seat according to the in-vehicle information, the seat target parameter, the road traffic environment information, and the out-of-vehicle information includes:

[0122] obtaining an occupant sign of an occupant located in the target seat according to the in-vehicle information;

[0123] determining collision condition information according to the road traffic environment information and the out-of-vehicle information;

[0124] inputting the occupant sign, the seat target parameter, and the collision condition information into a pre-trained occupant restraint parameter optimization model to determine the target restraint parameter.

[0125] The occupant sign data obtained inside the vehicle is obtained through in-vehicle information, including occupant sign percentiles, for example, dividing the occupant sign into 5 percentile, 50 percentile, and 95 percentile signs. For example, based on the millimeter wave radar or 3D camera arranged in the vehicle, the occupant three-dimensional model is generated through millimeter wave or 3D structured light technology, and the occupant three-dimensional model is compared with the model stored in the database to determine the occupant sign percentile.

[0126] In an embodiment, the seat target parameter represents an optimal seat position parameter obtained after optimization of the seat parameter, including one or more of a seat back angle, a cushion reclining angle, and a footrest angle. The crash condition information includes crash condition parameters such as a crash type, a crash speed, a relative lateral displacement, and an angle. For example, based on CAN bus data collection of vehicle speed, steering angle, brake state, etc., combined with GPS positioning information, the crash condition can be determined through an external sensor network. When the vehicle suddenly brakes and turns, the CAN bus records the brake force, speed change, steering angle, etc., combined with GPS data, to determine the crash type and crash speed; the vehicle position and speed information can also be provided through the vehicle-mounted GPS system; and various sensor data such as radar, lidar, and camera can be collected through the vehicle-mounted DAQ system.

[0127] The target constraint parameter represents a passenger constraint system parameter determined to meet the passenger safety requirement, including an airbag ignition time, a seat belt pretensioning time, a seat belt force limit value, and a seat belt waist belt pretensioning time.

[0128] The passenger constraint parameter optimization model represents a prediction model for determining the target constraint parameter, which learns the patterns in historical data to accurately predict the constraint parameter. The passenger individual characteristics, seat state, and crash environment are dynamically associated, and the target constraint parameter is determined based on comprehensive scene data. The constraint system parameter can adapt to different passenger characteristics and environmental conditions, avoiding inadequate protection due to changes in passenger posture.

[0129] Optionally, the crash condition information includes at least one of a crash type, a speed, a relative lateral displacement, and an angle.

[0130] And / or, the target constraint parameter includes at least one of an airbag ignition time, a seat belt pretensioning time, a seat belt force limit value, and a seat belt waist belt pretensioning time.

[0131] Optionally, before inputting the passenger vital signs, the seat target parameter, and the crash condition information into the pre-trained passenger constraint parameter optimization model to determine the target constraint parameter, the method further includes:

[0132] The acquired second input data set and the real constraint parameter of the passenger constraint system corresponding to the second input data set are used as a second training sample, and each second input data set includes a set of passenger vital sign training data, seat parameter training data, and crash condition training data.

[0133] Based on the second training sample, a second machine learning model is optimized and trained through a preset optimization algorithm until the second machine learning model meets a second preset convergence condition, and the passenger constraint parameter optimization model is obtained.

[0134] The occupant sign data represents the occupant sign parameter data used for model training; the seat parameter training data represents the seat parameter data used for model training; and the crash condition training data represents the crash environment data used for model training.

[0135] The real constraint parameter represents the occupant constraint system parameter and the seat parameter obtained in the actual crash experiment or the actual crash accident; the real crash result represents the occupant injury result obtained in the actual crash experiment; and the second training sample is composed of the occupant sign training data, the seat parameter training data, the crash condition training data, the real constraint parameter and the real crash result, forming a data set required for model training.

[0136] The second machine learning model represents an initial prediction model used for learning the relationship between the occupant sign, the seat parameter, the crash condition and the constraint parameter. The second preset convergence condition represents an evaluation standard for completion of model training, such as a model prediction error lower than a specific threshold or an accuracy rate reaching a specific requirement.

[0137] The occupant constraint parameter optimization model represents a final model obtained by training, and is used for determining the target constraint parameter matched with the occupant sign, the seat parameter and the crash condition.

[0138] The prediction model is constructed by using the historical data, so as to improve the objectivity and reliability of the constraint parameter determination, and avoid the limitation of setting the constraint parameter only by relying on subjective experience.

[0139] In an embodiment, a model for adapting complex data, such as a radial basis function model, a Kriging model or a support vector machine model, is used for training; and a multi-parameter and global optimization algorithm, such as a multi-island genetic algorithm or a particle swarm optimization algorithm, is used for optimization.

[0140] Optionally, as shown in Figure 5 controlling the occupant constraint system according to the target constraint parameter to perform occupant constraint, and performing occupant injury probability prediction based on the crash signal to obtain a prediction probability result, includes:

[0141] performing occupant injury probability prediction based on the crash signal and the crash condition information by using an occupant injury prediction model to obtain the prediction probability result, wherein the occupant injury prediction model is trained by using traffic accident data samples including data of the target vehicle.

[0142] The crash signal represents a signal generated by the vehicle during the crash process, including signals such as acceleration change of the vehicle body and pressure fluctuation, for example, acceleration of the vehicle body in a direction other than the driving direction and greater than a threshold value and / or pressure received exceeding a threshold value. In some embodiments, the crash signal can also be triggered by sensors such as radars or cameras provided on the vehicle, or triggered by a network connection signal.

[0143] The collision condition information represents environmental parameters when the collision occurs, including collision type, collision speed, relative lateral displacement, and angle information.

[0144] The occupant injury prediction model represents a prediction tool trained through traffic accident data, used to calculate the occupant injury probability.

[0145] The traffic accident data sample represents the collision scene and occupant injury record collected in the experiment or in the actual traffic accident. Incorporating real accident data into the injury prediction process makes the prediction result more consistent with the actual collision situation.

[0146] Based on experimental data or accident data, the prediction reliability is improved; the deviation caused by simply relying on simulation data is avoided; the injury prediction result is closer to the actual collision environment, and the pertinence and timeliness of the rescue response are improved.

[0147] As shown in Figure 6 , the embodiment of the present application provides an occupant protection system, including a signal layer, a model layer, a control layer, and an execution layer.

[0148] The signal layer is used to provide at least one of road traffic environment information, external information, and internal information in different stages based on vehicle-road cloud cooperation. The stages include a driving stage, an avoidance stage, a collision stage, and a rescue stage.

[0149] The model layer is used to construct a driving monitoring model, a seat parameter optimization model, an occupant restraint parameter optimization model, and an occupant injury prediction model deployed in the cloud based on artificial intelligence methods according to the requirements of different stages. The driving monitoring model is used to determine whether there is a potential collision danger according to the acquired road traffic environment information; it is also used to analyze the occupant injury probability according to the acquired external information and internal information when there is a potential collision danger, and obtain the injury probability result. The seat parameter optimization model is used to step adjust the seat position parameter when the injury probability of at least one part is greater than the first preset probability, until the injury probability of each part is less than the second preset probability, and obtain the target seat parameter, wherein the first preset probability is greater than or equal to the second preset probability. The occupant restraint parameter optimization model is used to determine the target restraint parameter according to the internal information, the target seat parameter, the road traffic environment information, and the external information. The occupant injury prediction model is used to predict the occupant injury probability based on the collision signal and obtain the prediction probability result.

[0150] The control layer is used to issue driving safety warning control, step-by-step sitting posture adjustment control, occupant restraint system parameter dynamic control, and accident rescue cooperation control instructions based on the model analysis result.

[0151] The execution layer is configured to perform corresponding actions based on the control instructions.

[0152] The passenger protection device provided by the embodiment of the present application comprises:

[0153] The driving safety warning control module is configured to determine whether the target vehicle has potential collision danger according to the acquired road traffic environment information.

[0154] The sitting posture adjustment control module is configured to perform injury probability analysis on a passenger on a target seat in the target vehicle according to the acquired road traffic environment information, off-vehicle information and in-vehicle information of the target vehicle when the target vehicle has potential collision danger, and obtain an injury probability result; and further configured to adjust a seat position parameter of the target seat until the injury probability result is less than a second preset probability when the injury probability result is greater than a first preset probability, wherein the first preset probability is greater than or equal to the second preset probability.

[0155] The passenger restraint system parameter dynamic control module is configured to determine a target restraint parameter of a passenger restraint system matched with the target seat according to the in-vehicle information, the seat target parameter, the road traffic environment information and the off-vehicle information; and further configured to control the passenger restraint system to perform passenger restraint according to the target restraint parameter in response to a collision signal, and perform passenger injury probability prediction based on the collision signal to obtain a prediction probability result.

[0156] The accident rescue coordination control module is configured to trigger an alarm rescue when the prediction probability result meets a preset condition.

[0157] As shown in Figure 7 The vehicle 700 provided by the embodiment of the present application comprises the passenger protection method and the passenger protection system as described above.

[0158] A vehicle 700 that can be a server or a client of the present application will now be described, which is an example of a hardware device that can be applied to various aspects of the present application. The vehicle 700 comprises various forms of digital electronic computer devices, such as a laptop computer, a desktop computer, a workstation, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The vehicle 700 can also comprise various forms of mobile devices, such as a personal digital assistant, a cellular phone, a smart phone, a wearable device, and other similar computing devices. The components shown herein, their connections, and relationships, and their functions, are merely examples and are not intended to limit realizations of the present application described and / or claimed herein.

[0159] The vehicle 700 includes a computing unit that can perform various appropriate actions and processes in accordance with a computer program stored in a read only memory (ROM) or a computer program loaded from a storage unit into a random access memory (RAM). In the RAM, various programs and data required for device operation can also be stored. The computing unit, the ROM, and the RAM are connected to each other through a bus. An input / output (I / O) interface is also connected to the bus.

[0160] The computer readable storage medium provided by the embodiment of the present application stores a computer program, and when the computer program is executed by a processor, the computer program implements the occupant protection method described above.

[0161] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the program can be stored in a computer readable storage medium. When the program is executed, the program can include the processes of the above-mentioned embodiment methods. The storage medium can be a magnetic disc, an optical disc, a read-only memory (ROM), a random access memory (RAM), or the like. In the present application, the units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e., they can be located in one place or distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment of the present application. In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0162] Although the present application is disclosed as above, the protection scope of the present application is not limited to this. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present application, and these changes and modifications will fall within the protection scope of the present application.

Claims

1. A method of occupant protection, characterized by, include: Based on the obtained road traffic environment information, determine whether the target vehicle poses a potential collision hazard; When the target vehicle has a potential collision hazard, the occupant in the target seat inside the target vehicle is subjected to injury probability analysis based on the obtained road traffic environment information, external information and internal information of the target vehicle, and the injury probability result is obtained. When the damage probability result is greater than a first preset probability, the seat position parameters of the target seat are adjusted until the damage probability result is less than a second preset probability, wherein the first preset probability is greater than or equal to the second preset probability, and the seat position parameters include the seat back angle, seat cushion angle, and footrest angle; adjusting the seat position parameters of the target seat until the damage probability result is less than the second preset probability when the damage probability result is greater than the first preset probability includes: When the probability of damage to at least one part is greater than the first preset probability, at least one of the angles of the seat back, the seat cushion, and the footrest of the target seat is adjusted in a stepwise manner. After each adjustment, the probability of injury to each part of the occupant located on the target seat is re-determined by the occupant injury prediction model. The seat position parameters are iteratively adjusted based on the re-determined injury probabilities until the probability of injury to each part is less than the second preset probability, thereby obtaining the target parameters of the seat.

2. The occupant protection method according to claim 1, characterized by, The occupant protection method also includes: Obtain the seat position parameters of the target seat so that the damage probability result is less than the second preset probability; The target constraint parameters of the occupant restraint system that matches the target seat are determined based on the in-vehicle information, the target seat parameters, the road traffic environment information, and the external vehicle information.

3. The occupant protection method according to claim 2, characterized by, After determining the target constraint parameters of the occupant restraint system matching the target seat based on the in-vehicle information, the target seat parameters, the road traffic environment information, and the external vehicle information, the method further includes: In response to a collision signal, the occupant restraint system is controlled to perform occupant restraint according to the target constraint parameters, and occupant injury probability is predicted based on the collision signal to obtain the prediction probability result; When the predicted probability result meets the preset conditions, an alarm and rescue are triggered.

4. The occupant protection method according to claim 1, characterized by, The road traffic environment information includes at least one of road information, road condition information, or weather information; the step of determining whether the target vehicle has a potential collision hazard based on the acquired road traffic environment information includes: Determine whether at least one of the road information, traffic condition information, or weather information is abnormal; If yes, then the target vehicle is determined to have a potential collision risk; if no, then the target vehicle is determined not to have a potential collision risk.

5. The occupant protection method according to any one of claims 1-4, characterized by, The occupant protection method also includes: When the target vehicle presents a potential collision hazard, a driving warning is triggered; and / or, According to the in-vehicle information, passenger sitting posture information of a passenger located on the target seat is determined, and it is determined whether the passenger sitting posture information meets a safe sitting posture condition; when the passenger sitting posture information does not meet the safe sitting posture condition, a safety warning is triggered.

6. The occupant protection method according to any one of claims 1-4, characterized by, The out-of-vehicle information is obtained based on an out-of-vehicle sensing device. The damage probability analysis of the passenger located on the target seat in the target vehicle based on the obtained road traffic environment information, out-of-vehicle information and in-vehicle information of the target vehicle includes: The road traffic environment information, out-of-vehicle information and in-vehicle information are input into the pre-trained passenger damage prediction model to determine the damage probability of each part of the passenger located on the target seat as the damage probability result.

7. The occupant protection method according to claim 6, characterized by, Before the road traffic environment information, out-of-vehicle information and in-vehicle information are input into the pre-trained passenger damage prediction model to determine the damage probability of each part of the passenger located on the target seat as the damage probability result, the method further includes: The obtained first input data set and the real damage probability of each part of the passenger corresponding to the first input data set are used as first training samples, and each first input data set includes a set of out-of-vehicle information samples, in-vehicle information samples and road traffic environment samples; The first machine learning model is trained through the first training samples until the first machine learning model meets a first preset convergence condition, and the passenger damage prediction model is obtained.

8. The occupant protection method according to claim 3, characterized by, The target restraint parameter of the passenger restraint system matched with the target seat according to the in-vehicle information, the seat target parameter, the road traffic environment information and the out-of-vehicle information includes: The passenger physical sign of the passenger located on the target seat is obtained according to the in-vehicle information; The collision condition information is determined according to the road traffic environment information and the out-of-vehicle information; The passenger physical sign, the seat target parameter and the collision condition information are input into a pre-trained passenger restraint parameter optimization model to determine the target restraint parameter.

9. The occupant protection method according to claim 8, characterized by, The collision condition information includes at least one of collision type, speed, relative lateral displacement and angle; And / or, the target restraint parameter includes at least one of airbag ignition time, seat belt pretensioning time, seat belt limiting value, seat belt waist pretensioning time.

10. The occupant protection method according to claim 8, characterized by, Before the passenger physical sign, the seat target parameter and the collision condition information are input into the pre-trained passenger restraint parameter optimization model to determine the target restraint parameter, the method further includes: The obtained second input data set and the real restraint parameter of the passenger restraint system corresponding to the second input data set are used as second training samples, and each second input data set includes a set of passenger physical sign training data, seat parameter training data and collision condition training data; Based on the second training samples, a second machine learning model is optimized and trained through a preset optimization algorithm until the second machine learning model meets a second preset convergence condition, and the passenger restraint parameter optimization model is obtained.

11. The occupant protection method according to claim 8, characterized by, The controlling the occupant restraint system to perform occupant restraint according to the target constraint parameter includes: The occupant injury probability prediction is performed based on the collision signal and the collision condition information through an occupant injury prediction model, and the prediction probability result is obtained, wherein the occupant injury prediction model is obtained by training traffic accident data samples including data of the target vehicle.

12. An occupant protection device characterized by comprising: The method comprises: The driving safety warning control module is configured to determine whether the target vehicle has potential collision danger according to the obtained road traffic environment information. The sitting posture adjustment control module is configured to perform injury probability analysis on an occupant on a target seat in the target vehicle according to the obtained road traffic environment information, off-vehicle information and in-vehicle information of the target vehicle when the target vehicle has potential collision danger, and obtain an injury probability result; and further configured to adjust a seat position parameter of the target seat when the injury probability result is greater than a first preset probability until the injury probability result is less than a second preset probability, wherein the first preset probability is greater than or equal to the second preset probability, and the seat position parameter includes a seat backrest angle, a seat cushion angle and a foot support angle. The adjusting the seat position parameter of the target seat when the injury probability result is greater than the first preset probability until the injury probability result is less than the second preset probability includes: when the injury probability of at least one part is greater than the first preset probability, stepwise adjusting at least one of the seat backrest angle, the seat cushion angle and the foot support angle of the target seat; after each adjustment, the injury probability of each part of the occupant on the target seat is re-determined through an occupant injury prediction model, and the seat position parameter is iteratively adjusted based on the re-determined injury probability until the injury probability of each part is less than the second preset probability, and a target seat parameter is obtained.

13. A vehicle characterized by comprising: The memory and the processor are included; The memory is configured to store a computer program; The processor is configured to implement the occupant protection method according to any one of claims 1-11 when executing the computer program.

14. A computer-readable storage medium, characterized in that, The storage medium has the computer program stored thereon, and the computer program, when executed by a processor, implements the occupant protection method according to any one of claims 1-11.

Images