Vehicle collision adaptive protection method and apparatus therefor

Abstract

The application provides a vehicle collision adaptive protection method and device, the method comprises the following steps: obtaining a predicted trajectory of a target around the vehicle; performing collision risk assessment according to the motion information of the vehicle and the predicted trajectory of the target to obtain a collision risk parameter; when the collision risk parameter meets a preset triggering condition, determining a collision type and obtaining a seat belt state of an occupant; determining an inflation strategy according to the collision type and the seat belt state; wherein the inflation strategy is used to indicate a priority inflation module in a plurality of protection pad modules arranged at different positions in the vehicle, an inflation sequence of each priority inflation module and a target internal pressure; controlling the inflation of each priority inflation module according to the inflation strategy to make each priority inflation module complete deployment before the collision occurs. The protection pad can complete deployment before the collision occurs, and the deployment position, inflation sequence and target internal pressure of the protection pad are matched with the collision type and the seat belt state of the occupant, thereby improving the protection effect on the occupant in the vehicle collision.

Description

Technical Field

[0001] This application relates to the field of vehicle safety technology, specifically to a vehicle collision adaptive protection method and device. Background Technology

[0002] During operation, vehicles may collide with other vehicles, pedestrians, cyclists, or road infrastructure. Occupant protection in a collision primarily relies on passive safety systems, such as seat belts and airbags. Airbag systems are typically activated by an airbag controller based on collision signals detected by acceleration sensors after a collision. The airbags inflate and deploy within tens of milliseconds following the impact to cushion and protect the occupants.

[0003] Because airbags only deploy after a collision, there is a time lag between ignition and inflation. During this time, collision energy begins to transfer to the passenger compartment, limiting the preparation time for the airbag to establish cushioning. Furthermore, airbag installation locations are typically fixed, and their deployment and inflation pressure are executed according to pre-set parameters. This makes it difficult to adjust for different collision types (frontal, side, rear-end, rollover) and varying occupant restraints (e.g., whether seatbelts are worn). Consequently, it's challenging to adapt airbags to the specific protection requirements of different collision scenarios and occupant states. Improving occupant protection in vehicle collisions remains a critical issue in vehicle safety. Summary of the Invention

[0004] In view of this, embodiments of this application provide a vehicle collision adaptive protection method and apparatus, which enables the protective pad to be deployed before a collision occurs, and matches the deployment part, inflation sequence and target internal pressure of the protective pad with the collision type and the occupant's seat belt status, so as to improve the protection effect on the occupant in a vehicle collision.

[0005] The first aspect of this application provides a vehicle collision adaptive protection method, including: Obtain the predicted trajectory of targets around the vehicle; Collision risk assessment is performed based on the vehicle's motion information and the target's predicted trajectory to obtain collision risk parameters; wherein, the collision risk parameters include at least one of collision time, expected collision overlap rate, and estimated collision energy. When the collision risk parameters meet the preset triggering conditions, the collision type is determined and the occupant's seat belt status is obtained; An inflation strategy is determined based on the collision type and the seatbelt status; wherein, the inflation strategy is used to indicate the priority inflation modules among multiple protective pad modules installed in different parts of the vehicle, the inflation sequence of each priority inflation module, and the target internal pressure; The inflation strategy is used to control the inflation of each priority inflation module so that each priority inflation module can be deployed before a collision occurs.

[0006] A second aspect of this application provides a vehicle collision adaptive protection device, including an environmental sensing component, a main control unit, multiple protective pad modules disposed in different parts of the vehicle, and a gas generator for supplying air to each of the protective pad modules. The environmental perception component is used to collect environmental perception data around the vehicle. The main control unit is connected to the environmental perception component and each of the gas generators, and is used to obtain the predicted trajectory of the target around the vehicle based on the environmental perception data, perform a collision risk assessment based on the vehicle's motion information and the predicted trajectory of the target to obtain collision risk parameters, determine the collision type when the collision risk parameters meet the preset trigger conditions, obtain the occupant's seat belt status, determine the inflation strategy based on the collision type and the seat belt status, and control the corresponding gas generator according to the inflation strategy so that the priority inflation module indicated by the inflation strategy completes deployment before the collision occurs. The collision risk parameters include at least one of collision time, expected collision overlap rate, and collision energy estimate; the inflation strategy is used to indicate the priority inflation modules among the plurality of protective pad modules, the inflation sequence of each priority inflation module, and the target internal pressure.

[0007] The first aspect of this application embodiment involves acquiring the predicted trajectory of targets around the vehicle; performing a collision risk assessment based on the vehicle's motion information and the predicted trajectory of the targets to obtain collision risk parameters; determining the collision type and acquiring the occupant's seatbelt status when the collision risk parameters meet preset triggering conditions; determining an inflation strategy based on the collision type and seatbelt status; controlling the inflation of each priority inflation module according to the inflation strategy, so that each priority inflation module completes deployment before the collision occurs, enabling the identification and quantification of collision risks before the collision occurs, ensuring that the protective pads installed in different parts of the vehicle are at the designed pressure and design form before the collision energy is transmitted to the cabin, and matching the deployment position and inflation sequence of the protective pads with the target internal pressure, collision type, and occupant's seatbelt status, thereby improving the vehicle's protection effect on occupants under different collision types and different occupant restraint states.

[0008] It is understandable that the beneficial effects of the second aspect mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart illustrating the vehicle collision adaptive protection method provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the collision risk parameter determination method provided in an embodiment of the present invention; Figure 3 This is a top view of the vehicle collision adaptive protection device provided in an embodiment of the present invention; Figure 4 This is a side view of the vehicle collision adaptive protection device provided in this embodiment of the invention, viewed from the driver's side. Detailed Implementation

[0011] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0012] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0013] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0014] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0015] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0016] The vehicle collision adaptive protection method provided in this invention can be applied to various types of passenger or commercial vehicles, including gasoline vehicles, hybrid vehicles, and pure electric vehicles. It is particularly suitable for vehicles with advanced driver assistance or autonomous driving functions, and can be executed by the vehicle's main control unit when running a computer program with corresponding functions. For consistency throughout the description, this invention defines an absolute left-right reference frame, with the vehicle's direction of travel as forward, the driver's side as left, and the passenger side as right. This invention does not impose any restrictions on the specific type of vehicle.

[0017] like Figure 1 As shown, the vehicle collision adaptive protection method provided in this embodiment of the invention includes the following steps S101 to S105 executed by the vehicle's main control unit: Step S101: Obtain the predicted trajectory of targets around the vehicle.

[0018] In this application, targets around the vehicle include other vehicles, pedestrians, cyclists, and other traffic participants who may collide with the vehicle. The predicted trajectory of the target is used to characterize the target's movement trend over a future period. When performing collision protection for the vehicle, multiple onboard sensors continuously collect environmental perception data around the vehicle, and track and predict the trajectories of each target based on this data to obtain the predicted trajectory of each target. For example, the multiple sensors may include millimeter-wave radar, lidar, wide-angle cameras, ultrasonic sensors, and inertial measurement units, etc. The specific configuration, data specifications, and connection relationships between each sensor and the main control unit will be described in detail in the device embodiments below. During vehicle operation, the main control unit periodically executes steps S101 to S105 to continuously monitor the collision risk in real time.

[0019] Step S102: Based on the vehicle's motion information and the target's predicted trajectory, a collision risk assessment is performed to obtain collision risk parameters; wherein, the collision risk parameters include at least one of collision time, expected collision overlap rate, and estimated collision energy.

[0020] In application, vehicle motion information represents the vehicle's own motion state, including the vehicle's planned trajectory, speed, wheel speed, and longitudinal, lateral, and vertical accelerations and yaw rates monitored in real time by the inertial measurement unit. Based on the vehicle's planned trajectory and the target's predicted trajectory, the predicted collision time and predicted collision point location where the two overlap in space can be determined, and three types of collision risk parameters—collision time, predicted collision overlap rate, and collision energy estimate—can be calculated in real time and in parallel. Among them, collision time represents the urgency of the collision from a temporal dimension, predicted collision overlap rate represents the coverage of the collision from a geometric dimension, and collision energy estimate represents the severity of the collision from an energy dimension. The specific calculation methods for these three parameters will be explained in subsequent embodiments.

[0021] Step S103: When the collision risk parameters meet the preset triggering conditions, determine the collision type and obtain the occupant's seat belt status.

[0022] In application, preset trigger conditions are used to define when the protection system intervenes. When the collision risk parameters meet the preset trigger conditions, it indicates that a collision is unavoidable and active protection needs to be implemented immediately before the collision occurs. At this time, the main control unit determines the collision type of the impending collision based on the motion relationship between the vehicle and the target, and reads the seat belt status of each occupant in the vehicle. It should be noted that since the seat belt status of each occupant may be different, the seat belt status is obtained one by one for each seat, and the subsequent inflation strategy is also determined and executed independently for the occupant corresponding to the seat, to ensure that each occupant receives protection that matches their restraint status.

[0023] Step S104: Determine an inflation strategy based on the collision type and the seat belt status; wherein the inflation strategy is used to indicate the priority inflation modules among multiple protective pad modules installed in different parts of the vehicle, the inflation sequence of each priority inflation module, and the target internal pressure.

[0024] In application, multiple protective pad modules are pre-installed in different parts of the vehicle interior. For ease of description, the roof module A, left side airbag module BL, right side airbag module BR, knee airbag module C, and rear head and neck protection module D are used as examples. The structure, installation position, and materials of each module will be described in detail in the device embodiments below. The main control unit pre-stores an inflation strategy logic table. Using the collision type and the occupant's seat belt status as lookup inputs, the priority inflation modules involved in the current protection, the inflation order of each priority inflation module, and their respective target internal pressure can be determined. An example of an inflation strategy logic table is shown in the table below: It should be noted that the target internal pressure values ​​in the table above correspond to the discrete pressure levels of the gas generator. The actual system output will be based on the pressure level closest to the target internal pressure. For example, a target internal pressure of 50 kPa corresponds to a high pressure of 55-60 kPa, and a target internal pressure of 40 kPa corresponds to a medium-high pressure of 40-45 kPa. When the target internal pressure is equidistant from two adjacent pressure levels, the higher pressure level will be selected to ensure that the actual protection strength is not lower than the strategy requirements. The division of pressure levels and their implementation will be explained in detail in subsequent embodiments.

[0025] Step S105: Control each priority inflation module to inflate according to the inflation strategy so that each priority inflation module can be deployed before the collision occurs.

[0026] In application, when the triggering time is reached, the main control unit immediately sends a protection system trigger command to the gas generator corresponding to each priority inflation module. Each priority inflation module inflates sequentially according to the inflation strategy and the target internal pressure, and completes deployment 50-150ms before the collision. Because the gas generator can inflate the protective pad to the target pressure within milliseconds, and the triggering time allows for hundreds of milliseconds of lead time relative to the moment of collision, the protective pad can reach the design pressure and design configuration before the collision energy is transmitted to the cockpit.

[0027] In this embodiment, by acquiring the predicted trajectory of the target and combining it with the vehicle's motion information to assess collision risk, the protection system can perceive and quantify the collision risk before the actual collision occurs. This changes the traditional passive safety system's operation mode, which only triggers based on collision sensor signals after a collision, thus gaining intervention time before a collision. By determining the inflation strategy based on two dimensions—collision type and seatbelt status—when the triggering conditions are met, the deployment location, deployment sequence, and target internal pressure of the protective pad are matched with the specific collision conditions and occupant restraint status. The principle is that the occupant's motion trend and vulnerable parts differ under different collision types, and unbelted occupants experience greater displacement during a collision, requiring higher restraint strength and the linkage of more protective pad modules. This overcomes the limitations of traditional airbag systems that do not distinguish between collision conditions and occupant restraint status. This overcomes the shortcomings of traditional airbags, which are primarily designed for frontal collisions and fixed deployment. By selectively deploying multiple protective cushion modules located in different parts of the vehicle according to an inflation strategy, targeted buffers can be formed in scenarios such as frontal collisions, side collisions, rear-end collisions, rollovers, and multi-angle collisions. This compensates for the deficiencies of traditional airbags, which are mainly designed for frontal collisions and are insufficient in protecting against lateral head movement and roof intrusion. By ensuring that each priority inflation module is deployed before the collision, the protective cushions are already at the designed pressure and design shape when the collision energy is transmitted to the cabin. Their buffering range and energy absorption effect are superior to traditional airbags that only begin to deploy after the collision. These features work together to expand occupant protection from passive buffering after a collision to active protection before a collision and adaptive adjustment during the collision process, improving the occupant protection effect in various collision scenarios.

[0028] In one embodiment, step S101 includes: Spatiotemporal alignment and fusion tracking are performed on environmental perception data collected by multiple sensors to obtain the state vector of the target; wherein, the state vector includes the target's position, velocity, heading angle and bounding box size; The historical state sequence of the target's state vectors over the past first preset time period is input into the trajectory prediction model to obtain the predicted position, predicted heading angle, and predicted bounding box size of the target at each predicted time in the future second preset time period, which are used as the predicted trajectory.

[0029] In applications, since the sampling times and installation locations of multiple sensors differ, it is necessary to first perform spatiotemporal alignment on the environmental perception data collected by each sensor. Specifically, hardware timestamps and software interpolation can be used to unify all sensor data into a vehicle center coordinate system with the vehicle center as the origin, achieving a time synchronization accuracy of less than 10ms. After completing spatiotemporal alignment, target fusion and tracking are performed. Specifically, the clustering results of the LiDAR point cloud can be used as a benchmark, associated with the visual recognition boxes output by the camera and the radar tracking points output by the millimeter-wave radar. The Joint Probabilistic Data Association (JPDA) algorithm and Extended Kalman Filter (EKF) are used to output a state vector list for each tracked target, including position, velocity, heading angle, and bounding box size. The state vector can also further include the target's acceleration. The update frequency of the state vector list is 100Hz, and the bounding box size includes the length, width, and height of the target bounding box.

[0030] In application, the trajectory prediction model can employ a lightweight long short-term memory (LSTM) network prediction model deployed in the main control unit. This model takes a historical state sequence of 10 frames from the past second as input. Each frame in the historical state sequence contains the target's position, velocity, heading angle, and bounding box size. The model outputs the predicted position, predicted heading angle, and predicted bounding box size of the target every 0.1 seconds within the next second. That is, both the first and second preset durations are 1 second, and the interval between each prediction moment is 0.1 seconds. This model is pre-trained and validated on simulation datasets and real vehicle datasets containing urban, highway, and intersection scenarios to ensure prediction accuracy under different traffic scenarios.

[0031] In this embodiment, spatiotemporal alignment unifies heterogeneous and asynchronous sensor data to the same spatiotemporal reference, eliminating temporal and spatial deviations between multi-source data and providing consistent input for subsequent fusion. By associating visual recognition boxes and radar tracking points based on lidar point cloud clustering results, the complementary characteristics of lidar's high ranging accuracy, camera's strong target classification capability, and millimeter-wave radar's stable velocity measurement and adaptability to adverse weather conditions are utilized. A joint probabilistic data association algorithm is used to resolve data association ambiguities when multiple targets cross-occlusion, and extended Kalman filtering is used to suppress observation noise, ensuring that the state vector has both integrity and stability. By inputting historical state sequences into a long short-term memory network for trajectory prediction, its ability to model temporal dependencies is utilized, and the predicted position, predicted heading angle, and predicted bounding box size are output simultaneously. This ensures that the predicted trajectory includes not only the target's future position but also the target's future attitude and shape information, providing a complete data foundation for accurate assessment of subsequent collision risks.

[0032] In one embodiment, such as Figure 2 As shown, step S102 includes: Step S201: Determine the predicted collision time and predicted collision speed based on the vehicle's motion information and the target's predicted trajectory; Step S202: Determine the collision time based on the relative distance and radial velocity between the vehicle and the target; Step S203: Based on the predicted position, predicted heading angle and predicted bounding box size corresponding to the predicted collision time in the predicted trajectory, determine the predicted bounding box of the target at the predicted collision time, project the vehicle's bounding box and the predicted bounding box onto the horizontal plane, and use the ratio of the intersection area of ​​the two projections to the area of ​​the vehicle's bounding box as the expected collision overlap rate. Step S204: Determine the estimated collision energy value based on the predicted collision speed and the mass category of the target.

[0033] In application, the predicted collision time and predicted collision speed are determined based on the vehicle's motion information and the target's predicted trajectory. Specifically, the moment when the planned trajectory of the vehicle and the predicted trajectory of the target overlap in space is determined as the predicted collision time, and the relative speed between the vehicle and the target at the predicted collision time is used as the predicted collision speed.

[0034] In one embodiment, the formula for calculating the collision time is: Where TTC represents the time of collision, and D represents the relative distance between the vehicle and the target. This represents the radial velocity between the vehicle and the target, i.e., the component of the relative velocity along the line connecting the two.

[0035] In application, the calculation of the predicted collision overlap rate specifically includes the following three sub-steps: First, the predicted position, predicted heading angle, and predicted bounding box size corresponding to the predicted collision time are extracted from the predicted trajectory to determine the predicted bounding box of the target at the predicted collision time, where the orientation of the predicted bounding box is determined by the corresponding predicted heading angle; then, the vehicle's bounding box and the predicted bounding box are projected onto a horizontal plane with the vehicle's center as the origin, where the size of the vehicle's bounding box is known and fixed; finally, the ratio of the intersection area of ​​the two rectangular projections to the area of ​​the vehicle's bounding box is calculated as the predicted collision overlap rate. Since the trajectory prediction model simultaneously outputs the predicted position, predicted heading angle, and predicted bounding box size at each prediction time, this overlap rate can be calculated accurately. It should be noted that when the predicted collision time does not coincide with any prediction time but is still within the prediction time domain, the corresponding predicted position, predicted heading angle, and predicted bounding box size can be obtained by interpolation of the prediction data of two adjacent prediction times. When the predicted collision time exceeds the prediction time domain of the trajectory prediction model, it is extrapolated based on the predicted state at the end of the prediction time domain according to the assumption of uniform motion, thus ensuring that the expected collision overlap rate can also be calculated within a longer warning time interval.

[0036] In one embodiment, the formula for calculating the expected collision overlap rate is: Wherein, POR represents the expected collision overlap rate. This represents the intersection area of ​​the predicted bounding box of the vehicle and the predicted bounding box of the target on the horizontal plane at the predicted collision moment. This represents the projected area of ​​the vehicle's bounding box on the horizontal plane.

[0037] In application, the collision energy estimate is obtained based on the predicted collision speed and the target's mass category. The target's mass category is mapped from the target's classification results. For example, pedestrians, cyclists, passenger cars, and commercial vehicles correspond to different typical mass categories, thus enabling rapid estimation of the collision kinetic energy level without relying on the actual mass measurement of the target.

[0038] In this embodiment, by pre-determining the predicted collision time and predicted collision velocity, a unified time reference is provided for the calculation of each collision risk parameter. This ensures that the collision time, predicted collision overlap rate, and collision energy estimate are calculated for the same predicted collision event, guaranteeing consistency among the parameters. By characterizing the risk in parallel from three dimensions—time, geometry, and energy—the collision time, predicted collision overlap rate, and collision energy estimate, respectively, any noise or inaccuracy of any single parameter can be cross-validated by the other parameters, reducing the probability of false triggering and missed triggering. By extrapolating based on the assumption of uniform motion when the predicted collision time exceeds the prediction time domain, the continuous availability of risk parameters over a longer warning time interval is guaranteed. By estimating energy based on mass category rather than actual mass, the direct measurement of the target mass is avoided while ensuring accuracy in the magnitude of the energy, reducing computational complexity.

[0039] In one embodiment, the preset triggering conditions include warning conditions and execution conditions; Step S103 includes: When the collision risk parameters meet the execution conditions, the collision type is determined and the occupant's seatbelt status is obtained; The warning conditions include the collision time being less than a first time threshold or the predicted collision overlap rate being greater than a first overlap rate threshold; the execution conditions include any one of the following: the collision time being less than a second time threshold and lateral avoidance is not feasible; the predicted collision overlap rate being greater than a second overlap rate threshold and the estimated collision energy being greater than a preset energy threshold; the lateral acceleration of the vehicle being greater than a preset lateral acceleration threshold and continuously increasing; the second time threshold being less than the first time threshold and the second overlap rate threshold being greater than the first overlap rate threshold.

[0040] In application, the thresholds can be set as follows: the first time threshold is 2.0s, the first overlap threshold is 30%, the second time threshold is 0.3s, the second overlap threshold is 60%, the preset energy threshold is 15kJ, and the preset lateral acceleration threshold is 0.7g. When the collision time is less than 2.0s or the expected collision overlap rate is greater than 30%, the warning conditions are met, and the system enters the first-level warning stage. When the collision time is less than 0.3s and the vehicle stability control system reports that lateral avoidance is no longer feasible, or when the expected collision overlap rate is greater than 60% and the estimated collision energy is greater than 15kJ, or when the inertial measurement unit detects that the vehicle's lateral acceleration is greater than 0.7g and continues to rise, indicating that a rollover is imminent, the execution conditions are met, and the system enters the second-level execution stage, which involves determining the collision type, determining the inflation strategy, and controlling the inflation deployment. It should be noted that for rollover conditions, the execution conditions are the same as the judgment conditions for rollover types. That is, once a rollover is identified, the execution conditions are directly met, and the system skips the first-level warning stage and directly enters the second-level execution stage to ensure timely response to accidents such as rollovers that develop extremely quickly.

[0041] In this embodiment, by decomposing the preset triggering conditions into two levels—early warning conditions and execution conditions—early warning and accurate execution are decoupled. The early warning conditions are relatively lenient to ensure sufficient lead time, while the execution conditions are relatively stringent to avoid irreversible false triggering. Through the AND logic of collision time and lateral avoidance feasibility, and the AND logic of overlap rate and energy estimation in the execution conditions, exceeding the limit of any single parameter is insufficient to trigger the event, thus suppressing the risk of false detonation caused by sensor noise or instantaneous occlusion. By using the lateral acceleration condition as an independent execution channel, rollover accidents without external collision targets are covered. By limiting the second time threshold to be less than the first time threshold and the second overlap rate threshold to be greater than the first overlap rate threshold, the two levels of conditions are ensured to be logically strictly progressive, and the early warning must occur before the execution.

[0042] In one embodiment, step S103 includes: When the collision risk parameters meet the preset triggering conditions, the impending collision is determined to be one of the following based on the relative motion direction between the vehicle and the target, the predicted collision point position between the vehicle and the target, and the classification result of the target: frontal collision, side collision, rear-end collision, rollover, and multi-angle collision. The occupant's seatbelt status is obtained by reading the detection signal output by the sensor integrated in the seatbelt buckle; The frontal collision, the side collision, and the rear-end collision are determined based on a preset angle range of the angle between the relative velocity vector between the vehicle and the target and the vehicle's direction of travel. The side collision is distinguished between left-side collision and right-side collision. The rollover is determined based on at least one of the vehicle's lateral acceleration and wheel speed difference.

[0043] In application, the preset angle range can be set as follows: when the angle between the relative velocity vector and the vehicle's forward direction is less than or equal to 30°, it is determined to be a frontal collision; when the angle is between 60° and 120°, it is determined to be a side collision, and further distinguishes between left-side and right-side collisions based on whether the predicted collision point is located on the left or right side of the vehicle; when the angle is between 150° and 210°, it is determined to be a rear-end collision; when the angle is outside the above ranges, or when multiple predicted collision points are detected simultaneously, it is determined to be a multi-angle collision. For rollover, since it is due to the vehicle's own attitude instability rather than a direct impact with the target, the vehicle's own dynamics are used for determination, specifically by detecting a lateral acceleration greater than 0.7g and continuously increasing through the inertial measurement unit, or by judging through abnormal changes in the speed difference between the left and right wheels. The collision type determination result serves as one of the lookup inputs for the inflation strategy logic table.

[0044] In the application, each seat belt buckle integrates a Hall sensor. When the latch is inserted into the buckle, the magnetic circuit state changes, and the Hall sensor outputs a high-level or low-level signal to the body controller. The main control unit reads the signal in real time through the CAN bus, thereby obtaining the seat belt status of the occupant in the corresponding seat as "fastened" or "unfastened".

[0045] In this embodiment, the collision type is determined by the preset angle range of the relative velocity vector angle, which requires only simple geometric calculations, has low computational load, and strong real-time performance, providing results in a very short time after the triggering conditions are met. By using the relative motion direction, predicted collision point position, and target classification results as the determination criteria, multi-source information is mutually verified, improving the robustness of the type determination under complex working conditions. Rollover is determined by lateral acceleration and wheel speed difference, covering the condition where the vehicle's own attitude is unstable without external impact targets. The seat belt status is obtained by reading the detection signal of the sensor integrated in the seat belt buckle, directly detecting the physical connection status of the latch and buckle, with short signal links and high reliability. The collision type and seat belt status together constitute the complete lookup input of the inflation strategy logic table, providing a basis for subsequently determining the inflation strategy differently according to the collision conditions and occupant status.

[0046] In one embodiment, the method further includes: When the collision risk parameters meet the warning conditions, at least one of the following shall be executed: control the gas generator of each of the protective pad modules to enter standby state, pretension the seat belt, issue a warning to the driver, and broadcast a collision warning signal to the vehicle network; The collision warning signal is used to enable the vehicle to perform at least one of path fine-tuning and steering compensation, and to adjust the seat to a protective posture; the vehicle's original airbag controller deploys the original airbags after a collision based on the triggering logic of the acceleration sensor, forming a time difference dual protection with each of the priority inflation modules that have completed deployment before the collision.

[0047] In application, the specific actions of the first-level warning stage are as follows: the protection system is activated, the gas generators of each protective pad module enter the standby state, that is, to charge the energy storage capacitor in the ignition circuit; the seat belt motor pretensions the seat belt to eliminate webbing slack; a warning is issued to the driver through the head-up display or seat vibration, prompting the driver to take active intervention; the main control unit broadcasts the collision warning signal to the vehicle network through the vehicle Ethernet, which can adopt the 100BASE-T1 standard.

[0048] In application, after receiving a collision warning signal, the chassis domain controller commands the vehicle stability control system to perform a final optimal path fine-tuning, such as applying single-sided braking, and commands the electric power steering system to provide a steering compensation torque of 0.5-1.0 Nm, so as to improve the collision posture and reduce the severity of the collision as much as possible under the premise that a collision is unavoidable; after receiving a collision warning signal, the body domain controller commands the electric seat to adjust to the optimal protective posture within 300ms, such as adjusting the seat back to 80° and raising the headrest to the highest position, so that the occupant's body returns to the design protection condition of the protective pad and seat belt.

[0049] In application, upon receiving the trigger command, the system's priority inflation modules deploy 50-150ms before the collision. Simultaneously, the vehicle's original traditional airbag controller, while also receiving the collision signal, maintains its acceleration sensor-based triggering logic and typically deploys the existing airbags 10-30ms after the collision. Thus, the protective pads deployed before the collision cover energy absorption before and in the initial stages of the collision, while the original airbags deployed after the collision cover energy absorption during the peak of the collision. These two deployments are sequential on the timeline, forming a time-difference dual protection.

[0050] In this embodiment, by putting the gas generator into standby mode during the warning phase, the ignition energy storage capacitor is pre-charged, shortening the ignition delay during the actual triggering; by pre-tensioning the seat belts during the warning phase, the occupant is restrained in the optimal position on the seat before the collision, coordinating with the subsequent deployment of the protective pad; by issuing a warning to the driver, the opportunity for active intervention is preserved; by broadcasting the collision warning signal to the entire vehicle network, the chassis, body, and restraint system enter the preparation state in parallel, with the preparation actions of each system overlapping in time rather than waiting sequentially, making full use of the limited time during the warning phase; through path fine-tuning and steering compensation, in Even when a collision is unavoidable, the system proactively optimizes the collision angle and contact position to reduce the collision energy transmitted into the cabin from the source. By adjusting the seats to a protective posture, it corrects the deviation from the design conditions of the protection system caused by factors such as reclining seats and low headrests, allowing the deployment shape of the protective pads to be rematched with the occupant's body position. By retaining the independent triggering logic of the original airbag controller, this system and the original airbag system are independent of each other and redundant. Even if one fails, the other will still operate according to its own logic. Through the time difference between the protective pads that deploy before the collision and the original airbags that deploy after the collision, the occupant experiences a two-stage buffering process, reducing the peak collision acceleration experienced by the occupant.

[0051] In one embodiment, step S104 includes: The inflation strategy is obtained by querying a preset inflation strategy logic table based on the collision type and the seat belt status. The seatbelt status includes a fastened state and an unfastened state; the inflation strategy logic table records the priority inflation modules, inflation order, and target internal pressure corresponding to different combinations of the collision type and the seatbelt status; for the same collision type that distinguishes between the fastened state and the unfastened state in the inflation strategy logic table, the number of priority inflation modules corresponding to the unfastened state is greater than the number of priority inflation modules corresponding to the fastened state, and the maximum value of each target internal pressure corresponding to the unfastened state is higher than the maximum value of each target internal pressure corresponding to the fastened state.

[0052] In application, the inflation strategy logic table is pre-stored in the main control unit, using the combination of collision type and seat belt status as the lookup index. Each row records the priority inflation module, inflation sequence, and target internal pressure corresponding to that combination. Specific values ​​can be found in the inflation strategy logic table described earlier. The lookup operation is performed after the execution conditions are met, resulting in minimal computational overhead and predictable and reproducible results. The values ​​in each row of the logic table can be optimized item by item through bench testing, simulation analysis, and real-vehicle calibration, and adapted to differences in vehicle model and cabin layout. As mentioned earlier, the seat belt status of each occupant is read independently, and the inflation strategy is determined by looking up the table separately for each occupant, without affecting each other.

[0053] In application, taking a frontal collision as an example, the priority inflation modules for the fastened state are the knee airbag module and the roof module, with the knee airbag module inflating first, and the target internal pressures being 40 kPa and 25 kPa respectively. For the unfastened state, the priority inflation modules increase to the roof module, knee airbag module, and rear head and neck protection module, with the roof module inflating first, and the target internal pressure for the roof module increasing to 50 kPa. This is because unfastened occupants lack webbing restraint, resulting in greater body displacement and more unpredictable movement direction during a collision, and more dispersed potential contact points with the vehicle's interior structure. Therefore, higher cushioning stiffness and a larger coverage area are required to support the occupant. For rollover scenarios, since the risk of occupant ejection exists for both fastened and unfastened occupants, the requirement for top cage protection remains the same, and the inflation strategy logic table uses the same strategy for both fastened and unfastened states.

[0054] In this embodiment, the inflation strategy is determined by querying a preset inflation strategy logic table. The computational overhead of the table lookup is minimal and the result is certain, meeting the real-time requirements within a very short time window before the collision. The values ​​of each row in the logic table are easy to verify and trace through testing and calibration. By configuring more priority inflation modules and higher maximum target internal pressure for the unbelted state, the larger displacement and more dispersed contact positions of the occupant when the seat belt restraint is missing are compensated, so that the protection strength matches the actual restraint state of the occupant and improves the protection effect for unbelted occupants.

[0055] In one embodiment, each of the protective pad modules is configured with a multi-stage ignition gas generator, which outputs multiple discrete pressure levels through a combination of different ignition stages. Step S105 includes: According to the inflation sequence, the gas generator of each priority inflation module is controlled to inflate at the pressure level closest to the target internal pressure among the plurality of discrete pressure levels, so that each priority inflation module can complete its deployment before the collision occurs.

[0056] In application, the gas generator can be a three-stage pyrotechnic generator. The three ignition units share the same combustion chamber, and the outlet of the combustion chamber is sealed to the corresponding protective gasket module's gas chamber. The ignition sequence of the three-stage ignition units is as follows: the first stage ignites within 3ms, and within 15ms, the protective gasket is filled to its initial state with an internal pressure of approximately 10 kPa; the second stage selectively ignites within 5-20ms, raising the internal pressure to 25-40 kPa; the third stage selectively ignites within 10-30ms, raising the internal pressure to 45-60 kPa. Through different ignition combinations—i.e., only the first stage, the first and second stages, the first and third stages, and the first, second, and third stages—four discrete pressure outputs can be achieved: low (10 kPa), low-medium (25-30 kPa), medium-high (40-45 kPa), and high (55-60 kPa). The final target pressure is determined by the target internal pressure in the inflation strategy logic table. The system selects the closest setting based on the values ​​in the table. When the target internal pressure is equidistant from two adjacent settings, the higher setting is selected uniformly, eliminating the need for continuous pressure adjustment. As an alternative, a combination of a single-stage generator and a pressure regulating valve can be used to achieve continuous pressure output. However, the three-stage discrete setting scheme has lower structural and control complexity, and this embodiment prioritizes the three-stage discrete setting scheme.

[0057] In this embodiment, a multi-stage ignition gas generator outputs discrete pressure levels with different ignition stage combinations, achieving variable control of the internal pressure of the protective pad with a pure pyrotechnic structure. This eliminates the need for continuous adjustment mechanisms such as proportional valves and pressure regulating valves, resulting in fewer moving parts and high reliability, fast response, and low cost under high-temperature, high-pressure transient conditions. The phased sequence of first-stage ignition to inflate the protective pad to its initial shape followed by selective pressurization in subsequent stages allows the protective pad to first be flexibly formed and then its rigidity increased as needed, avoiding secondary injuries to occupants caused by the instantaneous impact of high-pressure gas. By executing the pressure level closest to the target internal pressure and selecting a higher level at equidistant intervals, the actual output protection strength is ensured to always be no less than the strategic requirements. By sequentially triggering each priority inflation module according to the inflation sequence, the ignition driving capability is prioritized for the most critical protective parts of the current collision type, ensuring that the protective pads in critical areas are deployed first.

[0058] In one embodiment, the method further includes a seat posture adaptive mapping step; The seat posture adaptive mapping step includes: when the seat position changes, calculating the vector direction of each of the protective pad modules relative to the occupant's key anatomical parts, and adjusting the nozzle angle and deployment sequence of the gas generator supplying air to each of the protective pad modules according to the vector direction; wherein, the occupant's key anatomical parts include at least one of the head center of gravity, the sternal center, and the hip joint.

[0059] In this application, the system incorporates a 3D model of the seat. When any of the seat's slide rail position, height, tilt angle, or rotation angle changes, the seat posture adaptive mapping algorithm calculates in real time, based on this 3D model, the vector direction of each protective pad module relative to key anatomical locations such as the occupant's head center of gravity, sternal center, and hip joint. Based on this, it dynamically adjusts the nozzle angle and deployment sequence of the gas generator supplying air to each protective pad module, ensuring that the deployment direction and rhythm of the protective pads always face and conform to the occupant's actual body position. This step is particularly suitable for cockpit layouts in autonomous driving scenarios where the seat can rotate and the occupant's posture is flexible.

[0060] In this embodiment, by calculating the vector direction based on the key anatomical parts of the occupant, the change in seat position is transformed into a quantitative adjustment of the nozzle angle and deployment sequence. This allows the deployment shape of the protective pad to adapt to the seat posture in real time, solving the problem of reduced protection effect of fixed safety systems in free postures such as seat rotation and reclining in autonomous driving cockpits, and even causing injury to the occupant due to mismatch between the deployment direction and the occupant's body position. Using key anatomical points such as the head center of gravity, the center of the sternum, and the hip joint as mapping references, the deployment energy of the protective pad is concentrated on the parts of the human body that need the most protection and have the weakest tolerance, improving the targeted nature of the protection.

[0061] In one embodiment, the method further includes an occupant type identification step; The occupant type identification step includes: identifying the occupant type based on seat pressure distribution data and in-vehicle image data; when a forward-facing child safety seat is identified, reducing the target internal pressure of the protective pad module located in front of and above the forward-facing child safety seat or removing it from the priority inflation module, and activating the protective pad module located on the side for wrap-around protection.

[0062] In application, occupant type recognition is achieved through a combination of seat pressure distribution data collected by seat pressure distribution sensors and in-vehicle image data collected by in-vehicle infrared cameras. Based on this, the system determines whether the occupant in the seat is an adult, a child, or a child safety seat. Children can be identified based on height and profile estimation. If a forward-facing child safety seat is identified, the system reduces the target internal pressure of the knee airbag module in front of the child safety seat and the roof module above it, or excludes them from the priority inflation modules, which corresponds to disabling their high-pressure inflation mode. The system then activates the left and right side airbag modules to provide wrap-around protection for the child safety seat, avoiding secondary impacts to the child from the high-pressure inflation from the front and top.

[0063] In this embodiment, dual-mode recognition using seat pressure distribution data and in-vehicle image data allows pressure features and visual contour features to corroborate each other, reducing the probability of misjudging occupant type. By reducing the target internal pressure of the corresponding part or excluding it from the priority inflation module when a forward-facing child safety seat is identified, and activating the lateral protective pad module for wrap-around protection, secondary injury to children caused by inflation intensity designed for adult body size is avoided, making the protection method compatible with occupant type.

[0064] This invention also provides a vehicle collision adaptive protection device for performing the steps described in the vehicle collision adaptive protection method embodiments above. This device can be integrated into the vehicle as a standalone safety module assembly, or it can share some hardware with other vehicle control systems.

[0065] The vehicle collision adaptive protection device provided in this embodiment of the invention includes an environmental sensing component, a main control unit, multiple protective pad modules disposed in different parts of the vehicle, and a gas generator for supplying air to each of the protective pad modules. The environmental perception component is used to collect environmental perception data around the vehicle. The main control unit is connected to the environmental perception component and each of the gas generators, and is used to obtain the predicted trajectory of the target around the vehicle based on the environmental perception data, perform a collision risk assessment based on the vehicle's motion information and the predicted trajectory of the target to obtain collision risk parameters, determine the collision type when the collision risk parameters meet the preset trigger conditions, obtain the occupant's seat belt status, determine the inflation strategy based on the collision type and the seat belt status, and control the corresponding gas generator according to the inflation strategy so that the priority inflation module indicated by the inflation strategy completes deployment before the collision occurs. The collision risk parameters include at least one of collision time, expected collision overlap rate, and collision energy estimate; the inflation strategy is used to indicate the priority inflation modules among the plurality of protective pad modules, the inflation sequence of each priority inflation module, and the target internal pressure.

[0066] In application, the environmental perception component includes various types of sensors distributed across the vehicle body. Examples of sensor hardware configuration and data specifications are as follows: Six millimeter-wave radars are arranged in a configuration of three forward-facing, one rearward-facing, and two lateral-facing sensors, operating at a frequency of 76-81 GHz, with a detection range of 0.2m-250m and a speed resolution of 0.1 km / h, used for accurately calculating the relative speed and relative distance of targets; one forward-facing solid-state lidar with a wavelength of 905nm, a horizontal field of view of 120°, a vertical field of view of 25°, a frame rate of 10Hz, generating a 128-line point cloud for constructing a high-precision 3D drivable area model in front of the vehicle; and five wide-angle cameras are used in a configuration of four surround-view cameras. One front-facing sensor with a resolution of 2 megapixels and a dynamic range of 120dB outputs RGB images and YUV data streams processed by an image signal processor. This is used for classifying targets such as vehicles, pedestrians, and cyclists, as well as for lane line recognition. A total of 12 ultrasonic sensors are used, with a detection frequency of 48kHz and a detection range of 0.15m-4.5m. These are primarily used for parking and detecting obstacles at very close range. The inertial measurement unit incorporates a three-axis accelerometer and a three-axis gyroscope. The accelerometer has a range of ±25g, the gyroscope has a range of ±300° / s, and the sampling frequency is 100Hz. This unit is used to monitor the vehicle's longitudinal, lateral, and vertical accelerations and yaw rate in real time.

[0067] In application, the signal flow and connection relationship inside the device are as follows: the lidar and each camera are connected to the Ethernet port of the main control unit via the vehicle Ethernet to transmit point cloud data and image data. The vehicle Ethernet can adopt the 100BASE-T1 standard; each millimeter-wave radar, each ultrasonic sensor and inertial measurement unit are connected to the CAN transceiver port of the main control unit via the CAN bus to transmit target traces, distance echo and motion status data; the sensors in each seat buckle output detection signals to the body controller, and the main control unit reads the signal from the body controller in real time via the CAN bus; the ignition control output terminal of the main control unit is electrically connected to the ignition terminal of each gas generator via the ignition drive circuit. The ignition drive circuit is equipped with an energy storage capacitor, which is charged during the first-level warning; each gas generator is fixedly installed in the module housing of the corresponding protective pad module, and its gas outlet is sealed and connected to the gas chamber of the protective pad module. The gas generated after ignition is directly filled into the gas chamber through the gas outlet. In each of the above stages, environmental perception data flows from the sensor to the main control unit, and trigger commands flow from the main control unit to the ignition drive circuit and the gas generator, forming a complete signal chain from self-sensing and decision-making to execution.

[0068] Figure 3The figure shows a top view of the vehicle collision adaptive protection device provided in an embodiment of the present invention. The background is a top view of the vehicle with all four doors open. The distribution of each protective pad module inside the vehicle is shown: Roof module A is located in the roof area above the front passenger; left side airbag module BL is located in the side area corresponding to the B-pillar of the left door; right side airbag module BR is located in the side area corresponding to the B-pillar of the right door; knee airbag module C is located in the knee area below the dashboard on the driver's side, and in the area on the passenger side where the back of the front seat faces the knees of the rear passenger; rear head and neck protection modules D are shown in both the front and rear seat headrest areas; the marking box in the center of the dashboard indicates the installation area of ​​the main control unit.

[0069] In this embodiment, an environmental perception component, a main control unit, multiple protective pad modules, and a gas generator form a three-layered, sequentially connected link structure for perception, decision-making, and execution. Environmental perception data flows in unidirectionally, and trigger commands flow out unidirectionally, resulting in a clear link and controllable latency. Through the complementary configuration of millimeter-wave radar, lidar, cameras, ultrasonic sensors, and inertial measurement units in terms of detection band, field of view, and measurement range, long-range high-speed targets, short-range low-speed obstacles, and the vehicle's own attitude are all covered, achieving omnidirectional environmental perception without blind spots. The main control unit directly drives the gas generator via an ignition drive circuit, resulting in fewer trigger links and lower ignition latency, ensuring the millisecond-level response required for deployment before a collision. By integrating the inflation strategy into the main control unit, this device can be manufactured and assembled as an independent functional assembly, facilitating its adaptation to different vehicle models.

[0070] In one embodiment, the plurality of protective pad modules include at least two of the following: a canopy module, a left side airbag module, a right side airbag module, a knee airbag module, and a rear head and neck protection module. The roof module is integrated into the roof trim panel to cover the head, neck, and upper chest of the occupants; the left and right side airbag modules are integrated into the corresponding door trim panels or B-pillar trim panels, and their deployment shape is a C-shape with a reserved escape route; the knee airbag module is integrated into the knee pad of the dashboard or the back of the front seat; the rear head and neck protection module is integrated into the active headrest mechanism, which can move forward and upward a preset distance before the rear head and neck protection module inflates and deploys.

[0071] In application, the unfolded area of ​​ceiling module A is approximately 0.8m². 2The inflatable pad is 8-12cm thick. Its base layer is made of 420D high-strength nylon fabric, and the airtight layer is made of TPU film with a light-transmitting area. This area uses optical-grade polyurethane with a light transmittance greater than 90% and a haze of less than 5%, ensuring occupant visibility and psychological safety in non-collision conditions. The roof module A's housing is fixedly connected to the roof interior panel's frame. An internal guide tube is fixedly installed inside the housing. The protective pad is folded and stored inside the housing, with its air inlet sealed to the gas generator's outlet. When unfolded, the protective pad pops out at a 45° angle downwards and forwards along the internal guide tube, covering the occupant's head, neck, and upper chest, preventing collisions with the roof, A-pillars, and other hard structures.

[0072] In application, the deployment area of ​​the left side airbag module BL is approximately 0.4m². 2 The airbag module unfolds in a C-shape, with a third of the lower edge of the door reserved as an escape route. This area, approximately 300mm high, ensures unobstructed access for occupants or rescue personnel after an accident. The protective pad is made of silicone rubber-coated aramid fabric with a tear resistance greater than 5000N. The module housing is fixedly integrated into the left door trim panel or the left B-pillar trim panel. The right side airbag module BR is symmetrical in size, material, and installation method to the left side airbag module BL, and is fixedly integrated into the right door trim panel or the right B-pillar trim panel. The left and right side airbag modules BL and BR respectively buffer collision energy from the corresponding sides, protecting the occupants' ribs and shoulders on the corresponding sides.

[0073] In application, the deployed knee airbag module C covers an area extending from the occupant's knee to their ankle, with a deployed area of ​​approximately 0.25m². 2 The driver-side knee airbag module C is fixedly integrated into the dashboard knee pad, while the passenger-side knee airbag module C is fixedly integrated into the back of the front seats, facing the lower limbs of the front and rear occupants respectively. The rear head and neck protection module D is integrated into the active headrest mechanism, which includes a headrest body and a drive mechanism. The headrest body is movably connected to the seat back frame via a sliding guide rod, and the drive mechanism is fixedly installed on the seat back frame and driven by the sliding guide rod. In the event of a rear-end collision warning, the drive mechanism pushes the headrest body forward and upward by 50mm along the guide direction to reduce the gap between the headrest and the occupant's head. Subsequently, the protective pad stored in the headrest body inflates and deploys to wrap and support the back of the occupant's head.

[0074] Figure 4The figure shows the side view of the vehicle collision adaptive protection device provided in this embodiment of the invention on the driver's side. The background of the figure is the side view of the cabin after the driver's side door is opened. The steering wheel, dashboard, gear shift area, driver's seat, headrest and door frame are visible. The installation positions of each protective pad module on the driver's side are shown: the headliner module A is located in the headliner trim panel area above the driver's seat; the rear head and neck protection module D is located in the headrest of the driver's seat; the knee airbag module C is located in the knee protection panel area of ​​the dashboard below the steering column, facing the driver's knee and lower leg.

[0075] In this embodiment, five types of modules—roof, left and right sides, knees, and rear head and neck—are distributed three-dimensionally above, to the sides, front and lower, and rear of the cabin. Once deployed, each priority inflation module forms a cage-like enclosure for the occupants. Compared to traditional airbags that primarily provide frontal protection, this expands the protective coverage and the range of collision scenarios. The C-shaped deployment of the side airbag modules creates an escape route of approximately 300mm along the lower edge of the doors, providing lateral cushioning while ensuring occupants can escape after an accident and rescue personnel can enter, thus addressing both the needs of protective enclosure and escape / rescue. The active headrest mechanism employs a two-stage action: first moving 50mm forward and upward before inflating, first... Mechanical displacement reduces the gap between the headrest and the head, and then inflation unfolds to achieve flexible wrapping, significantly reducing the risk of whiplash injury to the cervical spine in rear-end collisions; the combination of 420D high-strength nylon fabric and TPU film as the base airtight layer and the selection of silicone rubber coated aramid fabric meet the differentiated requirements of the large-area pressure bearing of the roof module and the tear resistance of the side airbag module; through the optical-grade polyurethane light-transmitting area with a light transmittance of more than 90% and a haze of less than 5%, the roof module does not obstruct the light and vision when hidden and ready, avoiding visual obstruction and a sense of oppression for the occupants; each module is hidden and integrated within the interior panels, trim panels, knee pads or headrests, and does not affect the cabin shape and passenger space at all in non-collision states.

[0076] In one embodiment, the gas generator is a multi-stage ignition gas generator, which outputs multiple discrete pressure levels through a combination of different ignition stages. The main control unit is used to control the multi-stage ignition gas generator corresponding to each priority inflation module to inflate at the pressure level closest to the target internal pressure among the multiple discrete pressure levels, according to the inflation sequence.

[0077] In applications, the number of stages and gear settings of a multi-stage ignition gas generator can be set with reference to the aforementioned method embodiments. For example, a three-stage pyrotechnic gas generator can be used, which outputs four discrete pressure gears—low, medium-low, medium-high, and high—through different combinations of three-stage ignition. The main control unit controls the conduction and ignition timing of the ignition circuit in which each ignition stage is located through the ignition drive circuit.

[0078] In this embodiment, a multi-stage ignition gas generator is used to output the inflation pressure in discrete levels. The target internal pressure is approximated in stages using a simple and fast-responding pyrotechnic ignition scheme. No continuous pressure adjustment mechanism is required. While ensuring millisecond-level deployment speed, the inflation strategy meets the requirements of different target internal pressures in different parts and under different working conditions.

[0079] In one embodiment, the main control unit adopts a dual-core lockstep architecture; The device also includes a backup trigger module independent of the main control unit. The backup trigger module is used to monitor the acceleration signal output by the vehicle's inertial measurement unit, and when the acceleration values ​​at multiple consecutive sampling points exceed a preset trigger acceleration threshold and the main control unit does not receive a trigger signal within a preset response time, the backup trigger module is triggered to deploy through a redundant ignition circuit.

[0080] In applications, the main control unit can employ a dual-core lockstep microprocessor architecture, such as the Infineon TC39xx series. Its two processor cores execute the same code and compare the execution results frame by frame. If the comparison results are inconsistent, a random hardware fault is detected and a reset is triggered, ensuring the correctness of the decision-making calculations from the source. The backup trigger module is an analog circuit module independent of the main control unit. Its electrical connection is as follows: the original acceleration analog signal output of the inertial measurement unit is divided into two paths. One path is connected to the analog-to-digital converter input of the main control unit for normal motion state monitoring, and the other path is connected to the comparator circuit input of the backup trigger module. The trigger output of the backup trigger module is connected to the redundant ignition circuit via an isolated drive circuit. The redundant ignition circuit and the main ignition circuit on the main control unit side are connected in parallel to the ignition terminals of each gas generator. Explosion of any circuit can trigger the deployment of the corresponding protective pad module.

[0081] In application, the backup trigger module uses an independent high-speed sampling circuit to sample the raw acceleration analog signal output by the inertial measurement unit. The sampling rate is, for example, 1kHz. When the acceleration value within three consecutive sampling points, i.e., within a time window of approximately 3ms, reaches the preset trigger acceleration threshold, such as reaching the saturation output of 25g at the upper limit of the accelerometer range, and no trigger signal is received from the main control unit within the subsequent preset response time of 5ms, it is determined that the main control unit may have failed and a collision has indeed occurred. The backup trigger module directly drives the redundant ignition circuit, triggering the deployment of each protective pad module. With the help of the dual-redundant safety architecture composed of the dual-core lockstep and the independent backup trigger channel, and in deep collaboration with the vehicle's existing chassis and body systems, this device can still ensure basic protection functions under extreme fault conditions and can support ASIL-D functional safety level.

[0082] In this embodiment, the dual-core lockstep architecture of the main control unit intercepts random hardware faults within the processor before generating erroneous outputs by comparing the two cores step-by-step, ensuring the correctness of the triggering decision. By setting up an analog backup triggering module that is heterogeneous in principle and independent in power supply and circuitry from the main control unit, there is no common cause failure link between the main and backup triggering channels. Even if the main control unit fails as a whole, the backup channel can still complete the backup triggering based on the original acceleration signal. Through the dual criteria of multiple consecutive sampling points reaching a threshold and no main control triggering signal being received within a preset response time, the backup channel only intervenes when the collision characteristics persist and the main channel does not respond, effectively preventing the backup channel from being falsely triggered. By connecting the redundant ignition circuit in parallel with the main ignition circuit to the ignition end of the gas generator, either channel can independently complete the detonation, and there is no single point of failure in the ignition process. The above features together constitute a dual-redundant safety architecture, enabling the device to meet the active protection function before collision while having the reliability basis to support the ASIL-D functional safety level.

[0083] This application also provides an electronic device, including: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the steps in the above-described method embodiments.

[0084] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the steps in the above-described method embodiments.

[0085] This application provides a computer program product, including a computer program, which, when run on an electronic device, enables the electronic device to implement the steps described in the various method embodiments above.

[0086] It should be noted that the above-described device embodiments and method embodiments are based on the same concept. The specific functions and technical effects of the information interaction, execution process, etc. of the components in the device can be found in the relevant descriptions in the aforementioned method embodiment section, and will not be repeated here.

[0087] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0088] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A vehicle collision adaptive protection method, characterized in that, include: Obtain the predicted trajectory of targets around the vehicle; Collision risk assessment is performed based on the vehicle's motion information and the target's predicted trajectory to obtain collision risk parameters; wherein, the collision risk parameters include at least one of collision time, expected collision overlap rate, and estimated collision energy. When the collision risk parameters meet the preset triggering conditions, the collision type is determined and the occupant's seat belt status is obtained; An inflation strategy is determined based on the collision type and the seatbelt status; wherein, the inflation strategy is used to indicate the priority inflation modules among multiple protective pad modules installed in different parts of the vehicle, the inflation sequence of each priority inflation module, and the target internal pressure; The inflation strategy is used to control the inflation of each priority inflation module so that each priority inflation module can be deployed before a collision occurs.

2. The vehicle collision adaptive protection method as described in claim 1, characterized in that, The acquisition of the predicted trajectory of targets around the vehicle includes: Spatiotemporal alignment and fusion tracking are performed on environmental perception data collected by multiple sensors to obtain the state vector of the target; wherein, the state vector includes the target's position, velocity, heading angle and bounding box size; The historical state sequence of the target's state vectors over the past first preset time period is input into the trajectory prediction model to obtain the predicted position, predicted heading angle, and predicted bounding box size of the target at each predicted time in the future second preset time period, which are used as the predicted trajectory.

3. The vehicle collision adaptive protection method as described in claim 2, characterized in that, The collision risk assessment based on the vehicle's motion information and the target's predicted trajectory, to obtain collision risk parameters, includes: Based on the vehicle's motion information and the target's predicted trajectory, the predicted collision time and predicted collision speed are determined. The collision time is determined based on the relative distance and radial velocity between the vehicle and the target; Based on the predicted position, predicted heading angle, and predicted bounding box size corresponding to the predicted collision time in the predicted trajectory, the predicted bounding box of the target at the predicted collision time is determined. The bounding box of the vehicle and the predicted bounding box are projected onto the horizontal plane, and the ratio of the intersection area of ​​the two projections to the area of ​​the bounding box of the vehicle is used as the expected collision overlap rate. The estimated collision energy is determined based on the predicted collision velocity and the target's mass category.

4. The vehicle collision adaptive protection method as described in claim 1, characterized in that, The preset triggering conditions include warning conditions and execution conditions; When the collision risk parameters meet the preset triggering conditions, the collision type is determined, and the occupant's seatbelt status is obtained, including: When the collision risk parameters meet the execution conditions, the collision type is determined and the occupant's seatbelt status is obtained; The warning conditions include the collision time being less than a first time threshold or the predicted collision overlap rate being greater than a first overlap rate threshold; the execution conditions include any one of the following: the collision time being less than a second time threshold and lateral avoidance is not feasible; the predicted collision overlap rate being greater than a second overlap rate threshold and the estimated collision energy being greater than a preset energy threshold; the lateral acceleration of the vehicle being greater than a preset lateral acceleration threshold and continuously increasing; the second time threshold being less than the first time threshold and the second overlap rate threshold being greater than the first overlap rate threshold.

5. The vehicle collision adaptive protection method as described in claim 1, characterized in that, When the collision risk parameters meet the preset triggering conditions, the collision type is determined, and the occupant's seatbelt status is obtained, including: When the collision risk parameters meet the preset triggering conditions, the impending collision is determined to be one of the following based on the relative motion direction between the vehicle and the target, the predicted collision point position between the vehicle and the target, and the classification result of the target: frontal collision, side collision, rear-end collision, rollover, and multi-angle collision. The occupant's seatbelt status is obtained by reading the detection signal output by the sensor integrated in the seatbelt buckle; The frontal collision, the side collision, and the rear-end collision are determined based on a preset angle range of the angle between the relative velocity vector between the vehicle and the target and the vehicle's direction of travel. The side collision is distinguished between left-side collision and right-side collision. The rollover is determined based on at least one of the vehicle's lateral acceleration and wheel speed difference.

6. The vehicle collision adaptive protection method as described in claim 4, characterized in that, The method further includes: When the collision risk parameters meet the warning conditions, at least one of the following shall be executed: control the gas generator of each of the protective pad modules to enter standby state, pretension the seat belt, issue a warning to the driver, and broadcast a collision warning signal to the vehicle network; The collision warning signal is used to enable the vehicle to perform at least one of path fine-tuning and steering compensation, and to adjust the seat to a protective posture; the vehicle's original airbag controller deploys the original airbags after a collision based on the triggering logic of the acceleration sensor, forming a time difference dual protection with each of the priority inflation modules that have completed deployment before the collision.

7. The vehicle collision adaptive protection method as described in claim 1, characterized in that, The step of determining the inflation strategy based on the collision type and the seatbelt status includes: The inflation strategy is obtained by querying a preset inflation strategy logic table based on the collision type and the seat belt status. The seatbelt status includes a fastened state and an unfastened state; the inflation strategy logic table records the priority inflation modules, inflation order, and target internal pressure corresponding to different combinations of the collision type and the seatbelt status; for the same collision type that distinguishes between the fastened state and the unfastened state in the inflation strategy logic table, the number of priority inflation modules corresponding to the unfastened state is greater than the number of priority inflation modules corresponding to the fastened state, and the maximum value of each target internal pressure corresponding to the unfastened state is higher than the maximum value of each target internal pressure corresponding to the fastened state.

8. The vehicle collision adaptive protection method as described in claim 1, characterized in that, Each of the protective pad modules is equipped with a multi-stage ignition gas generator, which outputs multiple discrete pressure levels through combinations of different ignition stages; The step of controlling the inflation of each priority inflation module according to the inflation strategy, so that each priority inflation module can be deployed before the collision occurs, includes: According to the inflation sequence, the gas generator of each priority inflation module is controlled to inflate at the pressure level closest to the target internal pressure among the plurality of discrete pressure levels, so that each priority inflation module can complete its deployment before the collision occurs.

9. The vehicle collision adaptive protection method as described in claim 1, characterized in that, The method also includes a seat posture adaptive mapping step; The seat posture adaptive mapping step includes: when the seat position changes, calculating the vector direction of each of the protective pad modules relative to the occupant's key anatomical parts, and adjusting the nozzle angle and deployment sequence of the gas generator supplying air to each of the protective pad modules according to the vector direction; wherein, the occupant's key anatomical parts include at least one of the head center of gravity, the sternal center, and the hip joint.

10. The vehicle collision adaptive protection method as described in claim 1, characterized in that, The method also includes a passenger type identification step; The occupant type identification step includes: identifying the occupant type based on seat pressure distribution data and in-vehicle image data; when a forward-facing child safety seat is identified, reducing the target internal pressure of the protective pad module located in front of and above the forward-facing child safety seat or removing it from the priority inflation module, and activating the protective pad module located on the side for wrap-around protection.

11. A vehicle collision adaptive protection device, characterized in that, It includes an environmental sensing component, a main control unit, multiple protective pad modules installed in different parts of the vehicle, and a gas generator that supplies air to each of the protective pad modules; The environmental perception component is used to collect environmental perception data around the vehicle. The main control unit is connected to the environmental perception component and each of the gas generators, and is used to obtain the predicted trajectory of the target around the vehicle based on the environmental perception data, perform a collision risk assessment based on the vehicle's motion information and the predicted trajectory of the target to obtain collision risk parameters, determine the collision type when the collision risk parameters meet the preset trigger conditions, obtain the occupant's seat belt status, determine the inflation strategy based on the collision type and the seat belt status, and control the corresponding gas generator according to the inflation strategy so that the priority inflation module indicated by the inflation strategy completes deployment before the collision occurs. The collision risk parameters include at least one of collision time, expected collision overlap rate, and collision energy estimate; the inflation strategy is used to indicate the priority inflation modules among the plurality of protective pad modules, the inflation sequence of each priority inflation module, and the target internal pressure.

12. The vehicle collision adaptive protection device as described in claim 11, characterized in that, The plurality of protective pad modules include at least two of the following: a canopy module, a left side airbag module, a right side airbag module, a knee airbag module, and a rear head and neck protection module; The roof module is integrated into the roof trim panel to cover the head, neck, and upper chest of the occupants; the left and right side airbag modules are integrated into the corresponding door trim panels or B-pillar trim panels, and their deployment shape is a C-shape with a reserved escape route; the knee airbag module is integrated into the knee pad of the dashboard or the back of the front seat; the rear head and neck protection module is integrated into the active headrest mechanism, which can move forward and upward a preset distance before the rear head and neck protection module inflates and deploys.

13. The vehicle collision adaptive protection device as described in claim 11, characterized in that, The gas generator is a multi-stage ignition gas generator, which outputs multiple discrete pressure levels through combinations of different ignition stages. The main control unit is used to control the multi-stage ignition gas generator corresponding to each priority inflation module to inflate at the pressure level closest to the target internal pressure among the multiple discrete pressure levels, according to the inflation sequence.

14. The vehicle collision adaptive protection device as described in claim 11, characterized in that, The main control unit adopts a dual-core lockstep architecture; The device also includes a backup trigger module independent of the main control unit. The backup trigger module is used to monitor the acceleration signal output by the vehicle's inertial measurement unit, and when the acceleration values ​​at multiple consecutive sampling points exceed a preset trigger acceleration threshold and the main control unit does not receive a trigger signal within a preset response time, the backup trigger module is triggered to deploy through a redundant ignition circuit.

Images