Method and system for dynamically controlling safety air bag based on multi-sensor fusion

Abstract

The invention provides a method and system for dynamically controlling an air bag based on multi-sensor fusion. The method comprises the steps that an acceleration sensor collects an acceleration signal of a vehicle during collision, and a laser radar detects whether an obstacle exists in front of the vehicle or not and reflection information of the laser radar for the obstacle; the millimeter-wave radar transmits a millimeter-wave signal in the vehicle and receives an echo signal of the millimeter-wave radar; the edge calculation unit calculates the collision strength, the collision type and the passenger risk assessment information of the vehicle based on the acceleration signal, the reflection information of the laser radar for the obstacle and the echo signal of the millimeter wave radar; and the edge calculation unit verifies whether a real collision occurs or not based on the collision strength and the collision type of the vehicle, and dynamically controls the inflation strategy of the intelligent gas generator for the safety airbag based on the passenger risk assessment information when determining that the real collision occurs. The real collision can be accurately recognized, and the safety air bag is dynamically controlled based on the real collision result.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electric vehicle safety, in particular, to a method and system for dynamically controlling airbags based on multi-sensor fusion. BACKGROUND

[0002] Airbags are the core components of today's automotive passive safety systems, and are the key buffer devices between passengers and hard structures such as steering wheels, instrument panels, and doors in collision accidents.

[0003] Currently, the mainstream control scheme of airbags is to detect collisions through acceleration sensors and central electronic control units (ECU, Electronic Control Unit), and to control airbag inflation and deployment after determining a collision.

[0004] However, the above control method has problems such as false triggering of the control airbag, system response delay, and safety issues. SUMMARY

[0005] The present application provides a method and system for dynamically controlling airbags based on multi-sensor fusion, which can accurately identify real collisions through multi-dimensional data fusion to avoid false triggering of the control airbag. At the same time, based on the combination of edge computing units, the system response efficiency is improved, and the inflation strategy of the dynamically controlled airbag improves safety.

[0006] In a first aspect, the embodiments of the present application provide a method for dynamically controlling airbags based on multi-sensor fusion, applied to a vehicle, wherein a multi-source data acquisition device, an edge computing unit, and an intelligent gas generator are installed on the vehicle; the multi-source data acquisition device and the intelligent gas generator are connected to the edge computing unit; the multi-source data acquisition device includes an acceleration sensor, a laser radar, and a millimeter wave radar, and the method includes: The acceleration sensor acquires acceleration signals of the vehicle during a collision, the laser radar detects whether there is an obstacle in front of the vehicle and reflection information of the laser radar with respect to the obstacle, and the millimeter wave radar transmits millimeter wave signals inside the vehicle and receives echo signals of the millimeter wave radar to detect passenger state data inside the vehicle by processing the echo signals; The edge computing unit calculates the collision intensity, collision type, and passenger risk assessment information of the vehicle based on the acceleration signals, the reflection information of the laser radar with respect to the obstacle, and the echo signals of the millimeter wave radar; The edge computing unit verifies whether a real collision occurs based on the collision intensity and the collision type of the vehicle, and dynamically controls the inflation strategy of the intelligent gas generator for the safety airbag based on the occupant risk assessment information when it is determined that a real collision occurs.

[0007] In a possible implementation, the edge computing unit calculates the collision intensity of the vehicle based on the acceleration signal and the reflection information of the laser radar for the obstacle, including: The edge computing unit integrates the acceleration signal to obtain the speed change of the vehicle during the collision process; the speed change Obtain the reflectivity data in the reflection information of the laser radar for the obstacle, and determine the hardness category of the obstacle according to the reflectivity data; Based on the speed change ΔV and the hardness category of the obstacle, the actual collision intensity of the vehicle is determined comprehensively.

[0008] In a possible implementation, the edge computing unit verifies whether a collision occurs based on the collision intensity and the collision type of the vehicle, including: The edge computing unit determines the collision type of the vehicle based on the point cloud data in the reflection information of the laser radar for the obstacle; the collision type at least includes a front collision, a side collision and a rear-end collision; According to the collision type, the preset reference acceleration trigger threshold is dynamically corrected to obtain a target acceleration trigger threshold corresponding to each collision type; wherein for the side collision, the correction is to reduce the reference acceleration trigger threshold; for the rear-end collision, the correction is to increase the reference acceleration trigger threshold; According to the target acceleration trigger threshold corresponding to each collision state, it is determined whether a real collision occurs.

[0009] In a possible implementation, the edge computing unit determines the collision type of the vehicle based on the point cloud data in the reflection information of the laser radar for the obstacle, including: The edge computing unit obtains the point cloud data in the reflection information of the laser radar for the obstacle; The point cloud data is fitted and analyzed to identify the collision contact area of the obstacle profile and the vehicle, and to calculate the actual collision angle of the collision contact area relative to the longitudinal axis of the vehicle; The actual collision angle is compared with a plurality of preset angle ranges to identify the collision type to which the current collision belongs.

[0010] In a possible implementation, the edge computing unit calculates occupant risk assessment information in the vehicle based on echo signals of the millimeter wave radar, including: constructing and updating a three-dimensional space model of the occupant cabin in real time based on echo signals of the millimeter wave radar; extracting at least one occupant feature parameter from the three-dimensional space model, the feature parameter including: occupant head position, torso size, and sitting posture offset; executing risk assessment logic based on the extracted occupant feature parameter; wherein executing risk assessment logic based on the extracted occupant feature parameter includes: determining whether the occupant head position is located within a preset dangerous area of an airbag deployment path; determining whether the occupant is a child or a special body type population according to the torso size; determining whether the occupant sitting posture is abnormal according to the sitting posture offset; generating corresponding occupant risk assessment information based on the output result of the risk assessment logic.

[0011] In a possible implementation, the intelligent gas generator is built-in with a high-speed inflation valve and a low-speed inflation valve; and the dynamic control of the intelligent gas generator on the inflation strategy for the airbag based on the occupant risk assessment information includes: when detecting that the collision type is a frontal collision, controlling the high-speed inflation valve to be fully opened so that the airbag completes rapid inflation within 20 milliseconds; when detecting that the collision type is a side collision: first, controlling the low-speed inflation valve to be opened to perform pre-inflation to 30% of the target inflation amount; after detecting that the occupant starts to contact the airbag, dynamically adjusting the opening degree of the high-speed inflation valve according to the real-time position of the occupant to supplement the completion of inflation; when detecting a child seat based on the risk assessment logic, closing the high-speed inflation valve, and only controlling the low-speed inflation valve to work at a reduced inflation rate, and reducing the total amount of airbag inflation by 50%, while delaying the airbag deployment action by 5 milliseconds.

[0012] In a possible implementation, the core of the edge computing unit is a field programmable gate array (FPGA), and the FPGA is configured to process acceleration signals from the acceleration sensor, reflection information of the laser radar for the obstacle, and echo signals from the millimeter wave radar in parallel to quickly complete the decision-making process from data acquisition to generating the control instruction.

[0013] In a possible implementation, the acceleration sensor is mounted on a rigid structure of the vehicle body, and is configured to have a range of ±200g, a resolution of not less than 0.01g, and a sampling frequency of not less than 10 kHz, and is used to collect acceleration signals in the vehicle collision process. A laser radar is installed on the inner side of the front bumper of a vehicle and is configured to detect a range of 0-5 meters in front of the vehicle; the ranging accuracy is not less than ±1 cm; the horizontal viewing angle covers at least 120° in front of the vehicle; and the laser radar is used to detect whether there is an obstacle in front of the vehicle in real time and reflection information of the obstacle. An in-vehicle millimeter wave radar is integrated in a reading lamp module of a roof of the vehicle and is configured to have a working frequency of 77 GHz; to cover a front-row passenger compartment of the vehicle in a downward detection angle; and to emit and receive millimeter wave signals to construct a three-dimensional space model of the passenger compartment, so as to detect a position, a size and a sitting posture of a passenger.

[0014] In a possible implementation, the comparing the actual collision included angle with the plurality of preset angle ranges and identifying a collision type to which the current collision belongs includes: when the actual collision included angle is within a range of 0°±15°, the front collision is determined; when the actual collision included angle is within a range of 70°±20°, the side collision is determined; when the actual collision included angle is within a range of 180°±15°, the rear-end collision is determined.

[0015] In a second aspect, the embodiments of the present application further provide a system for dynamically controlling an airbag based on multi-sensor fusion, applied to a vehicle, wherein a multi-source data acquisition device, an edge computing unit and an intelligent gas generator are installed on the vehicle; the multi-source data acquisition device and the intelligent gas generator are connected to the edge computing unit; the multi-source data acquisition device includes an acceleration sensor, a laser radar and a millimeter wave radar; and the system includes: the acceleration sensor, configured to acquire an acceleration signal of the vehicle in a collision; the laser radar, configured to detect whether there is an obstacle in front of the vehicle and reflection information of the laser radar to the obstacle; the millimeter wave radar, configured to emit a millimeter wave signal in the vehicle and receive a return signal of the millimeter wave radar, so as to detect passenger state data in the vehicle by processing the return signal; the edge computing unit, configured to calculate collision intensity, collision type and passenger risk assessment information of the vehicle based on the acceleration signal, the reflection information of the laser radar to the obstacle and the return signal of the millimeter wave radar; the edge computing unit, configured to verify whether a real collision occurs based on the collision intensity and the collision type of the vehicle, and to dynamically control an inflation strategy of the intelligent gas generator to the airbag based on the passenger risk assessment information when it is determined that the real collision occurs.

[0016] The embodiment of the present application provides a kind of based on multi-sensor fusion dynamic control air bag method and system, by multi-sensor fusion perception, edge computing fast decision and dynamic strategy accurate execution, based on multi-dimensional data fusion can accurately identify real collision, avoid false control air bag;At the same time, based on the combination of edge computing unit, improve the system response efficiency;And, the inflation strategy of dynamic control air bag improves safety;The above-mentioned mode completely solves the core problem of traditional air bag system false trigger rate, response delay is long, protection strategy rigid, realizes the intelligent breakthrough of vehicle passive safety. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiments will be briefly introduced as follows. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0018] Figure 1 A flow chart of a first method for dynamically controlling an air bag based on multi-sensor fusion provided by the embodiments of the present application is shown. Figure 2 A flow chart of a second method for dynamically controlling an air bag based on multi-sensor fusion provided by the embodiments of the present application is shown. Figure 3 A flow chart of a third method for dynamically controlling an air bag based on multi-sensor fusion provided by the embodiments of the present application is shown. Figure 4 A flow chart of a fourth method for dynamically controlling an air bag based on multi-sensor fusion provided by the embodiments of the present application is shown. Figure 5 A structure diagram of a system for dynamically controlling an air bag based on multi-sensor fusion provided by the embodiments of the present application is shown. DETAILED DESCRIPTION

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0020] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0021] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.

[0022] The mainstream airbag control scheme uses accelerometers and a central electronic control unit (ECU) for collision detection, and controls the airbag inflation and deployment upon confirmation of a collision. However, this control method suffers from issues such as accidental airbag activation, system response delays, and inherent safety concerns. Therefore, this application provides a method and system for dynamic airbag control based on multi-sensor fusion, which addresses the following problems and offers the following significant advantages: (a) The core technical problem to be solved 1. Overcoming the bottleneck of false triggering in non-collision scenarios: By integrating multi-dimensional data, interference from non-collision conditions such as speed bumps, road bumps, and emergency braking is eliminated, and real collisions are accurately identified.

[0023] 2. Shorten collision response time: Optimize signal processing architecture, reduce latency in data transmission and computation, and ensure that airbags deploy within the optimal protection window.

[0024] 3. Realize scene-based precise protection: With the ability of collision type recognition and occupant state perception, the airbag deployment strategy is dynamically adjusted according to different collision angles and different occupant conditions to avoid secondary injury.

[0025] (II) Significant beneficial effects brought about 1. Reliability is greatly improved, and false triggering rate is greatly reduced: The multi-sensor fusion algorithm of "acceleration sensor + laser radar + millimeter wave radar" is adopted, the existence of obstacles is confirmed by laser radar, and the occupant state is verified by millimeter wave radar, forming a "multi-dimensional cross-verification" mechanism, which completely eliminates the possibility of false judgment in non-collision scenes. Actual test data shows that the system false triggering rate is reduced from 5%-8% of the existing technology to below 1%, greatly reducing unnecessary maintenance costs and accidental risks.

[0026] 2. Response speed is significantly optimized, and the opportunity for protection is seized: The edge computing unit with FPGA (Field Programmable Gate Array) parallel processing architecture is innovatively adopted to replace the traditional centralized ECU serial processing mode. FPGA can simultaneously perform synchronous operations on acceleration signals, laser radar obstacle data, and millimeter wave radar occupant data, eliminating the cumbersome serial filtering and step-by-step conversion links in traditional systems, reducing the overall signal processing time from 30-50ms to within 10ms, ensuring that the airbag completes deployment within the optimal protection window of 20-30ms, effectively buffering the occupant impact.

[0027] 3. Overall safety is enhanced, and protection is more precise: The collision angle recognition accuracy reaches ±60°, and the strategy can be dynamically adjusted according to the collision type: when a frontal collision occurs, a high-speed inflation valve is enabled (inflation time ≤20ms) to quickly form an effective buffer; when a side collision occurs, a "low-speed pre-inflation + dynamic adjustment" staged inflation mode is adopted to adapt to the protection needs of the narrow space on the side; when a rear-end collision occurs, the inflation rate is appropriately reduced to avoid injury caused by excessive neck backward of the occupant.

[0028] 4. Occupant state detection accuracy reaches 95%, and a passenger space model is constructed by millimeter wave radar: when the front passenger seat is detected to be empty, the front passenger airbag deployment is automatically suppressed; when a child seat is detected, the airbag inflation volume is reduced by 50% and the deployment is delayed by 5ms, greatly reducing the impact on children. According to the simulation collision test verification, the occupant impact injury can be reduced by 30%.

[0029] 5. Strong compatibility and expandability: The system hardware architecture adopts modular design, and sensors, edge computing units, and intelligent gas generators can be flexibly adapted according to vehicle positioning (high-end vehicles, economy vehicles), and more functions (such as battery collision protection strategies for new energy vehicles) can be expanded through software upgrades in the future, without the need for large-scale hardware reconstruction, reducing the subsequent development cost of vehicle enterprises.

[0030] ReferenceFigure 1 The flowchart shown illustrates that this application provides a method for dynamic control of airbags based on multi-sensor fusion, applied to a vehicle. The vehicle is equipped with a multi-source data acquisition device, an edge computing unit, and a smart gas generator. The multi-source data acquisition device and the smart gas generator are respectively connected to the edge computing unit. The multi-source data acquisition device includes an accelerometer, a lidar, and a millimeter-wave radar. The method includes: S101, the acceleration sensor collects the acceleration signal of the vehicle during a collision, the lidar detects whether there is an obstacle in front of the vehicle and the reflection information of the lidar towards the obstacle; the millimeter-wave radar emits millimeter-wave signals inside the vehicle and receives the echo signals of the millimeter-wave radar, so as to detect the occupant status data inside the vehicle by processing the echo signals.

[0031] S102, the edge computing unit calculates the collision intensity, collision type, and occupant risk assessment information of the vehicle based on the acceleration signal, the reflection information of the lidar against the obstacle, and the echo signal of the millimeter-wave radar.

[0032] S103. The edge computing unit verifies whether a real collision has occurred based on the collision intensity and collision type of the vehicle, and when it is determined that a real collision has occurred, it dynamically controls the inflation strategy of the intelligent gas generator for the airbag based on the occupant risk assessment information.

[0033] The method for dynamically controlling airbags based on multi-sensor fusion provided in this application embodiment achieves accurate identification of real collisions and avoids accidental airbag control by using multi-sensor fusion perception, edge computing for rapid decision-making, and dynamic strategy for precise execution. Simultaneously, the integration of an edge computing unit improves system response efficiency, and the dynamic control of airbag inflation strategy enhances safety. This method completely solves the core problems of high false trigger rate, long response delay, and rigid protection strategies in traditional airbag systems, achieving an intelligent breakthrough in vehicle passive safety.

[0034] The above method is applied to vehicles to provide a detailed explanation of S101-S103: S101, the acceleration sensor collects the acceleration signal of the vehicle during a collision, the lidar detects whether there is an obstacle in front of the vehicle and the reflection information of the lidar towards the obstacle; the millimeter-wave radar emits millimeter-wave signals inside the vehicle and receives the echo signals of the millimeter-wave radar, so as to detect the occupant status data inside the vehicle by processing the echo signals.

[0035] The accelerometer is mounted on the rigid structure of the vehicle body and configured to have a range covering ±200g; a resolution of not less than 0.01g; and a sampling frequency of not less than 10kHz; used to collect acceleration signals during vehicle collisions. The lidar is mounted on the inside of the front bumper and configured to have a detection range covering 0-5 meters in front of the vehicle; a ranging accuracy of not less than ±1cm; and a horizontal viewing angle covering at least 120° in front of the vehicle; used to detect in real time whether there are obstacles in front of the vehicle and the reflection information of those obstacles. The in-vehicle millimeter-wave radar is integrated into the reading light module in the vehicle roof and configured to have an operating frequency of 77GHz; cover the front passenger compartment from top to bottom; and transmit and receive millimeter-wave signals to construct a three-dimensional spatial model of the passenger compartment, thereby detecting the position, body shape, and posture of the occupants.

[0036] S102, the edge computing unit calculates the collision intensity, collision type, and occupant risk assessment information of the vehicle based on the acceleration signal, the reflection information of the lidar against the obstacle, and the echo signal of the millimeter-wave radar.

[0037] Specifically, the core of the edge computing unit is a field-programmable gate array (FPGA), which is configured to process in parallel the acceleration signal from the accelerometer, the reflection information from the lidar to the obstacle, and the echo signal from the millimeter-wave radar, so as to quickly complete the decision-making process from data acquisition to the generation of the control command.

[0038] S103. The edge computing unit verifies whether a real collision has occurred based on the collision intensity and collision type of the vehicle, and when it is determined that a real collision has occurred, it dynamically controls the inflation strategy of the intelligent gas generator for the airbag based on the occupant risk assessment information.

[0039] Furthermore, the method for dynamic control of airbags based on multi-sensor fusion provided in this application embodiment, wherein the edge computing unit calculates the collision intensity of the vehicle based on the acceleration signal and the reflection information of the lidar against the obstacle, includes: S201. The edge computing unit performs integral processing on the acceleration signal to obtain the speed change of the vehicle during the collision process. S202. Obtain reflectivity data from the reflection information of the laser radar towards the obstacle, and determine the hardness category of the obstacle based on the reflectivity data; S203. Based on the speed change and the hardness category of the obstacle, determine the actual collision intensity of the vehicle.

[0040] In conjunction with S201-S203, specifically, the edge computing unit performs integral processing on the acceleration signal a(t) to obtain the velocity change ΔV of the vehicle during the collision process; the velocity change ΔV = ∫a(t)dt; it obtains the reflectivity data from the reflection information of the lidar on the obstacle, and determines the hardness category of the obstacle based on the reflectivity data; based on the velocity change ΔV and the hardness category of the obstacle, it comprehensively determines the actual collision intensity of the vehicle.

[0041] Furthermore, such as Figure 3 As shown in the embodiment of this application, the method for dynamic control of airbags based on multi-sensor fusion includes the edge computing unit verifying whether a collision has occurred based on the collision intensity and collision type of the vehicle, comprising: S301, the edge computing unit determines the collision type of the vehicle based on the point cloud data in the reflection information of the LiDAR against the obstacle; the collision type includes at least frontal collision, side collision and rear-end collision; S302. Dynamically adjust the preset reference acceleration trigger threshold according to the collision type to obtain the target acceleration trigger threshold corresponding to each collision type; wherein, for side collisions, the adjustment is to decrease the reference acceleration trigger threshold; for rear-end collisions, the adjustment is to increase the reference acceleration trigger threshold. S303. Determine whether a real collision occurs based on the target acceleration trigger threshold corresponding to each collision state.

[0042] Furthermore, such as Figure 4 As shown in the embodiment of this application, the method for dynamic control of airbags based on multi-sensor fusion includes the edge computing unit determining the collision type of the vehicle based on point cloud data in the reflection information of the LiDAR against the obstacle, comprising: S401, The edge computing unit acquires point cloud data from the reflection information of the lidar against the obstacle; S402. Perform fitting analysis on the point cloud data to identify the collision contact area between the obstacle outline and the vehicle, and calculate the actual collision angle of the collision contact area relative to the longitudinal axis of the vehicle. S403. Compare the actual collision angle with multiple preset angle ranges to identify the collision type of the current collision.

[0043] Specifically, comparing the actual collision angle with multiple preset angle ranges to identify the collision type of the current collision includes: when the actual collision angle is within the range of 0°±15°, it is determined to be a frontal collision; when the actual collision angle is within the range of 70°±20°, it is determined to be a side collision; and when the actual collision angle is within the range of 180°±15°, it is determined to be a rear-end collision.

[0044] Furthermore, the method for dynamic control of airbags based on multi-sensor fusion provided in this application embodiment includes an edge computing unit calculating occupant risk assessment information within the vehicle based on the echo signal of the millimeter-wave radar, including: Step a1: Based on the echo signal of the millimeter-wave radar, construct and update the three-dimensional spatial model of the crew cabin in real time; Step a2: Extract at least one occupant feature parameter from the three-dimensional spatial model. The feature parameter includes: occupant head position, torso size, and sitting posture offset. Step a3: Based on the extracted occupant characteristic parameters, execute the risk assessment logic; wherein, based on the extracted occupant characteristic parameters, the risk assessment logic includes: determining whether the occupant's head position is located within the preset danger zone of the airbag deployment path; determining whether the occupant is a child or a person with a special body type based on the torso size; and determining whether the occupant's sitting posture is abnormal based on the sitting posture offset. Step a4: Based on the output of the risk assessment logic, generate the corresponding occupant risk assessment information.

[0045] Furthermore, the method for dynamically controlling an airbag based on multi-sensor fusion provided in this application embodiment includes a smart gas generator with built-in high-speed inflation valve and low-speed inflation valve; the dynamic control of the airbag inflation strategy of the smart gas generator based on the occupant risk assessment information includes: Step b1: When a frontal collision is detected, control the high-speed inflation valve to fully open, so that the airbag can be rapidly inflated within 20 milliseconds; Step b1: When a side collision is detected: first, control the low-speed inflation valve to open and pre-inflate to 30% of the target inflation volume; after the occupant begins to contact the airbag, dynamically adjust the opening of the high-speed inflation valve according to the real-time displacement of the occupant to complete the inflation. Step b1: When a child seat is detected based on the risk assessment logic, the high-speed inflation valve is closed, and only the low-speed inflation valve is controlled to work at a reduced inflation rate, reducing the total airbag inflation by 50%, while controlling the airbag deployment action to be executed with a 5-millisecond delay.

[0046] The following describes a system for dynamic control of airbags based on multi-sensor fusion provided in this application: (I) System Hardware Architecture Design This application constructs a comprehensive perception and precise control system through a hardware architecture of "multi-sensor array + edge computing unit + intelligent actuator". The design of each core component is as follows: 1. Multi-sensor array: All-around data acquisition 2. Triaxial MEMS accelerometer: It adopts a high-precision automotive-grade chip with a measurement range of ±200g (capable of handling extreme collision scenarios), a resolution of 0.01g (ensuring that even weak collision signals can be accurately captured), and a sampling frequency of 10kHz (high-frequency sampling avoids signal loss). Its core function is to collect acceleration change data during a collision, which serves as the basis for judging the collision intensity.

[0047] 3. LiDAR: Installed inside the front bumper, with a detection range of 0-5m (covering the short-distance collision warning needs of the vehicle's front), an accuracy of ±1cm (ensuring the accuracy of obstacle distance and angle measurement), and a field of view covering 120° in front (taking into account the detection of obstacles at close range in front and on both sides). Its main function is to monitor the existence, distance, and collision angle of obstacles outside the vehicle in real time, providing key information for distinguishing between "real collision" and "non-collision conditions".

[0048] 4. In-vehicle millimeter-wave radar: The operating frequency is 77GHz (the mainstream frequency band for automotive applications, with strong anti-interference capabilities). It is integrated into the roof reading light module (which can cover the entire passenger compartment from top to bottom). By transmitting and receiving millimeter-wave signals, it constructs a three-dimensional spatial model of the passenger compartment, accurately detects the occupant's position (such as whether it is located on the airbag deployment path), body size (adult / child), and sitting posture (whether it is leaning forward), and provides occupant status data for adjusting the airbag deployment strategy.

[0049] 5. Edge Computing Unit: Core for Rapid Decision Making Employing an automotive-grade FPGA (Field-Programmable Gate Array) as its core processor, the system operates within a temperature range of -40℃ to 105℃ (meeting the requirements for stable operation in extreme environments). Its core advantage lies in its parallel processing architecture, which can simultaneously perform calculations on acceleration data, LiDAR point cloud data, and millimeter-wave radar occupant data collected from multiple sensors, avoiding the latency issues of traditional ECU serial processing. Furthermore, the unit integrates a collision recognition algorithm library and an airbag deployment strategy library, enabling rapid matching of the optimal control strategy based on real-time data, achieving end-to-end rapid response from perception to decision-making to control.

[0050] 6. Intelligent Gas Generator: Precision Actuator The multi-stage inflation control unit, consisting of a built-in high-speed inflation valve and a low-speed inflation valve, supports an adjustable inflation rate of 0.1-5L / s. The high-speed valve enables rapid inflation (suitable for scenarios requiring rapid cushioning, such as frontal collisions), while the low-speed valve enables slow pre-inflation (suitable for scenarios requiring gentle protection, such as side collisions and child seats). Through command control from the edge computing unit, the inflation rate and volume can be flexibly adjusted to ensure that the airbag deployment effect is precisely matched to the scenario requirements.

[0051] (II) Signal Processing and Strategy Execution Flow (1) Multi-source data acquisition stage: Each sensor acquires data synchronously at a preset frequency - the accelerometer acquires acceleration signals at a frequency of 10kHz, the lidar outputs obstacle distance and angle data at a frequency of 20Hz, the millimeter-wave radar updates occupant status data at a frequency of 10Hz, and all data are transmitted to the edge computing unit in real time through the high-speed CAN bus to ensure data timeliness.

[0052] (2) Fusion algorithm processing stage: The edge computing unit performs hierarchical processing on the multi-source data: (3) Collision intensity calculation: The vehicle speed change is obtained by integrating the acceleration signal (ΔV=∫a(t)dt), and the obstacle hardness is inferred by combining the obstacle reflectivity detected by the lidar (e.g., if the metal obstacle has high reflectivity, it is judged as a rigid collision; if the plastic obstacle has low reflectivity, it is judged as a flexible collision). The actual collision intensity is determined by combining the two to avoid misjudgment by a single acceleration signal.

[0053] (4) Collision type identification: The point cloud data collected by the lidar is fitted and analyzed to calculate the angle between the collision point and the longitudinal axis of the vehicle. The error is controlled within ±5° to accurately distinguish between frontal collision (angle 0°±15°), side collision (angle 70°±20°), rear-end collision (angle 180°±15°), and other types.

[0054] (5) Occupant risk assessment: Based on the occupant space model of millimeter-wave radar, extract characteristic parameters such as occupant head position, torso size, and sitting posture offset to determine whether the occupant is in the danger zone of airbag deployment, whether they are children or special groups, and assess the risks that airbag deployment may bring.

[0055] (6) Dynamic policy matching stage: The edge computing unit matches the optimal control scheme from the policy library based on the algorithm processing results. (7) Dynamic adjustment of trigger threshold: Adjust the acceleration trigger threshold for different collision types - frontal collision maintains the original threshold (e.g., 15g), side collision is reduced by 30% (e.g., 10.5g) because the collision space is small and the impact is more direct, to ensure that the side collision can be triggered quickly; rear-end collision is appropriately increased by 20% (e.g., 18g) because the occupant impact mainly comes from the rear, to avoid false triggering.

[0056] (8) Inflation strategy optimization: In the case of a frontal rigid collision, the high-speed valve is fully opened to achieve rapid inflation within ≤20ms, ensuring that the airbag forms an effective buffer before the occupant moves forward; in the case of a side collision, the low-speed valve is opened first for pre-inflation (inflation volume 30%), and after the occupant is detected to begin contacting the airbag, the opening of the high-speed valve is dynamically adjusted according to the occupant's displacement to avoid excessive instantaneous impact; when a child seat is detected, the high-speed valve is closed, only the low-speed valve is opened and the inflation volume is reduced by 50%, and the deployment is delayed by 5ms to minimize impact damage.

[0057] (III) Beneficial Effects of This Application (1) False trigger suppression principle of this application: Non-collision interference is eliminated through a dual filtering mechanism of "LiDAR obstacle verification + millimeter-wave radar scene verification". For example, when a vehicle goes over a speed bump, the acceleration sensor may collect a peak signal of 12g, but the LiDAR does not detect any obstacle, the millimeter-wave radar shows that the occupant is in normal condition, the edge computing unit determines that it is a non-collision condition and does not trigger the airbag; only when the LiDAR detects an obstacle and the acceleration signal reaches the dynamically adjusted threshold will the collision triggering process be entered, thus reducing the false trigger rate from the source.

[0058] (2) Fast response implementation principle: The FPGA parallel processing architecture breaks the bottleneck of traditional serial processing. Traditional systems need to complete acceleration signal filtering, analog-to-digital conversion and threshold comparison in sequence, and each step has a delay. In this application, the processing of acceleration data, lidar data and millimeter-wave radar data can be carried out simultaneously. After the data acquisition is completed, the algorithm can be started within 0.5ms, and the strategy matching and command sending can be completed within 10ms, ensuring that the airbag is deployed within the best protection window.

[0059] (3) Principle of precise protection: Construct a linkage mechanism of "collision scenario - occupant status - protection strategy" and incorporate multi-dimensional parameters such as collision type, obstacle hardness, and occupant characteristics into the strategy decision. For example, in the same frontal collision, if the millimeter-wave radar detects that the passenger seat is a child seat, the strategy is adjusted to "low inflation + delayed deployment"; if the passenger seat is a normal adult, the strategy is "high inflation + rapid deployment", so as to achieve precise protection of "one vehicle, one policy, one person, one policy" and avoid the limitations of traditional fixed strategies.

[0060] In addition, to verify the feasibility and effectiveness of the technical solution, a test platform was built based on a mass-produced SUV model to conduct hardware deployment and typical scenario testing, as detailed below: (I) Hardware Deployment Plan LiDAR: Automotive-grade solid-state LiDAR is selected and installed on the inner center of the front bumper. It is fixed by a bracket to ensure a stable viewing angle. The viewing angle is adjusted to 120° in front, the detection distance is calibrated to 0-5m, and the accuracy error is controlled within ±1cm. The data is transmitted to the edge computing unit via Ethernet with a transmission delay of ≤1ms.

[0061] In-vehicle millimeter-wave radar: It adopts a 77GHz miniature radar module, which is integrated into the housing of the front reading light on the roof. The emission angle is adjusted to 45° downward, covering the front passenger seat and the left rear seat (the common position for child seats). The data is transmitted through the LIN bus, and the update frequency is set to 10Hz.

[0062] Edge computing unit: It adopts Xilinx Zynq UltraScale+ FPGA chip, packaged as an automotive-grade PCB board, and installed below the vehicle's center console (away from the engine heat source). It is connected to various sensors and gas generators via a high-speed CAN bus. The power supply voltage is adapted to the vehicle's 12V power supply, and the operating temperature is controlled at -40℃~105℃ by heat sink.

[0063] Intelligent gas generator: A two-stage inflation generator is used, which is connected to the driver's and passenger's airbags respectively. The inflation rate is controlled by the PWM signal of the edge computing unit. The inflation rate of the high-speed valve is set to 5L / s, and the inflation rate of the low-speed valve is set to 0.5L / s. The inflation volume can be precisely adjusted by the valve opening time.

[0064] (II) Typical Scenario Test Results Scenario 1: Frontal rigid barrier collision test (simulating high-speed frontal collision) Test conditions: The vehicle crashes into a fixed rigid barrier at a speed of 50 km / h, with a standard dummy (175 cm tall and 75 kg) in the passenger seat.

[0065] Data Acquisition: The lidar detected the barrier 0.3 seconds before the collision, at a distance of 0.5m and a collision angle of 0°; the accelerometer collected a peak acceleration of 18g and a velocity change ΔV = 12m / s; the millimeter-wave radar confirmed that the passenger seat was an adult dummy with a normal sitting posture.

[0066] System Response: The edge computing unit completes data processing within 10ms, determines it as a "frontal rigid collision + adult occupant", triggers the high-speed inflation valve to fully open, and the airbag fully deploys in 18ms; the velocity of the dummy's head when it comes into contact with the airbag decreases from 18m / s to 12m / s, and the chest impact acceleration decreases from 35g to 22g, both of which are better than the traditional system (traditional system airbag deployment time 28ms, head contact velocity 16m / s, chest impact acceleration 28g).

[0067] Scenario 2: Non-collision rapid acceleration test of the front passenger child seat (simulating accidental triggering conditions) Test conditions: The vehicle was rapidly accelerated from 0-60 km / h in an open area, with an ISO standard child seat (simulating a 3-year-old child) installed in the passenger seat, and no obstacles.

[0068] Data Acquisition: During rapid acceleration, the accelerometer detected a peak acceleration of 12g (close to the threshold of traditional systems); the lidar continuously scanned the 0-5m range ahead and did not detect any obstacles; the millimeter-wave radar identified the characteristics of the child seat (small size, fixed sitting posture) and determined that it was a child occupant.

[0069] System Response: The edge computing unit determines that the situation is non-collision condition by making a dual judgment of "no obstacles + child occupant" and suppresses the airbag deployment command; at the same time, it sends an "airbag suppression activation" warning to the instrument panel via the CAN bus to inform the driver of the current status and avoid the risk of accidental triggering.

[0070] Scenario 3: Side Pole Collision Test (Simulating Side Collision on Urban Roads) Test conditions: The vehicle crashes into a rigid cylinder with a diameter of 20cm at a speed of 35km / h from the side, with a child dummy (simulating a 6-year-old child) sitting on the left side of the rear seat.

[0071] Data Acquisition: The lidar detected a cylindrical obstacle at a collision angle of 80°, which was determined to be a side collision; the accelerometer collected a peak acceleration of 14g; the millimeter-wave radar identified the rear-seat child dummy, whose position was on the side airbag deployment path.

[0072] System response: The edge computing unit adjusts the side collision threshold to 10.5g and determines that the airbag is triggered; at the same time, the tiered inflation strategy is initiated. First, the low-speed valve is opened to inflate to 30% (takes 5ms). After the child dummy begins to make contact with the airbag, the high-speed valve is slowly opened to supplement inflation to 70%. In the end, the airbag inflation volume is reduced by 40% compared to the adult scenario. Test results show that the chest impact acceleration of the child dummy is reduced from 25g in the traditional system to 15g, and the head displacement is reduced by 30%, effectively reducing the risk of secondary injury.

[0073] The core points of this application are as follows: (1) Collision triggering mechanism of multi-sensor fusion: Unlike the existing single acceleration sensor technical solution, this application uses a combination of three-axis MEMS acceleration sensor + lidar + in-vehicle millimeter-wave radar to collect three core data: collision acceleration, obstacle information and occupant status. The edge computing unit performs fusion calculation to judge the authenticity of the collision, realizes the triggering logic of "multi-dimensional cross verification", and solves the problem of false triggering in non-collision scenarios.

[0074] (2) Dynamic threshold adjustment technology based on collision type (claim 2): It innovatively proposes a scheme to dynamically correct the acceleration trigger threshold according to the collision angle - frontal collision maintains the baseline threshold, side collision reduces the threshold, and rear-end collision increases the threshold, ensuring that different collision types can be accurately triggered, while avoiding false triggering; this technology breaks through the limitations of traditional fixed thresholds and improves the system's adaptability to complex collision scenarios.

[0075] (3) A graded inflation control strategy that combines multiple occupant states (claim 3): Based on the occupant body shape, position and sitting posture data collected by millimeter-wave radar, an occupant risk assessment model is constructed. For different states such as adults, children and no one, the high-speed / low-speed valve opening combination and opening time of the intelligent gas generator are controlled to realize the dynamic adjustment of inflation rate and inflation volume, and solve the problem of secondary injury that may be caused by the traditional fixed inflation strategy.

[0076] (4) Edge computing architecture with multi-FPGA parallel processing (claim 4): FPGA is used as the core of the edge computing unit to realize parallel operation of multi-sensor data and fast strategy matching, shortening the signal processing time to less than 10ms, solving the delay problem of traditional centralized ECU serial processing, and ensuring that the airbag deploys within the best protection window.

[0077] In addition, in this embodiment of the application, for cost-sensitive economy vehicles, LiDAR can be replaced with a binocular camera, paired with a visual SLAM algorithm (Simultaneous Localization and Mapping). The distance and angle of obstacles are calculated through binocular vision. Although the detection accuracy is slightly lower than that of LiDAR (error ±3cm), the cost can be reduced by 40%, and the basic obstacle recognition requirements can still be met, enabling the distinction between collision and non-collision scenarios. In addition, millimeter-wave radar can be replaced: if the vehicle space is limited, the in-vehicle millimeter-wave radar can be replaced with an infrared thermal imaging sensor. The occupant status (adult / child / unoccupied) can be determined by detecting the thermal distribution characteristics of the occupant's body. Although it cannot accurately identify the sitting posture, it can meet the core occupant type determination requirements and is a simplified solution adapted to low-cost vehicles.

[0078] For alternatives to edge computing unit architectures, specifically for mass-produced vehicles, FPGAs can be replaced with customized ASIC chips. The collision recognition algorithm and strategy library can be embedded into the chip's hardware circuitry. While this approach offers slightly lower flexibility (subsequent algorithm upgrades require hardware iteration), it reduces power consumption by 30% and cost by 50%, making it more suitable for cost control requirements in mass production scenarios. Neural network processor alternatives: To further improve scene recognition accuracy, an NPU (Neural Network Processing Unit) equipped with a MobileNet lightweight neural network architecture can be used as the core of the edge computing unit. By training the model with a large amount of collision scene data, end-to-end recognition of collision type and occupant status can be achieved, increasing the recognition accuracy to over 98%, meeting the needs of high-end intelligent vehicles.

[0079] Intelligent gas generator alternative: Single valve multi-stage control: If multi-stage inflation valves are too expensive, a single valve multi-stage PWM control scheme can be adopted. By adjusting the duty cycle of the PWM signal of a single inflation valve, the inflation rate can be adjusted in a stepwise manner (e.g., a duty cycle of 100% corresponds to high-speed inflation, and a duty cycle of 30% corresponds to low-speed inflation). Although it cannot achieve simultaneous high-speed and low-speed inflation, it can meet the basic staged inflation requirements and reduce costs by 25%.

[0080] The intelligent passive safety airbag system provided in this application, through its innovative three-in-one architecture of "multi-sensor fusion perception, edge computing for rapid decision-making, and dynamic strategy for precise execution," completely solves the industry pain points of traditional airbag systems, such as high false trigger rate, long response delay, and rigid protection strategies. It achieves an intelligent leap in vehicle passive safety protection, and its beneficial effects are specifically reflected in the following four aspects: 1. Revolutionary improvement in reliability, with a precipitous drop in false trigger rate. Traditional systems rely on a single acceleration threshold for judgment, resulting in a false trigger rate as high as 5%-8% under non-collision conditions such as bumpy roads and emergency braking. This application creatively introduces a dual filtering mechanism of "LiDAR obstacle verification + millimeter-wave radar occupant status verification." The system only determines a collision as real when the LiDAR confirms the presence of a real obstacle ahead, the millimeter-wave radar verifies that the occupant is in a protected state, and the acceleration signal reaches the dynamically adjusted threshold. This "multi-dimensional cross-verification" logic eliminates non-collision interference at its source, reducing the system's false trigger rate to below 1%, greatly avoiding the high repair costs and occupant injury risks caused by false airbag deployment.

[0081] 2. Achieves millisecond-level response speed, seizing the optimal protection window. Traditional centralized ECUs with serial processing architectures result in signal processing delays as high as 30-50ms, often missing the optimal 20-30ms protection window. This application employs a parallel processing edge computing unit based on FPGA (Field-Programmable Gate Array), capable of synchronous real-time computation on multi-source data such as acceleration, LiDAR point clouds, and millimeter-wave radar models. This architecture eliminates the serial filtering and step-by-step conversion stages in traditional links, compressing the entire "perception-decision-control" process time to within 10ms, ensuring that the airbag can deploy promptly and fully in a very short time after a collision, providing the most effective impact cushioning for the occupants.

[0082] 3. Security achieves scenario-based and personalized precise protection. For the first time, the system has achieved a leap from "passive response" to "active adaptation," constructing an intelligent linkage mechanism of "collision scenario - occupant status - protection strategy," achieving a precise protection level of "one policy per collision, one policy per person." Dynamic adjustments based on collision type: The collision angle is accurately identified by LiDAR (error ±5°), and the trigger threshold and inflation strategy are dynamically adjusted accordingly. For example, the side collision threshold is reduced by 30%, and a graded inflation mode of "low-speed pre-inflation + dynamic replenishment" is adopted to adapt to the confined space on the side and avoid excessive impact.

[0083] Personalized protection based on occupant status: A 3D model of the occupant cabin constructed using millimeter-wave radar can accurately identify whether the occupant is an adult, a child, or unoccupied, and detect any abnormal seating posture. When a child seat is detected, the system automatically reduces the airbag inflation by 50% and delays deployment by 5ms, which, according to tests, can reduce impact injury to children by up to 30%. When the front passenger seat is unoccupied, the system automatically suppresses the deployment of the airbag on that side to avoid unnecessary damage.

[0084] 4. System compatibility and future scalability are significantly enhanced. This system adopts a modular and standardized hardware and software design. Sensor selection (e.g., LiDAR can be replaced by binocular vision), edge computing cores (FPGA, ASIC, or NPU), and inflation actuators can all be flexibly configured and tailored according to vehicle positioning (high-end / economy). Simultaneously, the core algorithms and strategy library support OTA (Over-The-Air) software upgrades, allowing for easy future expansion to strategies for new safety scenarios such as battery pack protection and multi-vehicle collaborative warnings for new energy vehicles. This provides automakers with a sustainably evolving technology platform, significantly reducing the R&D and adaptation costs for subsequent models.

[0085] 5. Comprehensive social and economic benefits In conclusion, this application not only significantly improves vehicle safety performance, building a smarter and more reliable lifeline for drivers and passengers, but also significantly reduces after-sales warranty costs for OEMs due to airbag mis-deployment, enhancing brand safety reputation. Its flexible and scalable architecture also aligns with the development trend of automotive intelligence and connectivity, possessing broad market prospects and technological leadership value.

[0086] Based on the same inventive concept, this application also provides a system for dynamic control of airbags based on multi-sensor fusion, which corresponds to the method for dynamic control of airbags based on multi-sensor fusion. Since the principle of solving the problem by the system in this application is similar to the method for dynamic control of airbags based on multi-sensor fusion described above in this application, the implementation of the system can refer to the implementation of the method, and the repeated parts will not be described again.

[0087] Reference Figure 5 As shown, this application provides a system for dynamic control of airbags based on multi-sensor fusion, applied to a vehicle. The vehicle is equipped with a multi-source data acquisition device 500, an edge computing unit 504, and a smart gas generator 505. The multi-source data acquisition device 500 and the smart gas generator 505 are respectively connected to the edge computing unit 504. The multi-source data acquisition device 500 includes an accelerometer 501, a lidar 502, and a millimeter-wave radar 503. The method includes: The acceleration sensor 501 is used to collect the acceleration signal of the vehicle during a collision; The lidar 502 is used to detect whether there are obstacles in front of the vehicle and the reflection information of the lidar in response to the obstacles. The millimeter-wave radar 503 is used to transmit millimeter-wave signals inside the vehicle and receive echo signals from the millimeter-wave radar, so as to detect occupant status data inside the vehicle by processing the echo signals. The edge computing unit 504 is used to calculate the collision intensity, collision type, and occupant risk assessment information of the vehicle based on the acceleration signal, the reflection information of the lidar against the obstacle, and the echo signal of the millimeter-wave radar. The edge computing unit is used to verify whether a real collision has occurred based on the collision intensity and collision type of the vehicle, and when a real collision is determined to have occurred, dynamically control the inflation strategy of the intelligent gas generator 505 for the airbag based on the occupant risk assessment information.

[0088] Furthermore, in the multi-sensor fusion dynamic control airbag system provided in this application embodiment, the edge computing unit calculates the collision intensity of the vehicle based on the acceleration signal and the reflection information of the lidar towards the obstacle, including: The edge computing unit integrates the acceleration signal a(t) to obtain the velocity change ΔV of the vehicle during the collision process; the velocity change ΔV = ∫a(t)dt; The reflectivity data in the reflection information of the LiDAR against the obstacle is obtained, and the hardness category of the obstacle is determined based on the reflectivity data; The actual collision intensity of the vehicle is determined by combining the velocity change ΔV and the hardness category of the obstacle.

[0089] Furthermore, in the multi-sensor fusion dynamic control airbag system provided in this application embodiment, the edge computing unit verifies whether a collision has occurred based on the vehicle's collision intensity and collision type, including: The edge computing unit determines the collision type of the vehicle based on point cloud data in the reflection information of the LiDAR against the obstacle; the collision type includes at least frontal collision, side collision and rear-end collision. The preset baseline acceleration trigger threshold is dynamically adjusted according to the collision type to obtain the target acceleration trigger threshold corresponding to each collision type; wherein, for side collisions, the adjustment is to decrease the baseline acceleration trigger threshold; for rear-end collisions, the adjustment is to increase the baseline acceleration trigger threshold. Based on the target acceleration trigger threshold corresponding to each collision state, determine whether a real collision has occurred.

[0090] Furthermore, in the multi-sensor fusion dynamic control airbag system provided in this application embodiment, the edge computing unit determines the collision type of the vehicle based on point cloud data in the reflection information of the LiDAR against the obstacle, including: The edge computing unit acquires point cloud data from the reflection information of the lidar in response to the obstacle; The point cloud data is fitted and analyzed to identify the collision contact area between the obstacle outline and the vehicle, and the actual collision angle of the collision contact area relative to the longitudinal axis of the vehicle is calculated. The actual collision angle is compared with multiple preset angle ranges to identify the collision type of the current collision.

[0091] Furthermore, in the multi-sensor fusion dynamic control airbag system provided in this application embodiment, the edge computing unit calculates the occupant risk assessment information inside the vehicle based on the echo signal of the millimeter-wave radar, including: Based on the echo signal of the millimeter-wave radar, a three-dimensional spatial model of the crew cabin is constructed and updated in real time. At least one occupant feature parameter is extracted from the three-dimensional spatial model, the feature parameter including: occupant head position, torso size, and sitting posture offset; Based on the extracted occupant characteristic parameters, risk assessment logic is executed; wherein, based on the extracted occupant characteristic parameters, the risk assessment logic includes: determining whether the occupant's head position is located within the preset danger zone of the airbag deployment path; determining whether the occupant is a child or a person with a special body type based on the torso size; and determining whether the occupant's sitting posture is abnormal based on the sitting posture offset. Based on the output of the risk assessment logic, corresponding occupant risk assessment information is generated.

[0092] Furthermore, in the multi-sensor fusion-based dynamic control airbag system provided in this application embodiment, the intelligent gas generator has a built-in high-speed inflation valve and a low-speed inflation valve; the dynamic control of the intelligent gas generator for the airbag inflation strategy based on the occupant risk assessment information includes: When a frontal collision is detected, the high-speed inflation valve is fully opened, allowing the airbag to inflate rapidly within 20 milliseconds. When a side collision is detected: first, the low-speed inflation valve is opened to pre-inflate to 30% of the target inflation volume; after the occupant begins to contact the airbag, the opening of the high-speed inflation valve is dynamically adjusted according to the real-time displacement of the occupant to complete the inflation. When a child seat is detected based on the risk assessment logic, the high-speed inflation valve is closed, and only the low-speed inflation valve is controlled to operate at a reduced inflation rate, reducing the total airbag inflation by 50%, while the airbag deployment action is delayed by 5 milliseconds.

[0093] Furthermore, in the multi-sensor fusion dynamic control airbag system provided in this application embodiment, the core of the edge computing unit is a field-programmable gate array (FPGA). The FPGA is configured to process in parallel the acceleration signal from the accelerometer, the reflection information from the lidar to the obstacle, and the echo signal from the millimeter-wave radar, so as to quickly complete the decision-making process from data acquisition to generating the control command.

[0094] Furthermore, in the multi-sensor fusion dynamic control airbag system provided in this application embodiment, the acceleration sensor is installed on the rigid structure of the vehicle body and configured to have a range covering ±200g; a resolution of not less than 0.01g; and a sampling frequency of not less than 10kHz; used to collect acceleration signals during vehicle collision. A lidar unit is installed on the inside of the vehicle's front bumper and configured to have a detection range covering 0-5 meters in front of the vehicle; a ranging accuracy of no less than ±1 cm; and a horizontal viewing angle covering at least 120° in front of the vehicle; used to detect in real time whether there are obstacles in front of the vehicle and the reflection information of the obstacles. The in-vehicle millimeter-wave radar is integrated into the reading light module in the vehicle's roof and is configured to: operate at a frequency of 77GHz; cover the front passenger compartment of the vehicle with a top-down detection angle; and transmit and receive millimeter-wave signals to construct a three-dimensional spatial model of the passenger compartment, thereby detecting the position, body shape, and sitting posture of the occupants.

[0095] Furthermore, in the multi-sensor fusion dynamic control airbag system provided in this application embodiment, the step of comparing the actual collision angle with multiple preset angle ranges to identify the collision type of the current collision includes: When the actual collision angle is within the range of 0°±15°, it is determined to be a frontal collision; When the actual collision angle is within the range of 70°±20°, it is determined to be a side collision; When the actual collision angle is within the range of 180°±15°, it is determined to be a rear-end collision.

[0096] This application completely solves the core problems of traditional airbag systems, such as high false trigger rate, long response delay, and rigid protection strategy, by using multi-sensor fusion perception, edge computing for rapid decision-making, and dynamic strategy for precise execution, thus achieving an intelligent breakthrough in vehicle passive safety.

[0097] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the method embodiments, and will not be repeated here. In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some communication interfaces; the indirect coupling or communication connection of devices or modules can be electrical, mechanical, or other forms.

[0098] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0099] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0100] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0101] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for dynamic control of airbags based on multi-sensor fusion, applied to a vehicle, wherein the vehicle is equipped with a multi-source data acquisition device, an edge computing unit, and a smart gas generator; The multi-source data acquisition device and the intelligent gas generator are respectively connected to the edge computing unit; The multi-source data acquisition device includes: an accelerometer, a lidar, and a millimeter-wave radar, characterized in that the method includes: The accelerometer collects the acceleration signal of the vehicle during a collision; the lidar detects whether there is an obstacle in front of the vehicle and the reflection information of the lidar towards the obstacle; the millimeter-wave radar emits millimeter-wave signals inside the vehicle and receives the echo signals from the millimeter-wave radar, so as to detect the occupant status data inside the vehicle by processing the echo signals. The edge computing unit calculates the collision intensity, collision type, and occupant risk assessment information of the vehicle based on the acceleration signal, the reflection information of the lidar against the obstacle, and the echo signal of the millimeter-wave radar. The edge computing unit verifies whether a real collision has occurred based on the collision intensity and collision type of the vehicle, and when a real collision is determined to have occurred, it dynamically controls the inflation strategy of the intelligent gas generator for the airbag based on the occupant risk assessment information.

2. The method for dynamic control of airbags based on multi-sensor fusion according to claim 1, characterized in that, The edge computing unit calculates the collision intensity of the vehicle based on the acceleration signal and the reflection information of the lidar towards the obstacle, including: The edge computing unit integrates the acceleration signal to obtain the speed change of the vehicle during the collision process; the speed change; The reflectivity data in the reflection information of the LiDAR against the obstacle is obtained, and the hardness category of the obstacle is determined based on the reflectivity data; The actual collision intensity of the vehicle is determined by combining the velocity change ΔV and the hardness category of the obstacle.

3. The method for dynamic control of airbags based on multi-sensor fusion according to claim 1, characterized in that, The edge computing unit verifies whether a collision has occurred based on the vehicle's collision intensity and collision type, including: The edge computing unit determines the collision type of the vehicle based on point cloud data in the reflection information of the LiDAR against the obstacle; the collision type includes at least frontal collision, side collision and rear-end collision. The preset baseline acceleration trigger threshold is dynamically adjusted according to the collision type to obtain the target acceleration trigger threshold corresponding to each collision type; wherein, for side collisions, the adjustment is to decrease the baseline acceleration trigger threshold; for rear-end collisions, the adjustment is to increase the baseline acceleration trigger threshold. Based on the target acceleration trigger threshold corresponding to each collision state, determine whether a real collision has occurred.

4. The method for dynamic control of airbags based on multi-sensor fusion according to claim 3, characterized in that, The edge computing unit determines the collision type of the vehicle based on point cloud data from the reflection information of the LiDAR against the obstacle, including: The edge computing unit acquires point cloud data from the reflection information of the lidar in response to the obstacle; The point cloud data is fitted and analyzed to identify the collision contact area between the obstacle outline and the vehicle, and the actual collision angle of the collision contact area relative to the longitudinal axis of the vehicle is calculated. The actual collision angle is compared with multiple preset angle ranges to identify the collision type of the current collision.

5. The method for dynamic control of airbags based on multi-sensor fusion according to claim 1, characterized in that, The edge computing unit calculates occupant risk assessment information within the vehicle based on the echo signal from the millimeter-wave radar, including: Based on the echo signal of the millimeter-wave radar, a three-dimensional spatial model of the crew cabin is constructed and updated in real time. At least one occupant feature parameter is extracted from the three-dimensional spatial model, the feature parameter including: occupant head position, torso size, and sitting posture offset; Based on the extracted occupant characteristic parameters, risk assessment logic is executed; wherein, based on the extracted occupant characteristic parameters, the risk assessment logic includes: determining whether the occupant's head position is located within the preset danger zone of the airbag deployment path; determining whether the occupant is a child or a person with a special body type based on the torso size; and determining whether the occupant's sitting posture is abnormal based on the sitting posture offset. Based on the output of the risk assessment logic, corresponding occupant risk assessment information is generated.

6. The method for dynamic control of airbags based on multi-sensor fusion according to claim 1, characterized in that, The intelligent gas generator has a built-in high-speed inflation valve and a low-speed inflation valve; the dynamic control of the intelligent gas generator's inflation strategy for the airbag based on the occupant risk assessment information includes: When a frontal collision is detected, the high-speed inflation valve is fully opened, allowing the airbag to inflate rapidly within 20 milliseconds. When a side collision is detected: first, the low-speed inflation valve is opened to pre-inflate to 30% of the target inflation volume; after the occupant begins to contact the airbag, the opening of the high-speed inflation valve is dynamically adjusted according to the real-time displacement of the occupant to complete the inflation. When a child seat is detected based on the risk assessment logic, the high-speed inflation valve is closed, and only the low-speed inflation valve is controlled to operate at a reduced inflation rate, reducing the total airbag inflation by 50%, while the airbag deployment action is delayed by 5 milliseconds.

7. The method for dynamic control of airbags based on multi-sensor fusion according to claim 1, characterized in that, The core of the edge computing unit is a field-programmable gate array (FPGA), which is configured to process in parallel the acceleration signal from the accelerometer, the reflection information from the lidar to the obstacle, and the echo signal from the millimeter-wave radar, so as to quickly complete the decision-making process from data acquisition to the generation of the control command.

8. The method for dynamic control of airbags based on multi-sensor fusion according to claim 2, characterized in that, The acceleration sensor is mounted on a rigid structure of the vehicle body and is configured to have a range covering ±200g; a resolution of not less than 0.01g; and a sampling frequency of not less than 10kHz; it is used to collect acceleration signals during a vehicle collision. A lidar unit is installed on the inside of the vehicle's front bumper and configured to have a detection range covering 0-5 meters in front of the vehicle; a ranging accuracy of no less than ±1 cm; and a horizontal viewing angle covering at least 120° in front of the vehicle; used to detect in real time whether there are obstacles in front of the vehicle and the reflection information of the obstacles. The in-vehicle millimeter-wave radar is integrated into the reading light module in the vehicle's roof and is configured to: operate at a frequency of 77GHz; cover the front passenger compartment of the vehicle with a top-down detection angle; and transmit and receive millimeter-wave signals to construct a three-dimensional spatial model of the passenger compartment, thereby detecting the position, body shape, and sitting posture of the occupants.

9. The method for dynamic control of airbags based on multi-sensor fusion according to claim 4, characterized in that, The step of comparing the actual collision angle with multiple preset angle ranges to identify the collision type of the current collision includes: When the actual collision angle is within the range of 0°±15°, it is determined to be a frontal collision; When the actual collision angle is within the range of 70°±20°, it is determined to be a side collision; When the actual collision angle is within the range of 180°±15°, it is determined to be a rear-end collision.

10. A system for dynamic control of airbags based on multi-sensor fusion, applied to a vehicle, wherein the vehicle is equipped with a multi-source data acquisition device, an edge computing unit, and an intelligent gas generator; The multi-source data acquisition device and the intelligent gas generator are respectively connected to the edge computing unit; The multi-source data acquisition device includes: an accelerometer, a lidar, and a millimeter-wave radar, characterized in that the system includes: The acceleration sensor is used to collect the acceleration signal of the vehicle during a collision; The lidar is used to detect whether there are obstacles in front of the vehicle and the reflection information of the lidar in response to the obstacles; The millimeter-wave radar is used to transmit millimeter-wave signals inside the vehicle and receive echo signals from the millimeter-wave radar, so as to detect occupant status data inside the vehicle by processing the echo signals. The edge computing unit is used to calculate the collision intensity, collision type, and occupant risk assessment information of the vehicle based on the acceleration signal, the reflection information of the lidar against the obstacle, and the echo signal of the millimeter-wave radar. The edge computing unit is used to verify whether a real collision has occurred based on the collision intensity and collision type of the vehicle, and when a real collision is determined to have occurred, to dynamically control the inflation strategy of the intelligent gas generator for the airbag based on the occupant risk assessment information.

Images