Control method, system, device, electronic equipment and vehicle of vehicle

Abstract

Embodiments of the present application provide a vehicle control method, system, device, electronic equipment and vehicle, wherein the method comprises: obtaining running state information of the vehicle; determining at least one risk information of the vehicle based on the running state information, wherein the risk information is used to represent a risk degree of the vehicle losing external support force of the wheel and / or the vehicle rolling over; determining a safety performance index of the vehicle based on the at least one risk information, wherein the safety performance index is used to represent a safety degree of the vehicle in the case of the vehicle losing external support force of the wheel and / or the vehicle rolling over; and controlling the vehicle to run according to a control strategy matched with the safety performance index, wherein the safety degree of the vehicle controlled according to the control strategy is greater than the safety degree of the vehicle to be controlled. The present application solves the technical problem of low control safety of the vehicle.

Description

Technical Field

[0001] This application relates to the field of vehicle electronic safety control technology, and more specifically, to a vehicle control method, system, device, electronic device, and vehicle. Background Technology

[0002] Currently, vehicle rollover detection systems mostly rely on a single Inertial Measurement Unit (IMU). However, this method suffers from response lag, triggering only after the vehicle tilt angle exceeds a threshold, and failing to predict early signs of ground lift such as wheel load transfer. Furthermore, the sensor is rigidly coupled to the airbag control unit (ACU), resulting in a rigid architecture that cannot adapt to the trend of centralized domain controllers, limiting computing power upgrades and multi-source fusion. In summary, these methods lead to low rollover detection accuracy and poor timeliness. Therefore, the technical problem of low vehicle control safety remains.

[0003] There is currently no good solution to the above problems. Summary of the Invention

[0004] This application provides a vehicle control method, system, device, electronic device, and vehicle to at least address the technical problem of low vehicle control safety.

[0005] According to one aspect of the embodiments of this application, a vehicle control method is provided, wherein the method may include: acquiring vehicle operating status information; determining at least one risk information of the vehicle based on the operating status information, wherein the risk information is used to characterize the risk level of the vehicle's wheels losing external support and / or the vehicle rolling over; determining a vehicle safety performance index based on the at least one risk information, wherein the safety performance index is used to characterize the vehicle's safety level under the condition that the wheels lose external support and / or the vehicle rolls over; controlling the vehicle operation according to a control strategy matched with the safety performance index, wherein the safety level of the vehicle controlled according to the control strategy is greater than the safety level of the vehicle to be controlled.

[0006] Furthermore, based on the operating status information, at least one risk information of the vehicle is determined, including: determining first risk information based on the wheel rotation speed information in the operating status information, wherein the first risk information is used to represent the degree of risk of the vehicle losing external support force; and / or, determining second risk information based on the three-axis information in the operating status information, wherein the three-axis information is used to represent the vehicle's motion posture in three-dimensional space and / or the interaction state between the vehicle and the magnetic field of three-dimensional space, and the second risk information is used to represent the degree of risk of the vehicle rolling over when the wheels lose external support force.

[0007] Furthermore, based on the wheel rotation speed information in the operating status information, the first risk information is determined, including: based on the rotation speed information, determining the wheel rotation speed difference characteristics, wherein the rotation speed difference characteristics are used to represent the degree of deviation of the wheel rotation speed abnormality caused by load transfer or separation; and based on the rotation speed difference characteristics, determining the first risk information.

[0008] Further, based on the rotational speed information, the rotational speed difference characteristics of the wheels are determined, including: determining the reference vehicle speed based on the rotational speed information of the non-driving wheels in the vehicle's wheel set; determining the theoretical rotational speed difference of the wheels based on the reference vehicle speed, the vehicle's steering angle and wheelbase in the operating status information; using the theoretical rotational speed difference to correct the difference between the rotational speed information and the reference vehicle speed, obtaining the corrected difference; determining the rotational speed difference characteristics based on the corrected difference, wherein the rotational speed difference characteristics and the first risk information are positively correlated; and / or, based on the rotational speed difference characteristics, the first risk information is determined, including: outputting the first risk information in response to the rotational speed difference characteristics exceeding the rotational speed difference characteristic threshold.

[0009] Furthermore, based on the three-axis information in the operating status information, the second risk information is determined, including: inputting the three-axis information into the vehicle's attitude calculation model, using the attitude calculation model to calculate the vehicle's roll angle and pitch angle from the three-axis information; and determining the second risk information based on the roll angle and pitch angle.

[0010] Furthermore, using the attitude calculation model, the vehicle's roll angle and pitch angle are calculated from the three-axis information. This includes: after inputting the three-axis information into the attitude calculation model, using the angular velocity measured by the gyroscope in the attitude calculation model, predicting the vehicle's current attitude using a quaternion integration strategy, and compensating for the angular velocity using the zero-bias estimate of the gyroscope to obtain the actual angular velocity; combining the gravity component measured by the accelerometer and the geomagnetic field component measured by the magnetometer in the attitude calculation model to correct the current attitude, obtaining the corrected current attitude; during the correction of the current attitude, monitoring the changes in the three-axis information, where the changes are used to characterize whether the vehicle is in a complex operating condition; adjusting the corrected current attitude based on the changes, obtaining the adjusted current attitude; and calculating the roll angle and pitch angle from the adjusted current attitude.

[0011] Furthermore, based on the roll angle and pitch angle, second risk information is determined, including: determining the degree of change of angular rate for each of the roll angle and pitch angle, wherein the degree of change of angular rate is used to characterize the severity of the risk of rollover or pitch of the vehicle; and determining the second risk information based on the degree of change of angular rate.

[0012] Furthermore, the method also includes: determining vehicle runaway information based on operational status information, wherein the runaway information is used to represent the degree of energy mutation of the vehicle; and determining vehicle driving conditions based on operational status information and ground status information; determining the weights corresponding to the runaway information and at least one risk information based on the driving conditions; and determining vehicle safety performance indicators based on at least one risk information, including: using the weights corresponding to the runaway information and at least one risk information to perform a weighted summation of the runaway information and at least one risk information to obtain the safety performance indicators.

[0013] Furthermore, the method further includes: in response to a safety performance index being greater than or equal to a first safety performance index threshold and a safety performance index being less than a second safety performance index threshold, determining a control strategy matching the safety performance index as a first control strategy, wherein the first control strategy is used to represent rules for controlling the seat belts and / or side curtain airbags in the vehicle; in response to a safety performance index being greater than or equal to the second safety performance index threshold, determining a control strategy matching the safety performance index as a second control strategy, wherein the safety level of the vehicle to be controlled according to the second control strategy is less than the safety level of the vehicle to be controlled according to the first control strategy, and the second control strategy is used to represent rules for controlling the airbags in the vehicle; in response to a safety performance index being less than the first safety performance index threshold, returning to the step of obtaining operating status information.

[0014] Furthermore, in accordance with the control strategy matching the safety performance indicators, the vehicle operation is controlled, including: in response to the control strategy being a first control strategy, performing a tightening operation on the seat belt, and / or, performing an inflation operation on the side curtain airbags; in response to the control strategy being a second control strategy, performing a detonation operation on the airbags.

[0015] Furthermore, the method also includes: after the tightening operation and / or inflation operation are completed, if the safety level of the vehicle operating according to the first control strategy decreases, the airbag is detonated according to the second control strategy.

[0016] According to another aspect of the embodiments of this application, a vehicle control system is also provided. The system may include: a sensing module for acquiring vehicle operating status information; a control and calculation unit for determining at least one risk information of the vehicle based on the operating status information, wherein the risk information characterizes the degree of risk of the vehicle's wheels losing external support and / or the vehicle rolling over; determining a vehicle safety performance index based on the at least one risk information, wherein the safety performance index characterizes the degree of safety of the vehicle under the condition that the wheels lose external support and / or the vehicle rolls over; and an execution module for controlling the vehicle operation according to a control strategy matched with the safety performance index, wherein the safety degree of the vehicle controlled according to the control strategy is greater than the safety degree of the vehicle to be controlled.

[0017] Furthermore, the system also includes a fault diagnosis module, which triggers a degradation control strategy in response to the detection that the operating state of the sensing module is in a fault state. The degradation control strategy is used to represent the rules for maintaining the safe operating state of the system.

[0018] According to another aspect of the embodiments of this application, a vehicle control device is also provided. The device may include: an acquisition module for acquiring vehicle operating status information; a first determination module for determining at least one risk information of the vehicle based on the operating status information, wherein the risk information is used to characterize the risk level of the vehicle's wheels losing external support and / or the risk of the vehicle rolling over; a second determination module for determining the vehicle's safety performance index based on at least one risk information, wherein the safety performance index is used to characterize the vehicle's safety level under the condition that the wheels lose external support and / or the risk of the vehicle rolling over; and a control module for controlling the vehicle's operation according to a control strategy matched with the safety performance index, wherein the safety level of the vehicle controlled according to the control strategy is greater than the safety level of the vehicle to be controlled.

[0019] According to another aspect of the embodiments of this application, a vehicle is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0020] According to another aspect of the embodiments of this application, an electronic device is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0021] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0022] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0023] According to another aspect of the embodiments of this application, a computer program product is also provided, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the methods in various embodiments of this application.

[0024] According to another aspect of the embodiments of this application, a computer program is also provided, which, when executed by a processor, implements the methods of the various embodiments of this application.

[0025] In this embodiment, if vehicle safety control is required, vehicle operating status information can be obtained. Based on this operating status information, at least one risk information of the vehicle can be determined. Based on this at least one risk information, the vehicle's safety performance indicators can be determined. The vehicle operation can be controlled according to the control strategy matched with the safety performance indicators. In other words, in this embodiment, by fusing risk information such as wheel lift-off signs and vehicle rollover risk, a dynamic safety performance indicator is constructed, overcoming the bottlenecks of reliance on a single IMU and response lag; and based on the safety performance indicator, a hierarchical control strategy is matched to achieve early warning of wheel loss of external support and coordinated response control, thereby significantly enhancing the timeliness and accuracy of rollover risk identification, thus achieving the technical effect of improving vehicle control safety and solving the technical problem of low vehicle control safety. Attached Figure Description

[0026] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0027] Figure 1 This is a schematic diagram illustrating an application scenario of vehicle control according to an embodiment of this application;

[0028] Figure 2 This is a flowchart of a vehicle control method according to an embodiment of this application;

[0029] Figure 3 This is a schematic diagram of a vehicle airborne and rollover recognition control system according to an embodiment of this application;

[0030] Figure 4 This is a flowchart of a method for vehicle airborne and rollover recognition and airbag control based on the fusion of nine-axis inertial information and wheel speed, according to an embodiment of this application.

[0031] Figure 5 This is a schematic diagram illustrating the principle of attitude calculation using adaptive Kalman filtering according to an embodiment of this application;

[0032] Figure 6 This is a flowchart of an adaptive adjustment logic decision-making method according to an embodiment of this application;

[0033] Figure 7 This is a schematic diagram of a graded ignition control state transition according to an embodiment of this application;

[0034] Figure 8 This is a flowchart of a sensor fault diagnosis and system degradation processing method according to an embodiment of this application;

[0035] Figure 9(a) is a schematic diagram of an integrated architecture according to an embodiment of this application;

[0036] Figure 9(b) is a schematic diagram of a distributed architecture according to an embodiment of this application;

[0037] Figure 9(c) is a schematic diagram of a hybrid architecture according to an embodiment of this application.

[0038] Figure 10 This is a schematic diagram of a vehicle control system according to an embodiment of this application;

[0039] Figure 11 This is a schematic diagram of a vehicle control device according to an embodiment of this application;

[0040] Figure 12 This is a schematic diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0041] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present application.

[0042] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0043] Figure 1 This is a schematic diagram illustrating an application scenario of vehicle control according to an embodiment of this application, such as... Figure 1 As shown, the scenario described above may include terminal device 10, network 20, and vehicle 30. Terminal device 10 can be used to obtain safety control commands from vehicle users (e.g., drivers, passengers) regarding whether safety control is required. The terminal device can be a mobile phone, computer, or other device used by the driver in the vehicle, or an interactive interface used for user interaction. The safety control commands can be sent to vehicle 30 via network 20. At this point, vehicle 30 needs to execute steps S102 to S108 to implement the vehicle control process.

[0044] The following steps can be performed by vehicle 30: Step S102, obtain the vehicle's operating status information; Step S104, determine at least one risk information of the vehicle based on the operating status information; Step S106, determine the vehicle's safety performance indicators based on at least one risk information; Step S108, control the vehicle's operation according to the control strategy matched with the safety performance indicators.

[0045] In this embodiment, through steps S102 to S108, if vehicle safety control is required, vehicle operating status information can be obtained. Based on this operating status information, at least one risk information of the vehicle can be determined. Based on this at least one risk information, the vehicle's safety performance indicators can be determined. The vehicle's operation can be controlled according to the control strategy matched with the safety performance indicators. In other words, in this embodiment, by fusing risk information such as wheel lift-off signs and vehicle rollover risks, a dynamic safety performance indicator is constructed, overcoming the bottleneck of reliance on a single IMU and response lag; and based on the safety performance indicators, a hierarchical control strategy is matched to achieve early warning of wheel loss of external support and coordinated response control, thereby significantly enhancing the timeliness and accuracy of rollover risk identification, thus achieving the technical effect of improving vehicle control safety and solving the technical problem of low vehicle control safety.

[0046] According to an embodiment of this application, a vehicle control method embodiment is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0047] This embodiment provides a vehicle control method. Figure 2 This is a flowchart of a vehicle control method according to an embodiment of this application, such as... Figure 2 As shown, the method may include the following steps.

[0048] Step S202: Obtain the vehicle's operating status information.

[0049] In the technical solution provided by step S202 of this application embodiment, the operating state information can refer to a multi-source sensor data set used to characterize the dynamic mechanical behavior of the vehicle and its interaction with the external environment during driving. This operating state information can serve as input for vehicle motion feature extraction and risk assessment, and can also be referred to as vehicle dynamic perception data or motion state data. The motion state information may include three-axis linear acceleration signals, three-axis angular velocity signals, and three-axis geomagnetic field strength signals acquired by a nine-axis IMU, as well as independent wheel speed signals acquired by four wheel-end speed sensors. These signals collectively constitute a comprehensive characterization of vehicle attitude changes, wheel load distribution, motion stability, and environmental interference, providing quantitative basis for identifying wheel detachment from ground support and vehicle rollover trends.

[0050] In this embodiment, if vehicle safety control is required, the vehicle's operational status information can be acquired. This embodiment achieves comprehensive perception of the vehicle's dynamic behavior and its interaction with the external environment through the synchronous collection and structured acquisition of vehicle operational status information and the collaborative work of multiple sensor sources. This provides high-precision, timely raw input data for subsequent risk assessment and safety control.

[0051] Optionally, during the acquisition of vehicle operating status information, dedicated sensors pre-installed in the vehicle structure can be used to collect and transmit data in real time. First, a nine-axis inertial measurement unit (IMU) is installed on the vehicle's rigid body platform. It integrates a three-axis accelerometer, a three-axis gyroscope, and a three-axis magnetometer, continuously outputting three-axis linear acceleration signals, three-axis angular velocity signals, and three-axis geomagnetic field strength signals. The linear acceleration signal reflects the overall translational state of the vehicle, the angular velocity signal characterizes the rotational motion of the vehicle body around each axis, and the geomagnetic field strength signal is used to assist in attitude calculation and magnetic interference identification. Second, independent wheel speed sensors are arranged at the ends of each of the four wheels. Using electromagnetic induction or the Hall effect principle, these sensors measure the instantaneous rotational speeds of the left front wheel, right front wheel, left rear wheel, and right rear wheel, outputting four wheel speed signals to characterize the contact state and load distribution between each wheel and the ground. All sensor signals, after being conditioned by a local circuit, are driven by a synchronous clock and uploaded to the control and computing unit via an onboard bus, such as CAN-FD or Ethernet, at a fixed sampling period (e.g., 2ms).

[0052] In this embodiment, the aforementioned method of acquiring operational status information ensures that the vehicle maintains a high degree of consistency and temporal synchronization in its data supply even under complex operating conditions. Through the synchronous acquisition of heterogeneous data from multiple sensors, a three-dimensional perception system covering vehicle attitude, motion dynamics, and tire-ground interaction is constructed. This lays a solid, reliable, and traceable data foundation for subsequent ground clearance prediction, attitude calculation, and risk fusion, significantly improving the system's early identification capability for airborne and rollover events and overall control safety.

[0053] Step S204: Based on the operating status information, determine at least one risk information of the vehicle.

[0054] In the technical solution provided by step S204 of the embodiments of this application, risk information can be used to characterize the risk level of a vehicle's wheels losing external support and / or the risk of the vehicle rolling over. The aforementioned risk information can refer to a computational feature quantity used to quantify the dynamic safety threat level caused by the wheels losing ground support or the vehicle body tending to roll over during driving. Essentially, it is a risk assessment output with physical meaning and engineering interpretability, generated based on the fusion of multi-source operating state information. The aforementioned external support force can refer to the vertical reaction force applied by the ground to the wheels, used to maintain the contact state between the wheels and the ground, and can directly reflect the wheel load distribution. When the external support force is significantly reduced or disappears, it indicates that the wheels have entered a state of airborne motion, an early sign of vehicle rollover. The aforementioned rollover risk can refer to the dynamic instability trend of the vehicle body continuously rotating around the longitudinal or lateral axis and losing its stable posture due to the accumulation of lateral or pitching motions. Essentially, it is a structural overturning threat caused by loss of vehicle body posture control. Both constitute the causal chain of a vehicle rollover accident: loss of external support force triggers airborne motion, and loss of posture control in the airborne state evolves into rollover risk. This application embodiment achieves full-chain risk identification from wheel-end load changes to vehicle body attitude instability by simultaneously sensing both.

[0055] The aforementioned risk information can also be referred to as ground clearance risk factors. For example, these ground clearance risk factors can include a first ground clearance risk factor and a second ground clearance risk factor. The first ground clearance risk factor can be generated from wheel speed signals after speed difference analysis and steering angle compensation. It is used to characterize early signs of wheels losing ground support due to a sudden drop or complete loss of vertical load, reflecting the failure trend of tire-road contact. The second ground clearance risk factor can be calculated from the real-time roll angle, pitch angle, and rate of change output by the adaptive attitude calculation model using data from a nine-axis inertial measurement unit. It is used to characterize the rollover evolution trend caused by inertial motion of the vehicle body, reflecting the dynamic degree of overall vehicle attitude instability. These two ground clearance risk factors can independently characterize the causes and processes of rollover accidents from the tire-ground interaction layer and the vehicle body rigid dynamics layer, respectively. They are complementary in terms of time dimension and physical mechanism, jointly constituting a multi-dimensional risk mapping of the entire process of vehicle airborne-rollover evolution, and can serve as direct inputs for safety performance indicators.

[0056] In this embodiment, after acquiring the vehicle's operating status information, at least one risk information of the vehicle can be determined based on this information. The core of this embodiment, which determines at least one risk information based on vehicle operating status information, lies in extracting early signs of wheels losing external support and the dynamic trend of vehicle rollover from multi-dimensional sensor data. This forms a ground-leaving risk factor with physical meaning and temporal complementarity, providing a high-precision and robust decision-making basis for subsequent risk fusion and safety control. The above method can distinguish between two modes: single risk factor and dual-factor collaborative output, dynamically selecting based on vehicle operating conditions and sensor availability to ensure accurate characterization of rollover risk across all scenarios.

[0057] Optionally, if the risk information is the first ground clearance risk factor, it can be analyzed collaboratively based on the four-wheel wheel speed signals and steering angle information. The deviation between each wheel speed and a reference vehicle speed is calculated. This reference vehicle speed is determined by the minimum wheel speed among the non-driving wheels to eliminate driving slip interference. Then, a steering angle compensation model is introduced to correct the difference in inner and outer wheel speeds caused by steering, obtaining the corrected wheel speed deviation. When the corrected deviations of two wheels on a specific side simultaneously exceed a dynamic threshold and the duration exceeds a first time window, it is determined that the wheel group on that side has lost external support, and the value of the first ground clearance risk factor increases, representing an early sign of wheel tip leaving the ground. This factor is sensitive to tire load transfer and responds faster than changes in vehicle attitude, providing a warning tens of milliseconds before the wheel leaves the ground.

[0058] Optionally, if the risk information is a second takeoff risk factor, the real-time roll and pitch angles can be calculated using the three-axis angular velocity, three-axis acceleration, and three-axis magnetic field strength signals output by the nine-axis inertial measurement unit through an adaptive extended Kalman filter model. This model dynamically adjusts the observation noise covariance matrix based on the magnitudes of angular velocity, linear acceleration, and magnetic field strength. Under stable operating conditions, it uses the gravitational field and geomagnetic field to correct gyroscope integral drift, and under severe roll or strong magnetic interference, it suppresses abnormal observations to ensure attitude angle accuracy. The second takeoff risk factor is generated by a weighted combination of roll angle amplitude, angular velocity, and angular acceleration, reflecting the vehicle's rotational trend around its longitudinal axis and characterizing the degree of rollover evolution, possessing high stability and full attitude coverage capability.

[0059] Optionally, if the risk information includes both a first ground-lift risk factor and a second ground-lift risk factor, both factors are activated simultaneously when the vehicle speed exceeds a preset threshold (e.g., 10 km / h), the wheel speed sensors are functioning normally, and the magnetic field environment is stable. In this case, the first ground-lift risk factor provides a leading signal for wheel liftoff, while the second ground-lift risk factor provides a confirmation signal for vehicle rollover. These two factors form a "warning-confirmation" two-level verification mechanism, effectively reducing the false alarm rate and shortening the response time. If a wheel speed sensor failure is detected or the vehicle is operating at low speed, the system can automatically switch to a single-factor mode relying solely on the second ground-lift risk factor. In this mode, attitude calculation is used to maintain the rollover recognition function, ensuring the system's basic safety capabilities under degraded conditions.

[0060] In this embodiment, the dynamic risk information determination mechanism achieves multi-source collaborative early warning under high vehicle speed and good sensor conditions, significantly improving the early identification rate of airborne and rollover events; it can still maintain reliable attitude-based judgment under sensor malfunction or low-speed conditions, ensuring system functional continuity and safety redundancy. By distinguishing between single and dual-factor output strategies, this step achieves accurate and adaptive characterization of rollover risk under different operating environments and fault modes, providing a solid, reliable, and hierarchical risk input foundation for graded ignition control.

[0061] Step S206: Determine the vehicle's safety performance indicators based on at least one piece of risk information.

[0062] In the technical solution provided by step S206 of this application embodiment, the safety performance index is used to characterize the vehicle's safety level when the wheels lose external support and / or the vehicle is at risk of rollover. The safety performance index can refer to a standardized assessment quantity used to comprehensively quantify the overall safety threat level of a vehicle in scenarios where the wheels lose external support and / or rollover risk occurs. Essentially, it is a unified risk measurement output generated after adaptively fusing multi-dimensional ground-leaving risk factors. The aforementioned safety performance index can also be called the airborne rollover risk index. The magnitude of the safety performance index corresponds to the urgency and severity of the vehicle's evolution towards a rollover accident.

[0063] In this embodiment, after determining at least one risk information of the vehicle based on operational status information, the vehicle's safety performance index can be determined based on this at least one risk information. This embodiment generates the vehicle's safety performance index through an adaptive weighted fusion mechanism based on at least one risk information. This index is a quantitative output that comprehensively characterizes the degree of safety threat corresponding to the risk of wheels losing external support and / or the vehicle body rolling over; it is called the airborne rollover risk index. This step dynamically selects a fusion strategy based on the composition of the risk information (single or dual factors), and automatically adjusts the weights by combining sensor confidence and vehicle operating conditions to ensure that the safety performance index has accuracy, stability, and functional safety under all operating conditions.

[0064] Optionally, if the risk information includes a first ground-lift risk factor, and the current scenario is determined to be that the wheel speed sensor is normal, the attitude calculation confidence is low, or the magnetic field environment is abnormal, then the safety performance index is directly generated by normalizing the first ground-lift risk factor. The first ground-lift risk factor reflects early signs of wheel group leaving the ground, and increases as the wheel speed deviation on a specific side continues to exceed the limit. The first ground-lift risk factor is linearly or nonlinearly mapped to a standardized range of 0 to 1 as the airborne rollover risk index. If the risk information only includes a second ground-lift risk factor, and the current scenario is determined to be that the wheel speed signal is malfunctioning or the vehicle speed is below a preset threshold, then the safety performance index is calculated from the second ground-lift risk factor. This second ground-lift risk factor is based on the precise roll and pitch angles output by the adaptive Kalman filter, combined with the rate of change and energy accumulation trend of the second ground-lift risk factor, reflecting the vehicle rollover evolution trend. The second ground-lift risk factor is output as the airborne rollover risk index after nonlinear weighted combination, used to maintain the basic rollover recognition capability.

[0065] Optionally, if the risk information includes both a first and a second takeoff risk factor, a dual-factor collaborative evaluation mode is entered. Safety performance indicators are calculated using a weighted fusion formula. For example, the airborne rollover risk index is equal to the algebraic sum of the first and second takeoff risk factors multiplied by their respective adaptive weighting coefficients. The weighting coefficients are dynamically adjusted based on the trace of the attitude covariance matrix, vehicle speed, steering state, and magnetic field stability. When the attitude estimation confidence is high, the weight of the second takeoff risk factor is increased, strengthening the ability to confirm the rollover trend; when wheel speed is significantly abnormal and attitude uncertainty increases, the weight of the first takeoff risk factor is increased, prioritizing responses to early signs of wheel takeoff. This fusion process does not rely on fixed thresholds but achieves adaptive allocation of risk weights through real-time confidence feedback, ensuring that the output safety performance indicators possess both sensitivity and anti-interference capabilities under complex operating conditions.

[0066] In this embodiment, the construction method of the aforementioned safety performance indicators achieves a paradigm shift from single-source sensing to multi-source collaboration and from fixed thresholds to dynamic evaluation. In single-factor mode, the system's functional continuity is ensured under sensor failure or low-speed scenarios; in two-factor mode, "early warning-confirmation" collaborative verification is achieved, significantly improving the timeliness, accuracy, and resistance to false triggering of rollover risk identification. By deeply binding risk information with the system's operating status, safety performance indicators become core decision variables with both physical interpretability and engineering robustness, providing a scientific, reliable, and traceable quantitative basis for graded ignition strategies.

[0067] Step S208: Control the vehicle operation according to the control strategy matched with the safety performance indicators.

[0068] In the technical solution provided by step S208 of the embodiments of this application, the safety level of the vehicle controlled according to the control strategy is greater than the safety level of the vehicle to be controlled. The control strategy can refer to a set of logical rules that dynamically match and trigger a series of graded passive safety actions based on the numerical levels of safety performance indicators. Essentially, it maps quantified risk into a safety response sequence with temporal and hierarchical characteristics. This sequence is used to proactively enhance the protective effectiveness of the occupant restraint system in critical states where the vehicle's wheels lose external support or there is a risk of rollover, thereby achieving a significant improvement in safety.

[0069] The aforementioned control strategy does not employ a single, fixed ignition threshold. Instead, it can sequentially activate multiple levels of actions, such as seatbelt pretensioning, side curtain airbag pre-charging, and airbag ignition and deployment, based on the preset grading range of the safety performance indicators. Each level of action corresponds to a specific stage of risk evolution. For example, Level 1 control activates the seatbelt pretensioner when the risk index reaches the warning threshold, eliminating webbing slack and allowing the occupant to be in contact with the seat in advance. Level 2 control triggers the pre-charging circuit of the side curtain airbag generator when the risk index continues to rise and meets the time-delay condition, enabling the airbag to complete gas storage before deflagration and shortening the inflation delay from the usual 30ms to less than 10ms. Level 3 control executes the final deployment of the side curtain airbags, side airbags, and rollover protection airbags when the risk index exceeds the ignition threshold, constructing a complete three-dimensional occupant protection space.

[0070] In this embodiment, after determining the safety performance indicators based on at least one risk information, the vehicle operation can be controlled according to the control strategy matched with the aforementioned safety performance indicators. The core of this embodiment, which matches and executes corresponding control strategies based on safety performance indicators, lies in establishing a precise mapping relationship between risk levels and safe execution actions. Through multi-level, orderly, and time-controllable passive safety interventions, the response efficiency and protective integrity of the occupant restraint system are proactively improved during the evolution of the vehicle's risk of losing external wheel support or rolling over. Ultimately, the safety level of the vehicle after implementing control is significantly better than the original state without this strategy.

[0071] Optionally, a control strategy matching the safety performance indicators is determined, and the airborne rollover risk index is compared in real time with a preset graded ignition threshold matrix. This threshold matrix contains at least two safety threshold intervals: the first threshold interval corresponds to the warning level, and the second threshold interval corresponds to the ignition level. When the safety performance indicator is in the first threshold interval, the system determines that the risk is in the early evolution stage, and the matching control strategy is seat belt pretensioning and side curtain airbag pre-charging; when the safety performance indicator continuously exceeds the second threshold interval and meets the minimum duration criterion (e.g., 3ms), the system determines that the risk has entered the irreversible development stage, and the matching control strategy is to detonate the side curtain airbags, side airbags, and rollover protection airbags. The threshold setting is not fixed, but dynamically fine-tuned according to the current vehicle operating conditions: if the weight of the first ground-leaning risk factor is high, it indicates that the wheel ground-leaning characteristics are significant, and the threshold can be appropriately relaxed for a faster response; if the weight of the second ground-leaning risk factor is high and the attitude covariance is large, it indicates that the attitude information uncertainty is high, and the threshold is tightened accordingly to reduce the risk of false triggering. The selection of control strategies is entirely determined by the numerical values ​​of safety performance indicators and their time evolution trends, and precise matching of strategies is achieved through table lookup or analytical functions.

[0072] Optionally, vehicle operation is controlled based on a matching control strategy by sending structured instructions from the control and computing unit to the execution module. When the matching strategy is seatbelt pretensioning and side curtain airbag pre-charging, the control and computing unit activates the ignition circuit of the seatbelt pretensioner, enabling the pretensioner to tighten the webbing within tens of milliseconds, and simultaneously starts the capacitor charging or propellant preheating mechanism of the side curtain airbag generator to prepare energy reserves for subsequent airbag inflation. When the matching strategy is upgraded to airbag detonation, the control and computing unit triggers the ignition circuits of all associated airbag generators, enabling the side curtain airbags and side airbags to generate and expand gas in a very short time, forming a three-dimensional buffer space surrounding the occupant's head and torso. All execution actions are transmitted via hardwired or high-reliability bus commands to ensure uninterrupted communication under extreme conditions. The control process does not rely on manual intervention and is entirely autonomously decided by the system, with a secondary state confirmation performed before execution to prevent false triggering due to momentary interference. After all actions are executed, the system records the action type, trigger time, and safety performance index values ​​for subsequent fault analysis and system optimization.

[0073] In this embodiment, the above-described method achieves a paradigm shift from passive response to active protection. By establishing a precise hierarchical mapping between safety performance indicators and multi-level execution actions, the system initiates pretensioning and pre-charging before the risk reaches a lethal level, significantly improving occupant restraint efficiency and airbag response speed. Final detonation is then executed during the risk-deterioration phase, avoiding protective failure caused by premature or delayed ignition. This strategy not only significantly reduces head and torso injuries in rollover accidents but also reduces unnecessary airbag deployments through step-by-step execution, improving overall efficiency and functional safety. Ultimately, vehicles operating according to this control strategy demonstrate significantly better occupant protection in airborne and rollover scenarios than traditional systems relying solely on a single threshold trigger, achieving a substantial and quantifiable improvement in safety.

[0074] In steps S202 to S208 of this embodiment, if vehicle safety control is required, vehicle operating status information can be obtained. At least one risk information of the vehicle can be determined based on this operating status information. The vehicle's safety performance indicators can be determined based on this at least one risk information. The vehicle's operation can be controlled according to the control strategy matched by the safety performance indicators. In other words, in this embodiment, by fusing risk information such as wheel lift-off signs and vehicle rollover risks, a dynamic safety performance indicator is constructed, overcoming the bottleneck of reliance on a single IMU and response lag; and based on the safety performance indicators, a hierarchical control strategy is matched to achieve early warning from wheel loss of external support to coordinated response control, thereby significantly enhancing the timeliness and accuracy of rollover risk identification, and thus achieving the technical effect of improving vehicle control safety, solving the technical problem of low vehicle control safety.

[0075] The embodiments of this application will be described in detail below with reference to the steps described above.

[0076] As an optional implementation, step S204 involves determining at least one risk information for the vehicle based on the operating status information, including: determining first risk information based on the wheel rotation speed information in the operating status information, wherein the first risk information is used to represent the degree of risk of the vehicle losing external support; and / or determining second risk information based on the three-axis information in the operating status information, wherein the three-axis information is used to represent the vehicle's motion posture in three-dimensional space and / or the interaction state between the vehicle and the magnetic field in three-dimensional space, and the second risk information is used to represent the degree of risk of the vehicle rolling over when the wheels lose external support.

[0077] In this embodiment, during the process of determining at least one risk information based on the operating status information, the first risk information can be determined based on the rotational speed information in the operating status information. Alternatively, the second risk information can be determined based on the three-axis information in the operating status information. Here, the rotational speed information refers to a sequence of digital signals characterizing the instantaneous angular velocity of each wheel, which is collected and output in real time by wheel speed sensors installed at the wheel ends of the four wheels. Physically, it is a discrete sampling expression of the linear velocity at the wheel-ground contact point in the time dimension. This rotational speed information can reflect the dynamic mechanical interaction between the tire and the road surface and can be used to identify early signs of abnormal wheel load distribution, loss of vertical support, or wheel detachment from the ground. In this embodiment, the rotational speed information can refer to four independent wheel speed signals from the left front wheel, right front wheel, left rear wheel, and right rear wheel of the vehicle. The numerical changes are not affected by the vehicle's attitude, and it has high temporal resolution and strong transient response characteristics, serving as a sensing basis for determining whether the wheel has lost external support.

[0078] Optionally, the aforementioned first risk information can refer to a characteristic quantity calculated based on rotational speed information after rotational speed deviation analysis and steering angle compensation, used to quantify the degree of risk of wheel loss of external support force. Its physical meaning is a dynamic trend index of tire-ground contact failure. This first risk information does not directly measure force values, but rather compares the deviation of each wheel's rotational speed information from the reference vehicle speed, combines this with dynamic threshold correction based on steering conditions, and identifies a coordinated trend of significant rotational speed anomalies simultaneously occurring on a specific side of the wheels, thereby inferring a concentrated loss of vertical load. The value of the first risk information increases monotonically with the increasing tendency of the wheel to leave the ground, exhibiting a response speed superior to changes in vehicle attitude, and can provide a warning of airborne events tens of milliseconds in advance. This first risk information can also be referred to as the first airborne risk factor or the wheel-end airborne sign index.

[0079] Optionally, the aforementioned triaxial information can refer to a composite sensor dataset collected and output by a nine-axis inertial measurement unit, containing triaxial acceleration signals, triaxial angular velocity signals, and triaxial magnetic field strength signals. Physically, it represents a full-dimensional expression of the interaction between the vehicle's rigid body motion state and the local geomagnetic environment in three-dimensional space. Specifically, the triaxial acceleration signals reflect the vehicle's overall linear acceleration (including the gravitational component). The triaxial angular velocity signals characterize the vehicle's rotational speed around the coordinate axes. The triaxial magnetic field strength signals reflect the change in the geomagnetic field vector at the vehicle's location, used to identify external magnetic disturbances and the stability of the attitude reference.

[0080] Optionally, the aforementioned second risk information can refer to a feature quantity generated based on three-axis information through adaptive extended Kalman filtering attitude calculation. This feature is used to quantify the degree of rollover risk when the wheels lose external support. Physically, it represents the dynamic evolution intensity of vehicle attitude instability and rotational accumulation. This second risk information can be calculated by nonlinear weighted fusion of the calculated real-time roll angle, pitch angle, and their first and second derivatives (angular velocity and angular acceleration). It reflects the trend of continuous rollover around the longitudinal or lateral axis, and its value increases with increasing rollover angular velocity and the greater the deviation of the angle from the horizontal plane. The second risk information has strong resistance to instantaneous interference and accurate long-term drift compensation, making it a core basis for determining whether rollover has entered an irreversible stage. This second risk information can also be called the second ground-lift risk factor or attitude rollover evolution index.

[0081] In this embodiment, determining at least one risk information for the vehicle based on operational status information is a crucial preliminary step for accurately identifying vehicle airborne and rollover risks. Through a dual-path parallel or optimal mechanism, two independent but complementary physical signs are extracted from the wheel-ground interaction layer and the vehicle body rigid dynamics layer, respectively, generating first and second risk information. This provides a highly reliable and timely input source for the subsequent construction of safety performance indicators. The above process dynamically selects single-path or dual-path output based on the vehicle's sensor availability and operating condition characteristics, ensuring the completeness and robustness of risk perception across all scenarios.

[0082] Optionally, in determining the first risk information based on wheel speed information from the operating status information, wheel speed signals of the four wheels can be collected to form a complete dataset of speed information. A reference vehicle speed is calculated, where the reference speed is taken from the minimum wheel speed value among the non-driving wheels to eliminate abnormal wheel speed interference caused by drive slippage or power distribution. Based on this, the deviation value between each wheel speed and the reference vehicle speed is calculated, and vehicle steering angle information is introduced. The theoretical wheel speed difference caused by turning is compensated through a preset geometric model to obtain the corrected wheel speed deviation. When the corrected deviation of two wheels on a specific side (left or right) simultaneously exceeds the dynamically set threshold for a duration exceeding the first time window, the system determines that the wheel group on that side has lost cooperative load, and the value of the first risk information increases accordingly, representing an early sign of wheel detachment from ground support. The above process can achieve early warning without relying on vehicle body posture, but can be achieved through changes in wheel end motion state, with the characteristics of fast response and strong resistance to vibration interference.

[0083] Optionally, during the process of determining the second risk information based on the three-axis information in the operational status information, three-axis acceleration signals, three-axis angular velocity signals, and three-axis magnetic field strength signals can be collected simultaneously to form a complete three-axis information set. Using an adaptive extended Kalman filter model, with the angular velocity signal as the state prediction input and the gravity component measured by the accelerometer and the geomagnetic field component measured by the magnetometer as the observation correction input, the roll and pitch angles of the vehicle body in the navigation coordinate system are calculated in real time. The aforementioned adaptive extended Kalman filter model dynamically adjusts the observation noise covariance matrix according to the magnitudes of angular velocity, linear acceleration, and magnetic field strength. During smooth driving, it maximizes the use of gravity and geomagnetic references to correct gyroscope drift, and suppresses abnormal observations during severe rollover or strong magnetic interference, ensuring long-term stability of the attitude angles. The second risk information is generated by a nonlinear weighted combination of the calculated roll angle, pitch angle, angular velocity, and angular acceleration, reflecting the evolution trend of the vehicle body continuously rotating around the longitudinal or transverse axis in an airborne state, characterizing the dynamic intensity of the rollover risk.

[0084] In this embodiment, the method described above constructs a dual-channel risk perception mechanism for wheel-end lift-off prediction and attitude rollover confirmation through parallel processing of rotation speed information and triaxial information. The first risk information captures the initial signal of wheel lift-off with a millimeter-level response speed, gaining crucial early warning time; the second risk information confirms the actual process of rollover evolution through highly stable attitude calculation, avoiding false triggering caused by misjudgment from a single sensor. The two can operate independently, ensuring the system still possesses basic rollover recognition capabilities even when wheel speed sensors fail or the magnetic field environment is abnormal; they can also be fused collaboratively to achieve dual protection of early warning and accurate confirmation. This process significantly improves the reliability of recognition under complex operating conditions (e.g., bumpy roads, strong magnetic interference, low-speed driving), providing high-quality risk input with physical interpretability, time synchronization, and functional safety redundancy for subsequent hierarchical control strategies, fundamentally overcoming the inherent defects of traditional single-sensor response lag and attitude drift.

[0085] As an optional implementation, determining the first risk information based on the wheel rotation speed information in the operating status information includes: determining the wheel rotation speed difference characteristics based on the rotation speed information, wherein the rotation speed difference characteristics are used to represent the degree of deviation of the wheel rotation speed abnormality caused by load transfer or separation; and determining the first risk information based on the rotation speed difference characteristics.

[0086] In this embodiment, during the process of determining the first risk information based on rotational speed information, the rotational speed difference characteristics of the wheels can be determined based on the rotational speed information. The first risk information can be determined based on these rotational speed difference characteristics. These rotational speed difference characteristics refer to a set of quantitative indicators characterizing the degree of deviation of each wheel's rotational speed from the vehicle's baseline state, obtained by calculating the difference between the wheel speed signals of the four wheels and a reference vehicle speed. Physically, they represent the relative difference in dynamic rotational speed caused by vertical load transfer, partial ground contact, or momentary airborne movement of the wheels. These rotational speed difference characteristics can indirectly infer changes in the stability of the wheel's contact with the ground through relative changes in wheel speed signals, and are a sensitive and direct kinematic sign of the loss of external support for the wheel group. For example, these rotational speed difference characteristics can include the first deviation values ​​between the left front wheel, right front wheel, left rear wheel, and right rear wheel and the reference vehicle speed. During the calculation process, the vehicle steering angle is introduced for dynamic compensation to eliminate theoretical wheel speed differences caused by differences in wheelbase and turning radius during normal turning, ensuring that the retained deviations only reflect abnormal rotational speed anomalies caused by load loss or airborne movement. The aforementioned speed characteristic difference can also be called the wheel end load imbalance index. The value increases as a specific side wheel simultaneously exhibits a significant positive or negative speed deviation, characterizing the coordinated trend of the vertical force of the tire on that side decreasing or losing contact.

[0087] In this embodiment, determining the first risk information based on the wheel rotation speed information in the operational status information is the core step in achieving early identification of vehicle liftoff risk. Through a hierarchical processing mechanism, rotation speed difference features characterizing abnormal wheel loads are extracted from the rotation speed information (raw wheel speed signals). Then, based on the spatial synergy and temporal persistence of the rotation speed difference features, first risk information that quantifies the degree of risk of the wheel losing external support is constructed. This process does not rely on a vehicle dynamics model but achieves high-sensitivity, low-latency perception of the tire's tendency to leave the ground through the relative changes in the wheel-end motion state.

[0088] Optionally, during the process of determining the speed difference characteristics based on speed information, wheel speed signals of all four wheels can be collected simultaneously to form complete speed information. The minimum wheel speed value among the non-driving wheels is used as the reference vehicle speed to avoid non-true speed deviations caused by drive slippage, torque distribution, or differential action. The difference between the current wheel speed of each wheel and the reference vehicle speed is calculated to generate four sets of first deviation values, which constitute the basic data units of the speed difference characteristics. To eliminate wheel speed differences caused by normal steering conditions, the vehicle's steering angle signal can be introduced. Combined with the vehicle's wheelbase and geometric model, the expected wheel speed difference caused by theoretical steering is calculated, and this expected wheel speed difference is subtracted from the first deviation values ​​of each wheel to obtain the corrected speed difference characteristics. Through the above correction process, it can be ensured that the speed difference characteristics reflect asymmetric speed changes caused by abnormal vertical loads or wheel lift-off, rather than those caused by conventional handling behavior, thereby improving the physical specificity and environmental robustness of the characteristics. The speed difference characteristics are ultimately expressed as four steering-compensated wheel speed deviation values, which are a direct kinematic expression of wheel load state changes.

[0089] Optionally, in determining the first risk information based on the speed difference characteristics, the corrected speed difference characteristics can be analyzed using both spatial coordination and temporal persistence criteria. It monitors whether the corrected speed difference characteristics of the two wheels on a specific side (left or right) simultaneously exceed a dynamically set threshold. This dynamically set threshold is adaptively adjusted according to vehicle speed and road conditions to accommodate normal fluctuations under different driving environments. When the speed difference characteristics of both wheels on a specific side continuously exceed the threshold, and the duration exceeds a preset first time window (e.g., 50ms), it is determined that the wheel group on that side has experienced a loss of coordinated load and has a tendency to lift off the ground. At this time, the numerical value of the first risk information is increased, and its magnitude is non-linearly positively correlated with the deviation amplitude, duration, and synchronicity of both sides exceeding the dynamically set threshold. If a single wheel is abnormal or both wheels are asynchronous, it is determined to be local slippage or sensor disturbance, and the first risk information remains at a low value. The water-filling process achieves a transition from "single-point anomaly" to "group coordinated instability," ensuring that the first risk information is only activated when there is a genuine risk of rollover, effectively suppressing false triggering.

[0090] In this embodiment, a wheel liftoff prediction mechanism independent of vehicle body posture is constructed through refined processing of the aforementioned rotation speed information and collaborative discrimination of rotation speed difference features. This method breaks through the traditional lag recognition mode that relies on vehicle body roll angle, and can complete risk identification in the early stage, before the wheels are fully airborne and the vehicle body is significantly tilted, by using collaborative features of abnormal wheel end rotation speeds, significantly advancing the warning time window. Simultaneously, through steering angle compensation and dynamic threshold mechanisms, interference caused by normal turning and uneven road surfaces is effectively filtered out, ensuring that the first risk information only responds to real load loss events. This process achieves high sensitivity and high specificity in sensing airborne risks, providing reliable, timely, and low-false-alarm front-end input for the construction of subsequent safety performance indicators, significantly improving the system's active protection capabilities in typical rollover scenarios such as high-speed airborne events and slope run-off.

[0091] As an optional implementation, determining the wheel speed difference characteristics based on rotational speed information includes: determining the vehicle's reference speed based on the rotational speed information of the non-driving wheels in the vehicle's wheel set; determining the theoretical wheel speed difference based on the reference speed, the vehicle's steering angle and wheelbase in the operating status information; correcting the difference between the rotational speed information and the reference speed using the theoretical speed difference to obtain a corrected difference; determining the rotational speed difference characteristics based on the corrected difference, wherein the rotational speed difference characteristics and the first risk information are positively correlated; and / or, determining the first risk information based on the rotational speed difference characteristics includes: outputting the first risk information in response to the rotational speed difference characteristics exceeding a rotational speed difference characteristic threshold.

[0092] In this embodiment, during the process of determining the wheel speed difference characteristics based on rotational speed information, the vehicle's reference speed can be determined based on the rotational speed information of the non-driving wheels in the vehicle's wheel set. The theoretical speed difference can be determined based on the aforementioned reference speed, the vehicle's steering angle, and wheelbase from the operating status information. The theoretical speed difference can be used to correct the difference between the rotational speed information and the reference speed, resulting in a corrected difference. This corrected difference can be defined as the speed difference characteristic. During the process of determining the first risk information based on the speed difference characteristic, if the speed difference characteristic exceeds a speed difference characteristic threshold, the first risk information can be output.

[0093] The reference speed refers to a benchmark value calculated from the minimum rotational speed signals of the non-driving wheels, based on the concentrated wheel speed signals of the vehicle's wheels. It represents the actual forward speed of the ground-contacting wheels under the current driving conditions. Physically, it is a motion reference benchmark that approximates the vehicle's true speed, unaffected by drive slippage or power distribution interference. This reference speed can be used to eliminate misjudgments of rotational speed deviations caused by drive wheel slippage, torque distribution differences, or differential action, ensuring that subsequent rotational speed difference analysis only reflects wheel speed anomalies caused by vertical load transfer or ground clearance. The reference speed can also be called the minimum non-driving wheel speed benchmark.

[0094] Optionally, the aforementioned steering angle can refer to the real-time measured value output by the vehicle's steering system, characterizing the deflection angle of the front wheels relative to the vehicle's longitudinal axis. Physically, it represents the change in curvature of the vehicle's current trajectory, used to quantify the intensity of the vehicle's ongoing steering motion. During a turn, due to the different radii of motion of the inner and outer wheels, even if the vehicle moves at a constant speed, the rotational speeds of the four wheels will theoretically differ. The steering angle serves to distinguish between the wheel speed difference caused by normal steering and the abnormal rotational speed difference caused by the wheels leaving the ground. The aforementioned wheelbase can refer to the straight-line distance between the center of the front axle and the center of the rear axle.

[0095] Optionally, the theoretical speed difference can refer to the theoretical speed difference that the left and right wheels should produce due to the different radii of their travel trajectories under ideal, slip-free conditions, calculated using a vehicle steering kinematics model based on the steering angle, reference vehicle speed, and wheelbase. This theoretical speed difference can be used to predict the expected speed difference between the left and right wheels during normal steering. The difference between the aforementioned speed information and the reference vehicle speed can refer to the direct arithmetic difference between the real-time wheel speed signal of each wheel and the reference vehicle speed. Physically, it represents the absolute deviation of that wheel from the reference vehicle speed, without considering the theoretical differences caused by steering. This difference can include two parts: one part is the theoretical wheel speed difference caused by normal steering, and the other part is the asymmetric speed anomaly caused by reduced wheel load or wheel detachment from the ground.

[0096] Optionally, the corrected difference can refer to the residual deviation obtained by subtracting the theoretical speed difference from the difference between the speed information and the reference vehicle speed. Physically, it represents the abnormal speed component caused by changes in wheel vertical load or wheel detachment from the ground after eliminating normal steering effects. The corrected value can reflect unexpected motion state changes caused by asymmetrical load distribution or tire contact loss, characterizing the kinematic features of the lift-off trend. The aforementioned speed difference characteristic threshold can be a dynamic critical value set to determine whether a wheel has entered a lift-off risk state. It can be adaptively adjusted based on vehicle speed, road conditions, and system confidence to distinguish between normal fluctuations and genuine anomalies. The aforementioned speed difference characteristic threshold can be a function that varies with operating conditions, ensuring no false triggering at low speeds, on bumpy or slippery surfaces, while possessing sufficient sensitivity at high speeds, on unpaved surfaces, or in airborne scenarios. If the corrected wheel speed deviation on a specific side exceeds the aforementioned speed difference characteristic threshold, and the duration meets the time window requirements, it can be determined that the wheel group on that side has a coordinated lift-off trend, thereby activating the first risk information.

[0097] In this embodiment, determining the wheel speed difference characteristics based on rotational speed information and further generating first risk information is the core process for achieving early perception of vehicle liftoff risk. This process can extract high signal-to-noise ratio features from the original wheel speed signal that clearly characterize the wheel's tendency to leave the ground, thus accurately identifying the loss of external support for the wheels before the vehicle's posture changes significantly. The entire process uses non-drive wheels as a reference, steering geometry as compensation, and dynamic thresholds as the decision-making mechanism, ensuring that the speed difference characteristics only respond to real liftoff events, avoiding misjudgments caused by steering interference and road surface disturbances, and providing highly timely and reliable risk input for subsequent safety control.

[0098] Optionally, in determining the reference vehicle speed based on the rotational speed information of the non-driving wheels, wheel speed signals of the left front wheel, right front wheel, left rear wheel, and right rear wheel can be collected, and sensor data belonging to the non-driving wheels can be identified. In four-wheel drive or all-wheel drive vehicles, non-driving wheels can refer to the wheels on the rear or front axle that do not receive power output; in rear-wheel drive or front-wheel drive vehicles, they are the two wheels on the non-driving axle. The minimum value among the wheel speeds of the aforementioned non-driving wheels is selected as the reference vehicle speed, which is defined as the benchmark quantity that is closest to the actual relative speed to the ground under the current driving condition. Choosing the minimum value instead of the average value is to avoid artificially high wheel speeds caused by drive wheel slippage, differential slippage, or uneven power distribution, ensuring that the reference vehicle speed truly reflects the ground contact state, thereby providing a stable and physically consistent reference benchmark for subsequent deviation calculations.

[0099] For example, the reference speed can be determined using the following formula:

[0100]

[0101] in, It can be used to indicate reference vehicle speed; It can be used to represent the wheel speed signal (rotation information) of the left non-drive wheel, that is, the instantaneous rotational speed value collected by the wheel speed sensor installed on the left non-drive axle of the vehicle (such as the rear wheel of a front-wheel drive vehicle and the front wheel of a rear-wheel drive vehicle). It can be used to represent the wheel speed signal (rotation information) of the right non-drive wheel, that is, the instantaneous rotational speed value collected by the wheel speed sensor installed on the right non-drive axle of the vehicle; It can be used to represent the operation of taking the minimum value between the wheel speed signal of the left non-driving wheel and the wheel speed signal of the right non-driving wheel.

[0102] Optionally, in the process of determining the theoretical speed difference based on reference vehicle speed, vehicle steering angle, and wheelbase, a kinematic compensation model is established to eliminate wheel speed differences caused by normal steering. Using the vehicle steering angle and wheelbase as input parameters, the theoretical speed difference between the left and right wheels due to their different trajectory radii during turning is calculated according to planar kinematics formulas. This theoretical speed difference is a purely geometric derivation, dependent on vehicle structure and control input, and does not involve tire slippage or load changes. Its purpose is to quantify the expected wheel speed distribution under ideal no-slip, no-ground-lift conditions, providing a theoretical basis for subsequently removing predictable components from the original wheel speed difference, ensuring that subsequent analysis focuses only on abnormal motion anomalies.

[0103] For example, the theoretical speed difference can be determined using the following formula:

[0104]

[0105] in, It can be used to represent the theoretical speed difference, which can be the theoretical difference in wheel speed that should exist in the left and right wheels when the vehicle is in a pure steering motion state, due to the different radii of the travel trajectories of the left and right wheels, calculated based on the vehicle's geometry and kinematics. It can be used to represent the steering angle of a vehicle; It can be used to indicate wheelbase.

[0106] Optionally, in the process of correcting the difference between the rotational speed information and the reference vehicle speed using the theoretical rotational speed difference, accurate extraction of rotational speed difference characteristics can be achieved. The original difference between the wheel speed of each wheel and the reference vehicle speed can be calculated. This difference can include two parts: one part is the theoretical difference caused by normal steering, and the other part is the abnormal deviation caused by the transfer of vertical load on the wheel or its removal from the ground. Subtracting the theoretical rotational speed difference from the original difference, retaining only the residual deviation, is the corrected difference. This correction process is essentially a "de-steering" process, making the rotational speed difference characteristics no longer affected by steering wheel operation, reflecting the asymmetric motion anomalies caused by changes in tire vertical force or removal from the ground. The corrected difference is a direct component of the rotational speed difference characteristics; the further its value deviates from zero, the more significant the loss of support force of that wheel or the wheel on that side.

[0107] Optionally, in determining the corrected difference as the rotational speed difference feature, spatial synergy analysis can be performed on the corrected difference between the left and right wheels to form a comprehensive feature quantity characterizing the group's tendency to leave the ground. The corrected differences between the two left wheels and the two right wheels are obtained separately as two dimensions of the rotational speed difference feature. This rotational speed difference feature does not depend on the anomaly of a single wheel, but requires that the corrected difference between the two wheels on a specific side (left or right) simultaneously deviate significantly from zero in the same direction, indicating that the wheel group on that side exhibits a synergistic loss of load. This rotational speed difference feature is in scalar or vector form. The amplitude and persistence of this feature are directly related to the severity of wheel derailment and show a clear positive correlation with the first risk information: the larger the corrected difference, the stronger the synchronicity between the two sides, and the longer the duration, the higher the rotational speed difference feature value, and the higher the first risk information, achieving an effective mapping from the original signal to risk quantification.

[0108] Optionally, during the process of outputting the first risk information in response to the speed difference feature exceeding the speed difference feature threshold, a dynamic speed difference feature threshold can be preset. This threshold is adaptively adjusted based on the current vehicle speed, road surface type, and system confidence level to ensure high sensitivity in high-risk scenarios such as high speeds and unpaved roads, and to suppress false triggering in low-speed, slippery, or bumpy road conditions. When the speed difference feature of a specific wheel continuously exceeds the speed difference threshold for a duration exceeding a preset first time window (e.g., 50ms), it is determined that the wheel on that side has entered a cooperative ground-lift state, and the first risk information is immediately output.

[0109] In this embodiment, the method described above constructs a highly robust and physically interpretable wheel lift-off recognition process through a four-pronged mechanism: non-driving wheel reference, steering geometry compensation, residual extraction, and collaborative discrimination. This process effectively distinguishes between three scenarios: normal steering, road bumps, and actual airborne movement, enabling the identification of early signs of wheel loss of ground support tens of milliseconds before any change in vehicle posture. Compared to traditional rollover recognition methods that rely on vehicle angular velocity or roll angle, this method avoids posture integral drift, dynamic interference, and response lag, significantly improving the system's early warning capabilities for typical rollover scenarios such as high-speed airborne movement and slope run-off. Its output first risk information, as a leading risk source independent of the posture system, provides a highly reliable and timely key input for subsequent multi-source fusion decision-making, fundamentally enhancing the active protection capabilities and functional safety of the vehicle's passive safety system.

[0110] As an optional implementation, determining the second risk information based on the three-axis information in the operating status information includes: inputting the three-axis information into the vehicle's attitude calculation model, using the attitude calculation model to calculate the vehicle's roll angle and pitch angle from the three-axis information; and determining the second risk information based on the roll angle and pitch angle.

[0111] In this embodiment, during the process of determining the second risk information based on triaxial information, the triaxial information can be input into the attitude calculation model, which then calculates the vehicle's roll and pitch angles from the triaxial information. Based on these roll and pitch angles, the second risk information is then determined.

[0112] The aforementioned attitude calculation model refers to a mathematical calculation framework used to reconstruct the real-time spatial orientation of the vehicle relative to the inertial reference frame in three-dimensional space from the three-axis acceleration signals, three-axis angular velocity signals, and three-axis magnetic field strength signals output from the nine-axis inertial measurement unit. Its core functions are to eliminate sensor noise, compensate for integral drift, and suppress dynamic interference, thereby outputting high-precision and highly robust vehicle attitude angles. In this embodiment, the attitude calculation model can employ an Adaptive Extended Kalman Filter (AKF) architecture. Unlike traditional fixed-parameter Kalman filters, it dynamically adjusts the process noise and observation noise covariance matrix based on the vehicle's current motion state (e.g., angular velocity magnitude, linear acceleration magnitude) and environmental interference level (e.g., magnetic field strength deviation). This maximizes the use of absolute references from gravity and the geomagnetic field to correct gyroscope drift during smooth driving, and automatically reduces the trust in abnormal sensor observations during severe rollovers or strong magnetic interference, instead relying on high-bandwidth angular velocity integrals to maintain attitude continuity.

[0113] Optionally, the aforementioned roll angle can refer to the tilt angle of the vehicle body relative to the horizontal plane in the roll direction, formed by the vehicle rotating around its own longitudinal axis (X-axis). Physically, it represents the degree to which the vehicle rolls laterally in the lateral plane and is a key state variable characterizing whether the vehicle rolls to the left or right. The aforementioned roll angle can be obtained by the attitude calculation model based on the combined results of gravity components measured by a three-axis accelerometer and integration by a three-axis gyroscope. When the vehicle is driving normally, the roll angle is close to zero. When the vehicle rolls over or rotates laterally after being airborne, the roll angle increases rapidly and continues to accumulate. The rate of change of the roll angle (i.e., roll angular velocity) and the rate of change of acceleration (i.e., roll angular acceleration) together constitute the evolution characteristics of the vehicle's lateral dynamic instability. The aforementioned roll angle can also be called the vehicle roll angle.

[0114] Optionally, the aforementioned pitch angle can refer to the tilt angle of the vehicle body relative to the horizontal plane in the longitudinal direction, formed by the vehicle rotating around its own lateral axis (Y-axis). Physically, it represents the degree to which the vehicle nods or pitches in the longitudinal plane, and is a core parameter characterizing the longitudinal attitude changes of the vehicle in complex conditions such as airborne, ramp-up, and fall. The pitch angle is obtained by the attitude calculation model through joint estimation of the gravity component observed by the three-axis accelerometer and the integral of the three-axis angular velocity. The pitch angle is zero when the vehicle is traveling at a constant speed in a straight line. When the front or rear wheels of the vehicle leave the ground first, or when a jump or dive occurs, the pitch angle will significantly deviate from zero and show a rapid changing trend. The evolution trend of the pitch angle, together with the roll angle, constitutes a complete description of the overall roll motion of the vehicle. Especially in complex conditions such as forward or backward rollovers, or complex rotations after airborne conditions, the amplitude and rate of change of the pitch angle are effective bases for judging the roll direction and energy accumulation level.

[0115] In this embodiment, determining the second risk information based on the three-axis information in the operational status information is the central link in achieving accurate identification of vehicle rollover risk. The above process uses the AKF attitude calculation model as the core algorithm. Through the coordinated processing of three-axis acceleration, three-axis angular velocity, and three-axis magnetic field strength signals, it continuously calculates the vehicle's real-time roll and pitch angles in space. Based on the amplitude, rate of change, and dynamic stability of these two attitude angles, a second risk information characterizing the vehicle's rollover trend is constructed. This process overcomes the drift and misjudgment problems of traditional single sensors under high-dynamic and strong-interference conditions, achieving continuous, stable, and high-precision perception of vehicle rollover motion, providing a reliable attitude domain risk basis for the graded ignition decision of the airbag system.

[0116] Optionally, during the process of calculating the vehicle's roll and pitch angles from three-axis information using the attitude calculation model, the three-axis acceleration, angular velocity, and magnetic field strength signals output by the nine-axis inertial measurement unit can be simultaneously acquired and input into the AKF attitude calculation model. The aforementioned attitude calculation model can describe the vehicle's attitude in quaternion form, with the state vector containing quaternion components and a three-axis gyroscope zero-bias estimate. It predicts the attitude at the next moment based on the angular velocity integral using a nonlinear state equation, and constructs an observation model using the gravity component observed by the accelerometer and the geomagnetic field component observed by the magnetometer to correct the predicted attitude. During the correction process, the model dynamically adjusts the observation noise covariance matrix and the system noise covariance matrix based on the real-time calculated angular velocity magnitude, linear acceleration magnitude, and magnetic field strength magnitude. This ensures that under stable operating conditions, the absolute references of the accelerometer and magnetometer are fully trusted to suppress gyroscope drift, while under severe rolling or strong magnetic interference, the trust in abnormal observations is reduced, prioritizing high-bandwidth angular velocity integrals to maintain attitude continuity. After recursive iteration, the model outputs the final estimated quaternion attitude. Then, through coordinate transformation from quaternion to Euler angles, the roll angle and pitch angle of the vehicle at the current moment are accurately calculated. Both are continuous, smooth, and gimbal lock-free real-time attitude calculation results, constituting the direct physical input of the second risk information.

[0117] Optionally, in the process of determining the second risk information based on the roll and pitch angles, the calculated roll and pitch angles can be dynamically evaluated in multiple dimensions to construct a quantitative index that comprehensively reflects the degree of rollover risk. This process not only examines the instantaneous amplitudes of the roll and pitch angles but also simultaneously analyzes their rate of change (angular velocity) and acceleration (angular acceleration) to identify the initiation, development, and acceleration stages of rollover. When the roll angle exceeds a preset threshold and continues to increase, or when the pitch angle undergoes a violent roll accompanied by high angular acceleration, it is determined that the vehicle has entered the rollover energy accumulation stage. Simultaneously, the confidence level of the current attitude estimate is assessed by combining the covariance matrix trace output by the attitude calculation model. If the covariance increases, it indicates that the filter is in a state of strong interference or high dynamics, and the amplification rate of the second risk information will be appropriately reduced to prevent misjudgment due to attitude estimation uncertainty. The value of the second risk information is positively correlated with the combined deviation of the roll angle and pitch angle. The upward trend reflects the increased irreversibility of the vehicle roll motion. When the condition of continuously exceeding the threshold within the time window is met, the above-mentioned second risk information is officially output and participates in the risk fusion decision, providing a high-confidence criterion in the attitude domain for subsequent airbag staged ignition.

[0118] In this embodiment, an adaptive extended Kalman filter model is used to achieve high-precision and robust real-time calculation of vehicle roll and pitch angles, effectively solving the problems of attitude drift and divergence under extreme conditions such as severe rollover, strong magnetic interference, and dynamic acceleration interference. Compared with traditional rollover recognition methods that rely solely on angular velocity or fixed thresholds, this method can not only capture the early attitude evolution characteristics of rollover, but also maintain the stability and continuity of attitude estimation across the entire operating range, greatly improving the system's ability to recognize complex rollover patterns (such as rollover accompanied by forward rollover, and rotation after airborne). The second risk information, as a dynamic risk quantity based on the evolution of attitude in real space, complements the first risk information in the wheel speed domain, jointly constructing a dual-channel decision-making mechanism of "leader prediction - attitude confirmation," significantly enhancing the response accuracy and functional safety of the airbag system in complex accident scenarios, and providing an irreplaceable algorithmic foundation for achieving millisecond-level precise protection.

[0119] As an optional implementation, an attitude calculation model is used to calculate the vehicle's roll angle and pitch angle from three-axis information. This includes: after inputting the three-axis information into the attitude calculation model, using the angular velocity measured by the gyroscope in the model, predicting the vehicle's current attitude using a quaternion integration strategy, and compensating for the angular velocity using the gyroscope's zero-bias estimate to obtain the actual angular velocity; combining the gravity component measured by the accelerometer and the geomagnetic field component measured by the magnetometer in the attitude calculation model to correct the current attitude, obtaining the corrected current attitude; during the correction process, monitoring the changes in the three-axis information, where the changes characterize whether the vehicle is in a complex operating condition; adjusting the corrected current attitude based on the changes, obtaining the adjusted current attitude; and calculating the roll angle and pitch angle from the adjusted current attitude.

[0120] In this embodiment, during the calculation of the vehicle's roll and pitch angles from the three-axis information, the angular velocity measured by the gyroscope in the attitude angle calculation model can be used to predict the vehicle's current attitude through a quaternion integration strategy. The angular velocity can be compensated using the gyroscope's zero-bias estimate to obtain the actual angular velocity. The current attitude can be corrected by combining the gravity component measured by the accelerometer and the geomagnetic field component measured by the magnetometer in the attitude calculation model, resulting in a corrected current attitude. During the correction process, changes in the three-axis information can be detected, and the corrected current attitude can be adjusted based on these changes, resulting in an adjusted current attitude. The roll and pitch angles are then calculated from the adjusted current attitude.

[0121] Angular velocity, measured directly by a three-axis gyroscope, is a three-dimensional vector representing the instantaneous rotational rate of the vehicle around its three orthogonal axes (roll, pitch, and yaw). This three-dimensional vector reflects the vehicle's angular motion relative to inertial space within the current sampling period. The quaternion integration strategy refers to a numerical calculation rule that uses continuous time integration of the angular velocity signal based on quaternion mathematical representation to recursively update the vehicle's attitude. Essentially, it maps angular velocity to the attitude evolution process through quaternion differential equations to avoid the gimbal lock problem that occurs at Euler angle poles. This quaternion integration strategy uses unit quaternions as attitude state variables and performs one-step prediction of the current attitude through quaternion multiplication operations, achieving a continuous and singular-free description of the vehicle's spatial orientation.

[0122] Optionally, the aforementioned current attitude can refer to the quaternion estimate, representing the vehicle's spatial orientation at the current moment, recursively predicted from the optimal attitude at the previous moment using a quaternion integration strategy, and has not yet been corrected by sensor observations. This current attitude relies solely on the gyroscope angular velocity integral and does not incorporate absolute reference information from the accelerometer and magnetometer. Therefore, it is prone to systematic drift due to the accumulation of gyroscope zero bias during long-term operation. The aforementioned zero bias estimate can refer to the three-dimensional compensation amount used in the attitude solution model to estimate the inherent zero-point drift of the three-axis gyroscope online. This zero bias estimate is used to correct the original angular velocity measurement, eliminating systematic deviations caused by temperature changes, device aging, or initial calibration errors. As part of the state vector, the zero bias estimate is updated synchronously with the attitude quaternion during Kalman filtering, aiming to make the compensated angular velocity reflect the vehicle's actual rotational motion as accurately as possible. The zero bias estimate is a key compensation parameter for achieving long-term attitude stability and can also be called the gyroscope drift compensation amount.

[0123] Optionally, the aforementioned actual angular velocity can refer to the corrected angular velocity, representing the vehicle's true rotational motion, obtained by subtracting the zero-bias estimate from the original angular velocity measurement. This actual angular velocity eliminates the influence of inherent gyroscope errors and is an accurate input for the quaternion integration strategy, ensuring that the attitude prediction process is based on a rotational rate close to physical reality. The aforementioned gravity component can refer to the specific force component caused by Earth's gravity, measured by a three-axis accelerometer in the vehicle coordinate system. Physically, it is the portion of the acceleration vector sensed by the sensor that is generated by gravity when the vehicle is stationary or in uniform linear motion. In the attitude calculation model, the gravity component serves as observed information to correct attitude deviations. Because gravity has a known direction in the navigation coordinate system—for example, pointing towards the Earth's center in the North-East-Down (NED) coordinate system—the degree of attitude deviation can be calculated by comparing the predicted attitude after rotation with the measured acceleration, thereby correcting the error. The aforementioned geomagnetic field component can refer to the projection vector of the Earth's magnetic field in the vehicle coordinate system, measured by a triaxial magnetometer. Its physical meaning is the component of the geomagnetic induction intensity at the vehicle's location along the vehicle's axial direction.

[0124] Optionally, the corrected current attitude refers to the quaternion estimate, which is closer to the vehicle's true attitude, obtained by using gravity and geomagnetic field components as observation information and applying Kalman gain weighting correction to the predicted attitude in the attitude calculation model. The corrected current attitude integrates the high-bandwidth dynamic response of the gyroscope with the absolute reference characteristics of the accelerometer and magnetometer. The aforementioned changing state can refer to a comprehensive discriminant quantity calculated based on real-time three-axis information, characterizing the vehicle's current external environment and the intensity of its motion. This quantity encompasses the dynamic trends of angular velocity magnitude, linear acceleration magnitude (after removing gravity), and magnetic field strength magnitude, used to identify whether the vehicle is in complex conditions such as severe rollover, strong magnetic interference, or road bumps. The adjusted current attitude can refer to the final attitude estimate, with adaptive robustness, recalculated based on the corrected current attitude and the complex conditions identified by the changing state, after dynamically adjusting the observation noise covariance matrix. When a violent rollover is detected, the accelerometer observation weight is reduced to suppress linear acceleration interference; when strong magnetic interference is detected, magnetometer correction is disabled to avoid abnormal magnetic field pull; when road bumps are detected, the accelerometer noise is moderately increased to suppress high-frequency vibration disturbances.

[0125] In this embodiment, calculating the vehicle's roll and pitch angles from three-axis information using an attitude calculation model is the core technical path for building a high-precision, robust rollover risk identification capability. This process uses the AKF algorithm framework and achieves continuous, stable, and interference-resistant estimation of the vehicle's spatial attitude through a closed-loop mechanism of "prediction-correction-adaptive adjustment." The entire process starts with quaternion integral prediction of the gyroscope angular velocity, obtains the true rotational input through zero-bias compensation, fuses the accelerometer gravity component and the magnetometer geomagnetic field component for attitude correction, and then dynamically adjusts the filter trust weights based on the changing state of the three-axis information, outputting an adaptively optimized attitude estimation result from which the roll and pitch angles are extracted. This process not only overcomes the performance bottleneck of traditional fixed-parameter filtering in dynamic environments but also achieves adaptive convergence and physical consistency of attitude calculation under complex conditions, providing a solid and reliable mathematical foundation for the generation of secondary risk information.

[0126] Optionally, after inputting the three-axis information into the attitude calculation model, the current attitude of the vehicle is predicted using the angular velocity measured by the gyroscope in the attitude calculation model through a quaternion integration strategy. This means the system uses quaternions as the mathematical expression of attitude, takes the raw angular velocity output by the gyroscope as the dynamic input, and performs time-recursive prediction of the vehicle attitude through quaternion differential equations. This quaternion integration strategy, based on quaternion multiplication, transforms the rotational effect of angular velocity in the vehicle coordinate system into the continuous evolution of attitude quaternions, thereby achieving a non-singularity, continuous description of the vehicle's spatial orientation. This prediction process does not rely on external sensors and is based on the time integration of angular velocity, thus exhibiting high-frequency response characteristics and rapidly capturing instantaneous changes in vehicle attitude.

[0127] For example, quaternions can be used as the attitude description method to avoid gimbal lock problems and achieve continuous description of all attitude angles. Define the system's state vector. A 7-dimensional column vector:

[0128]

[0129] in, To describe the unit quaternion of the vehicle body coordinate system relative to the navigation coordinate system, satisfying the normalization constraint. . This is a random zero-bias estimate for the three-axis gyroscope, used for online real-time compensation of the gyroscope's constant drift. The navigation coordinate system mentioned above can be the NED system or the East-North-Up (ENU) system.

[0130] For another example, the angular velocity shown below can be obtained by measuring a gyroscope: ,in, It can be used to represent angular velocity, that is, a three-dimensional angular velocity measurement synchronously acquired by a gyroscope (e.g., a three-axis gyroscope); It can be used to represent the angular velocity component measured by the gyroscope along the roll axis (X-axis) in the vehicle coordinate system; It can be used to represent the angular velocity component measured by the gyroscope along the pitch axis (Y-axis) in the vehicle coordinate system; It can be used to represent the angular velocity component measured by the gyroscope along the yaw axis (Z-axis) in the vehicle coordinate system.

[0131] Optionally, in the process of compensating for the angular velocity using the gyroscope's zero-bias estimate to obtain the actual angular velocity, the online estimated zero-bias estimate of the three-axis gyroscope is used. From the original angular velocity Subtract from the middle to obtain the corrected actual angular velocity. .in, It can be used to represent the zero-bias estimate, that is, the three-dimensional vector of the zero-bias estimate of a three-axis gyroscope; It can be used to represent the zero-bias estimate of the gyroscope in the roll axis (X-axis) direction of the vehicle coordinate system; It can be used to represent the zero bias estimate of the gyroscope in the pitch axis (Y-axis) direction of the vehicle coordinate system; This can be used to represent the zero-bias estimate of the gyroscope in the yaw axis (Z-axis) direction of the vehicle coordinate system. The zero-bias estimate is updated synchronously with the state vector of the attitude calculation model, essentially providing dynamic compensation for systematic deviations caused by temperature drift, device aging, or initial errors in the gyroscope. The actual angular velocity reflects the physical state of the vehicle's actual rotational motion and is the accurate input to the quaternion integration strategy. Without this compensation, gyroscope drift will cause the attitude prediction to deviate significantly from the true value within seconds, leading to rollover misjudgments. Therefore, the actual angular velocity is a necessary intermediate quantity connecting the original sensor signal and high-precision attitude estimation, ensuring the long-term stability of attitude prediction.

[0132] For example, the measured angular velocity can be compensated using the zero-bias estimate to obtain an estimate of the true angular velocity. .

[0133]

[0134] in, It can be used to represent quaternion multiplication, and to combine two quaternions to achieve rotational composition; It can be used to represent the derivative of the vehicle body attitude quaternion with respect to time, that is, the rate of change of the quaternion with time, and is used to describe the instantaneous evolution rate of the spatial attitude of the vehicle during continuous motion. A unit quaternion can be used to represent the spatial attitude of a vehicle relative to an inertial navigation coordinate system (such as the NED system) at a given moment. .

[0135] Alternatively, assume that the zero bias is a random walk process:

[0136]

[0137] in, This indicates that the rate of change of the gyroscope's zero bias over time is zero, that is, It follows a random walk process in time, and changes extremely slowly, approximating a constant or only affected by tiny random perturbations.

[0138] After discretization, the one-step predicted state is obtained. And one-step prediction of covariance matrix :

[0139]

[0140]

[0141] in, It can be used to represent the state estimate at time k based on time k-1 and the angular velocity measured by the gyroscope at the current time. The prior state estimate obtained by predicting the system state transition function f(); It can be used to represent a state transition function, which estimates the state based on the optimal state at the previous time step. Compared with the gyroscope angular velocity measurement at the current moment The attitude and zero bias prediction (zero bias estimate) at the current moment are calculated using a quaternion integral dynamics model. It can be used to represent the covariance estimated at time k based on the state at the previous time. The prediction covariance matrix obtained by propagating the dynamic characteristics of the system; It can be used to represent state transition matrices; It can be used to represent the system process noise covariance matrix, and to characterize the sources of uncertainty in the state transition model that are not accurately modeled, including the random slow drift of the gyroscope zero bias, the numerical error of the quaternion integration process, and the nonlinear residual of the system.

[0142] Optionally, in the process of correcting the current attitude by combining the gravity component measured by the accelerometer and the geomagnetic field component measured by the magnetometer in the attitude calculation model, the gravity component observed by the accelerometer in the vehicle coordinate system and the geomagnetic field component observed by the magnetometer can be used as reference information. These are compared with the residuals obtained after coordinate transformation of the predicted attitude to determine the theoretical gravity and geomagnetic field directions. A Kalman filter mechanism is used to calculate the correction gain, thereby weighting the predicted attitude. The gravity component is used to correct low-frequency drift in the roll and pitch angles, while the geomagnetic field component is used to correct the yaw angle and assist in roll correction. This correction process significantly suppresses gyroscope integration errors under steady operating conditions, achieving long-term attitude accuracy maintenance. The corrected current attitude is the final intermediate estimate that integrates dynamic response and absolute reference.

[0143] For example, accelerometers and magnetometers can be introduced as observation sources to correct for drift caused by gyroscope integration. Observation vector A 6-dimensional column vector:

[0144]

[0145] in, It can be used to represent the observation vector jointly output by the accelerometer and magnetometer at time k. It is a 6-dimensional column vector, which means it can be the corrected current state. It can be used for the gravitational component, that is, to represent the specific force vector measured by the accelerometer in the vehicle coordinate system. For example, it can be the total acceleration sensed by the sensor, which includes the synthesis of gravitational acceleration and non-gravitational linear acceleration. It can be used to represent the geomagnetic field vector measured by the magnetometer in the vehicle body coordinate system, that is, the projection of the local Earth's magnetic field sensed by the sensor in the vehicle body coordinate system, that is, the geomagnetic field component. It can be used to represent quaternions The defined rotation matrix from the vehicle coordinate system to the navigation coordinate system; It can be used to represent the standard gravitational acceleration vector in the navigation coordinate system; It can be used to represent the standard reference vector of the local geomagnetic field in the navigation coordinate system, and is a known geospatial constant. It can be used to represent the observation noise vector.

[0146] Optionally, during the monitoring of changes in triaxial information, the angular velocity magnitude, the linear acceleration magnitude after removing the influence of gravity, and the magnetic field strength magnitude can be calculated in real time. Based on their dynamic characteristics, it can be determined whether the vehicle is currently in an undesirable operating environment such as severe rollover, strong magnetic interference, or road bumps. The angular velocity magnitude reflects the degree of rotation, the linear acceleration magnitude reflects the non-gravitational acceleration disturbance of the vehicle body, and the magnetic field strength magnitude reflects anomalies in the geomagnetic field environment. The changes in state do not directly participate in attitude state updates, but rather serve as external environment perception signals to trigger subsequent adaptive adjustment mechanisms. This ensures that the filter does not diverge due to over-reliance on sensors in interference environments, nor does it lag due to excessive conservatism.

[0147] Optionally, during the adjustment of the corrected current attitude based on the changing state, the variance values ​​of each component in the observation noise covariance matrix can be dynamically adjusted according to the classification results of the changing state, thereby changing the Kalman gain's confidence in gravity and geomagnetic field observations. When the angular velocity magnitude exceeds the high dynamic threshold, the accelerometer observation noise is increased, the weight of attitude correction is reduced, and the filter relies on gyroscope integration to maintain continuity. When the magnetic field strength deviates from the local geomagnetic reference beyond a preset range, the magnetometer observation noise is set to a maximum value, and magnetic field correction is temporarily disabled. When a sharp fluctuation in linear acceleration is detected, the accelerometer noise is moderately increased to suppress high-frequency turbulence interference. The adjusted current attitude is the final attitude estimation result after environmental adaptive optimization.

[0148] Optionally, in the process of calculating the roll and pitch angles from the adjusted current attitude, the attitude estimate expressed by the adaptively adjusted quaternions can be used to calculate the vehicle's roll and pitch angles in the navigation coordinate system using the standard quaternion-to-Euler angle conversion formula. The roll angle reflects the degree of lateral tilt of the vehicle about the longitudinal axis (X-axis), and the pitch angle reflects the degree of forward and backward tilt of the vehicle about the transverse axis (Y-axis). The above conversion process is a deterministic analytical operation that does not introduce additional errors. Its output is a continuous, smooth, and gimbal-lock-free two-dimensional attitude parameter, which is directly used to construct the second risk information. The roll and pitch angles are the final outputs of the attitude calculation model and are also an intuitive and physical quantitative representation of the vehicle's rollover state, providing direct input for subsequent risk fusion and ignition control.

[0149] In this embodiment, the above method constructs a highly robust and dynamically adaptive vehicle attitude calculation process through a five-layer closed-loop mechanism consisting of quaternion integral prediction, zero-bias compensation, multi-source observation correction, change state perception, and adaptive noise adjustment. This process effectively eliminates gyroscope drift and maintains attitude accuracy over long periods when the vehicle is traveling smoothly. Under complex conditions such as severe rollovers, strong magnetic interference, or bumpy roads, it intelligently suppresses the influence of unreliable sensors, prioritizing the continuity and stability of attitude, thereby solving the divergence and hysteresis problems of traditional fixed-parameter filtering in extreme scenarios.

[0150] As an optional implementation, determining the second risk information based on the roll angle and pitch angle includes: determining the degree of change of the angular rate of the roll angle and pitch angle, wherein the degree of change of the angular rate is used to characterize the severity of the risk of rollover or pitch of the vehicle; and determining the second risk information based on the degree of change of the angular rate.

[0151] In this embodiment, during the process of determining the second risk information based on the roll and pitch angles, the angular velocity variation of each of the roll and pitch angles can be determined. The second risk information can be determined based on these angular velocity variation degrees. The angular velocity variation degree can refer to the second derivative of the roll and pitch angles with respect to time, i.e., attitude angular acceleration, used to characterize the dynamic severity of attitude angle changes during rollover or pitch, reflecting the vehicle's rotational acceleration or deceleration characteristics around the lateral and longitudinal axes. This angular velocity variation degree can characterize the instantaneous dynamic intensity of attitude evolution and is a key dynamic characteristic for identifying the early acceleration phase of rollover risk; it can also be called attitude angular disturbance rate or angular acceleration response intensity. This angular velocity variation degree can directly correlate with the suddenness of vehicle instability and the rate of energy accumulation.

[0152] This embodiment, which determines the second risk information based on roll and pitch angles, includes two consecutive and logically progressive steps: quantifying the severity of vehicle attitude dynamic evolution by calculating the angular rate of change of roll and pitch angles; and constructing second risk information reflecting the intensity of roll or pitch risk based on the amplitude and trend of this angular rate of change. This process does not rely on absolute thresholds of attitude angles but focuses on their dynamic rate of change, thereby improving sensitivity and response speed to early vehicle instability.

[0153] Optionally, in determining the angular rate changes of the roll and pitch angles, the corresponding angular acceleration values ​​can be obtained by acquiring the real-time roll and pitch angles calculated using an adaptive extended Kalman filter and performing a second differential operation on the first-order time derivative (i.e., angular velocity). The aforementioned angular rate change is the second derivative of the attitude angle with respect to time, characterizing the instantaneous acceleration capability of the vehicle body's rotational motion around the lateral and longitudinal axes, directly reflecting the dynamic severity of the attitude evolution when the vehicle encounters a rollover or forward / backward roll.

[0154] Optionally, in determining the second risk information based on the degree of change in angular rate, the degree of change in angular rate of roll angle and pitch angle can be mapped to the corresponding dynamic risk components, and a single second ground-lift risk factor can be generated by using a nonlinear weighting function or a lookup table mapping method. The above mapping relationship is pre-calibrated according to the vehicle dynamics characteristics. When the degree of change in angular rate exceeds a preset dynamic threshold, the risk component grows exponentially to highlight the abrupt risk characteristics under high acceleration conditions. At the same time, a time window integration mechanism is introduced to suppress instantaneous pulse interference, ensuring that the risk information has temporal continuity and decision reliability. The output second risk information is a continuous numerical quantity, positively correlated with the dynamic intensity of attitude angle changes, and is used for subsequent adaptive fusion with the wheel speed risk factor.

[0155] In this embodiment, the method described above, by extracting the second-order dynamic features of the attitude angle, fundamentally improves the assessment of vehicle rollover and pitch risks from "static position judgment" to "dynamic trend prediction." This method does not rely on the absolute amplitude of the attitude angle, avoiding misjudgments caused by sensor drift or low-frequency interference. It significantly enhances the system's early identification capability under high-dynamic conditions such as vehicle airborne conditions, lateral acceleration, and compound rollovers, providing highly sensitive, low-latency, and robust dynamic risk input for graded airbag ignition strategies.

[0156] As an optional implementation, the method further includes: determining vehicle runaway information based on operating status information, wherein the runaway information is used to represent the degree of energy mutation of the vehicle; and determining the vehicle's driving condition based on operating status information and ground status information; determining the weights corresponding to the runaway information and at least one risk information based on the driving condition; step S206, determining the vehicle's safety performance index based on at least one risk information, including: using the weights corresponding to the runaway information and at least one risk information respectively, performing a weighted summation on the runaway information and at least one risk information to obtain the safety performance index.

[0157] In this embodiment, vehicle loss of control information can be determined based on operational status information, and vehicle driving conditions can be determined based on operational status information and ground status information. The weights corresponding to the loss of control information and at least one risk information can be determined based on the driving conditions. In determining safety performance indicators based on at least one risk information, the weights corresponding to the loss of control information and at least one risk information can be used to perform a weighted summation to obtain the safety performance indicators.

[0158] Among these, "runaway information" refers to quantitative indicators characterizing the instantaneous rate of change of the vehicle's total kinetic energy and rotational kinetic energy, calculated based on vehicle operating status information. This runaway information can be a runaway factor based on changes in the vehicle's total energy, with its core form being the rate of change of the vehicle's total mechanical energy derived from the integral of linear acceleration and angular velocity. This runaway information is used to measure the degree of discontinuous energy accumulation or release when the vehicle deviates from a stable driving state, such as the dramatic conversion characteristics of kinetic and potential energy under extreme conditions like airborne jumps, side rollovers, or high-speed collisions. The aforementioned ground state information can refer to road physical characteristic data indirectly inferred from sensors or provided by external environment models, including spatial and frictional characteristic parameters such as road surface slope angle, lateral inclination, adhesion coefficient, presence of slopes or ditches, and whether the area is a metal structure. This state information is used to determine whether the constraint relationship between the vehicle and the ground remains intact.

[0159] Optionally, the aforementioned driving conditions can refer to the vehicle's current dynamic operating mode, determined comprehensively based on operational status information and ground status information, encompassing typical state categories such as smooth driving, normal turning, emergency braking, road bumps, wheel liftoff, side slippage, airborne flight, and slope fall. These driving conditions can serve as a semantic classification of the vehicle's physical environment and dynamic behavior, used to dynamically adjust the decision priority of various risk factors. The aforementioned weights can refer to dynamic adjustment coefficients assigned to runaway information and at least one risk information in the calculation of safety performance indicators, determined by the current driving conditions, used to characterize the relative decision importance of different information sources in a specific scenario. For example, in the vehicle airborne condition, the weights of the ground-lift risk factor based on wheel speed and the runaway information based on total energy change increase, while the weight of attitude angle information decreases due to potential magnetic interference; in the severe rollover condition, the weight of runaway information significantly increases to dominate ignition decision.

[0160] In this embodiment, a tiered safety decision-making process for extreme vehicle dynamic events is constructed through multi-source information collaborative analysis. Based on vehicle operating status information, runaway information characterizing the degree of energy mutation is calculated. Then, ground state information is fused to identify the current driving condition. Weights are dynamically allocated between the runaway information and at least one risk information based on the driving condition. Safety performance indicators are generated through weighted fusion, providing a scientific and robust decision-making basis for the graded ignition of the airbag system. The above process achieves a closed-loop, adaptive, and scenario-based evolution from raw sensor data to safety decision output.

[0161] Optionally, in determining runaway information based on operational status information, dynamic data such as the vehicle's three-axis linear acceleration, three-axis angular velocity, and vehicle speed can be collected in real time to calculate the time derivative of the vehicle's total mechanical energy, i.e., the instantaneous rate of change of the sum of translational kinetic energy and rotational kinetic energy. This runaway information, also known as the energy mutation factor, reflects the rapid accumulation or release of energy during non-steady-state events such as airborne, rollover, sideslip, or impact. Its calculation does not rely on absolute energy values ​​but focuses on the gradient of energy change, thus possessing early sensitivity to high-dynamic events such as high-speed hill climbs, slope jumps, and continuous rollovers, and can capture dangerous stages where the attitude angle has not yet exceeded limits but the energy has already become significantly unbalanced.

[0162] Optionally, in determining the driving condition based on operational and ground status information, the vehicle's motion characteristics (such as speed, roll angle, and wheel speed difference) and road environment characteristics (such as slope, road surface adhesion coefficient, and the presence of slopes or metal structures) are fused using multi-dimensional features. A pre-defined rule base or machine learning classifier is then used to determine the current driving scenario category, such as smooth driving, normal steering, bumpy road surface, wheels off the ground, airborne, or slope fall. This driving condition is a semantic state label, which serves to distinguish the risk evolution mechanism under different physical constraints. It is a logical prerequisite for adaptive weight allocation of risk factors, ensuring that decision-making behavior remains consistent with the real driving environment.

[0163] Optionally, in determining the weights corresponding to the runaway information and at least one risk information based on the driving conditions, the corresponding allocation coefficients can be dynamically selected from a predefined weight mapping table according to the identified driving conditions. For example, in the airborne flight condition, the weight of the runaway information is significantly increased to emphasize the dominance of energy mutation, while the weight of the wheel speed risk factor remains high because the wheels are completely off the ground; in the condition of strong magnetic interference accompanied by rollover, the weight of the attitude angle risk factor is automatically reduced, and the decision is jointly dominated by the runaway information and wheel speed information. This weight is not a fixed value, and its adjustment mechanism is based on prior knowledge of the physical correlation between the driving conditions and each information source, ensuring that the system can still prioritize the most reliable information channel in complex and variable environments.

[0164] Optionally, in the process of weighting and summing the out-of-control information and at least one risk information according to their respective weights, the numerical values ​​of each information source can be linearly fused according to their current credibility and importance to form a single continuous risk assessment result.

[0165] For example, the safety performance index (comprehensive risk index) can be determined using the following formula:

[0166]

[0167] in, It can be used to represent safety performance indicators, such as the comprehensive airborne and rollover risk index of a vehicle at time t; It can be used to represent the wheel speed risk weight coefficient that is dynamically adjusted based on driving conditions; It can be used to represent the first off-ground risk factor; It can be used to represent the attitude risk weight coefficient based on attitude estimation confidence and dynamic adjustment of environmental disturbances; It can be used to represent the second ground clearance risk factor calculated based on the vehicle body roll angle, pitch angle and their rate of change; It can be used to represent the runaway risk weight coefficient that is dynamically adjusted based on driving conditions; It can be used to represent the rate of change of total mechanical energy of the vehicle body calculated based on the integral of linear acceleration and angular velocity of the vehicle, that is, the runaway factor based on the change of total energy (translation + rotation) of the vehicle body.

[0168] In this embodiment, the above method constructs a safety performance evaluation system with high robustness and strong scene adaptability through energy mutation perception, environmental semantic recognition, dynamic weight allocation, and multi-source fusion decision-making. This method breaks through the traditional static discrimination mode that relies solely on attitude angle thresholds, incorporating the essential characteristics of vehicle dynamics and external environmental constraints into a unified evaluation framework. This significantly improves the recognition accuracy and response timeliness under extreme conditions such as airborne, compound rollover, and strong interference. Simultaneously, through adaptive weight adjustment, it ensures the continuity and reliability of safety functions even when sensors malfunction or operating conditions change abruptly.

[0169] As an optional implementation, the method further includes: in response to a safety performance index being greater than or equal to a first safety performance index threshold and a safety performance index being less than a second safety performance index threshold, determining a control strategy matching the safety performance index as a first control strategy, wherein the first control strategy is used to represent rules for controlling the seat belts and / or side curtain airbags in the vehicle; in response to a safety performance index being greater than or equal to the second safety performance index threshold, determining a control strategy matching the safety performance index as a second control strategy, wherein the safety level of the vehicle to be controlled according to the second control strategy is less than the safety level of the vehicle to be controlled according to the first control strategy, and the second control strategy is used to represent rules for controlling the airbags in the vehicle; in response to a safety performance index being less than the first safety performance index threshold, returning to the step of obtaining operating status information.

[0170] In this embodiment, if the safety performance index is greater than or equal to a first safety performance index threshold and less than a second safety performance index threshold, the control strategy matching the safety performance index can be determined as the first control strategy. Conversely, if the safety performance index is greater than or equal to the second safety performance index threshold, the control strategy matching the safety performance index can be determined as the second control strategy. Conversely, if the safety performance index is less than the first safety performance index, the system can return to obtain the operating status information.

[0171] The first safety performance index threshold can refer to a preset critical value used to trigger the initial response level of the vehicle's passive safety system. This first safety performance index threshold can also be called the first-level risk trigger threshold, corresponding to the initial stage of the transition from potential to apparent vehicle instability risk. When the safety performance index reaches or exceeds the first safety performance index threshold, it indicates that the vehicle has entered the early stage of high-risk evolution, but has not yet reached the critical point of structural failure or serious injury to occupants. The design basis of the first safety performance index threshold is the time margin required for seatbelt pretensioning and side curtain airbag pre-charging, ensuring that the preparatory actions of the occupant restraint system are completed before airbag deployment. Essentially, it serves as the boundary criterion for the "warning-preparation" stage.

[0172] Optionally, the aforementioned second safety performance index threshold can refer to a higher-order preset critical value used to trigger the final ignition action of the vehicle airbag. This second safety performance index threshold can also be called a secondary risk trigger threshold, corresponding to the emergency stage where the vehicle has entered an irreversible high-energy rollover or airborne state, and the occupants face the risk of direct collision or ejection. The setting of this threshold must ensure that the system only executes airbag deployment after sufficient accumulation of risks across multiple dimensions, such as attitude angle, energy mutation, and wheel speed leaving the ground, to avoid false triggering, while allowing sufficient time for the airbag to complete the inflation process. Its function is to achieve the final response boundary of "execution-protection".

[0173] Optionally, the first control strategy can refer to the set of graded protection actions executed after the safety performance index reaches the first safety performance index threshold. It can also be called the first-level control or pre-protection strategy. The core content is to activate the seat belt pretensioner and / or the side curtain airbag pre-charging circuit. The first control strategy does not involve airbag detonation, but rather tightens the occupant position through mechanical restraint and pre-starts the capacitor charging or ignition preheating of the airbag gas generator to shorten the subsequent airbag response delay and improve the overall protection efficiency.

[0174] Optionally, the second control strategy can refer to the final set of active protection actions executed after the safety performance index reaches the second safety performance index threshold. It can also be called secondary control or main protection strategy. The core content is to detonate the side airbag, rollover protection airbag and associated restraint device, which form a rigid buffer space through high-speed inflation to directly absorb collision energy and limit occupant displacement.

[0175] In this embodiment, a tiered response mechanism based on safety performance indicators is constructed to achieve refined and phased safety control for extreme dynamic events in vehicles. This process, based on the comparison results of safety performance indicators with two preset thresholds, sequentially triggers the first control strategy, the second control strategy, or returns to the data acquisition state, forming a three-level progressive protection logic of "early warning – preparation – execution." This mechanism, while ensuring occupant safety, effectively avoids system malfunctions caused by misjudgment or premature intervention, ensuring that the airbag system can implement the most appropriate and timely protective measures at different stages of risk evolution.

[0176] Optionally, in response to a safety performance index being greater than or equal to a first safety performance index threshold and less than a second safety performance index threshold, a control strategy matching this index is determined as the first control strategy. This strategy is the first-level response action of the vehicle's passive safety system, and its core execution content is to activate the seat belt pretensioner and the side curtain airbag pre-charge mechanism. The seat belt pretensioner tightens the webbing within milliseconds, eliminating the slack space between the occupant and the seat, placing the occupant in a more stable restrained position; the side curtain airbag pre-charge mechanism activates the capacitor charging or ignition preheating circuit of the gas generator, enabling the airbag to have immediate deployment capability before it is fully inflated, reducing the time from the issuance of the command to full deployment from the conventional 30 milliseconds to less than 10 milliseconds. This strategy does not trigger airbag detonation, aiming to establish a physical and electrical preparedness state for subsequent possible severe collisions or rollovers. Its design is based on the fact that there is still a response window of several milliseconds for the occupant in the initial stage of a rollover, improving the final protection efficiency through non-destructive intervention.

[0177] Optionally, in response to a safety performance index being greater than or equal to a second safety performance index threshold, a control strategy matching this index is determined as the second control strategy. This strategy represents the system's highest-level response action, and its core execution involves detonating the side airbags, rollover protection airbags, and related restraint devices. High-pressure gas is used to inflate the airbags in a very short time, forming a rigid buffer structure to absorb impact energy and limit the displacement of the occupant's head and torso. The triggering condition for this strategy requires that the risk has accumulated to an irreversible, lethal level, and the duration meets the de-shaking requirements, ensuring it is not driven by a transient event. Essentially, it is the final decision by which the system transitions from a "prepared state" to "active protection," characterized by a one-time, irreversible, and high-energy release. The execution standards of this strategy are stricter than the first control strategy, aiming to provide the highest level of physical protection at the critical moment when the occupant faces the risk of direct collision or ejection. Even in the event of a misjudgment, its consequences take precedence over the serious injuries or fatalities caused by a missed detection.

[0178] Optionally, in response to a safety performance index falling below a first safety performance index threshold, the system returns to the initial step of acquiring operational status information and continues to cyclically collect and process multi-source sensor data from the vehicle. This behavior aims to maintain the system's low-power operation and avoid unnecessary resource consumption and control actions for non-risk conditions. Within this range, the vehicle is in a low-risk state such as stable driving, normal steering, or slight bumps, and the safety performance index has not reached the critical condition for activating any protective measures. If seatbelt pretensioning or curtain airbag pre-charging is prematurely triggered at this time, it will lead to passenger discomfort, reduced lifespan, and increased false alarm rate, while also increasing the electrical load on the entire vehicle, violating the design principle of "necessity first" for safety systems.

[0179] In this embodiment, the aforementioned graded response process, through clear threshold division and strategy binding, achieves a precise transition of airbag control from "no action—pre-protection—main protection," significantly improving the system's decision-making rationality and execution reliability under complex and variable operating conditions. This mechanism not only avoids the false triggering and response lag problems that are prone to occur in traditional single-threshold systems, but also maximizes protection efficiency through the timing coordination of "pre-charging" and "main detonation," while effectively suppressing invalid actions in low-risk stages, reducing system false alarm rate and energy consumption.

[0180] As an optional implementation, step S208 involves controlling the vehicle operation according to a control strategy that matches the safety performance indicators, including: in response to the control strategy being a first control strategy, performing a tightening operation on the seat belt and / or performing an inflation operation on the side curtain airbags; and in response to the control strategy being a second control strategy, performing a detonation operation on the airbags.

[0181] In this embodiment, during the process of controlling vehicle operation according to the control strategy matched with safety performance indicators, if the control strategy is the first control strategy, the seat belts can be tightened and / or the side curtain airbags can be inflated. If the control strategy is the second control strategy, the airbags in the vehicle can be detonated. This embodiment, after determining the control strategy corresponding to the matched safety performance indicators, drives the vehicle's passive safety devices through precise execution commands, achieving closed-loop control from risk assessment to physical protection. This execution process, based on the hierarchical differences between the first and second control strategies, implements differentiated and sequential action responses for the seat belts, side curtain airbags, and airbags, ensuring that the most appropriate protection measures are used at different risk stages. This avoids premature intervention causing unnecessary interference while ensuring sufficient occupant protection when fatal risks occur, thereby achieving the accuracy, effectiveness, and reliability of the safety system response.

[0182] Optionally, when the control strategy is the first control strategy, during the seatbelt tightening operation, the seatbelt pretensioner, upon receiving the control command corresponding to the first control strategy, can apply tension instantaneously within 10 to 30 milliseconds via a pyrotechnic or electric drive mechanism. This eliminates the slack between the seatbelt webbing and the seat, firmly restraining the occupant's body against the seat back, reducing the magnitude of forward or lateral displacement during subsequent rollovers or collisions. This tightening action does not rely on airbag inflation, operates completely independently, and is activated only when the safety performance index reaches the first threshold. Its function is to establish a stable initial restraint state for the occupant, providing a more controllable occupant posture basis for subsequent side curtain airbag deployment or airbag detonation, and preventing protection failure due to occupant loose displacement.

[0183] Optionally, under the first control strategy, during the inflation operation of the side curtain airbags, the airbags are not fully deployed. Instead, a pre-charging process is initiated for the side curtain airbag gas generator. This includes activating the capacitor charging circuit, preheating the ignition agent, or pre-pressurizing the gas storage chamber, putting the airbags in a physically "ready" state. This ensures that once the second control strategy is triggered, the airbags can complete the entire inflation process within 5 to 10 milliseconds, reducing the traditional 30-millisecond response time from command to full deployment to the extreme. This pre-charging behavior does not produce visible airbag expansion, nor does it apply external pressure. It only completes internal energy storage and ignition link activation. The design goal is to eliminate the inherent delay of the airbag system and ensure that the side curtain airbags can deploy synchronously with the occupant's displacement during the rollover event, achieving maximum protection coverage.

[0184] Optionally, under the second control strategy, during the airbag detonation process, the central control unit can send a high-energy current signal to the airbag ignition circuit to trigger the rapid combustion of the solid propellant in the gas generator. This generates a large amount of inert gas, completing the full inflation of the airbag within 60 milliseconds, forming a rigid buffer structure covering the occupant's head and sides. This action is executed immediately after the safety performance index reaches the second safety performance index threshold and meets the duration de-shaking requirements, exhibiting irreversible, high-energy release, and one-time triggering characteristics. The airbag detonation, seatbelt tightening, and side curtain airbag pre-charging are triggered synchronously or slightly later, forming a collaborative protection network. Through a dual mechanism of constraint and buffering, it maximizes the absorption of impact energy, disperses load, and limits head and chest displacement, preventing occupants from directly colliding with hard objects inside the vehicle or being ejected from the vehicle.

[0185] In this embodiment, a progressive safety protection system from "constraint preparation" to "active buffering" is achieved through the aforementioned hierarchical and time-coordinated control commands. The first control strategy completes the pre-fixation of occupant posture and pre-activation of the airbag system in a non-destructive and low-energy manner, laying the physical and temporal foundation for subsequent protective actions. The second control strategy performs decisive intervention after risk confirmation, directly providing an energy absorption mechanism. The entire process avoids the isolated execution of single actions. Through the coordinated linkage of seat belts, side curtain airbags, and airbags, a three-dimensional protection network covering the entire occupant is constructed. While ensuring high safety, it minimizes the negative impact of false triggering, fully meeting the stringent requirements of automotive passive safety systems for response accuracy, functional reliability, and ASIL D functional safety level.

[0186] As an optional implementation, the method further includes: after the tightening operation and / or inflation operation are completed, if the safety level of the vehicle operating according to the first control strategy decreases, the airbag is detonated according to the second control strategy.

[0187] In this embodiment, if the vehicle's safety level decreases after the tightening and / or inflation operations are completed, the airbags can be detonated according to the second control strategy. This embodiment does not terminate the safety assessment after executing the first control strategy, but continuously monitors the vehicle's dynamic evolution trend, constructing a closed-loop enhancement mechanism of "pre-protection—reassessment—ultimate response." This mechanism allows the system to immediately trigger the second control strategy and execute the airbag detonation operation if it detects further deterioration in the vehicle's safety status after the seatbelt tightening and side curtain airbag pre-charge are completed, thereby achieving dynamic response and proactive escalation to high-risk events. This process breaks through the static mode of "single judgment, fixed execution" in traditional hierarchical control, giving the system adaptive reasoning capabilities in the time dimension, significantly improving occupant protection effectiveness in complex, continuous rollover, or airborne scenarios.

[0188] Optionally, the determination of whether the vehicle's safety level has decreased after the execution of the first control strategy can be achieved by continuously collecting and analyzing operational status information and the real-time changing trends of safety performance indicators. After the seat belts have tightened and the side curtain airbags have pre-charged, the system enters a high-frequency monitoring mode, continuously calculating the derivatives of safety performance indicators, the rate of change of roll angle, the energy accumulation slope, and the persistence of abnormal wheel speed. If, within milliseconds after pre-charging, the safety performance indicators continue to rise and approach or exceed the threshold of the second safety performance indicator, or if the roll angle acceleration increases sharply, the vehicle body experiences continuous rollover motion, or the wheel-off-ground state is not alleviated or even aggravated, the system determines that the vehicle's safety level has not stabilized due to the first control strategy, but has instead continued to deteriorate. This judgment does not rely on a single parameter threshold, but comprehensively assesses the rate of risk evolution and dynamic trends, ensuring that it identifies a real risk escalation rather than transient fluctuations, thereby avoiding misjudgments caused by signal jitter.

[0189] Optionally, upon confirming a decrease in vehicle safety, a second control strategy is immediately activated to detonate the airbags. This operation is initiated by the control unit issuing a detonation command the moment the risk is determined to be deteriorating, triggering the ignition circuit of the airbag gas generator, causing the airbag to inflate and form a rigid protective structure in a very short time. This detonation action is not constrained by the previous execution of the first control strategy; its triggering logic is independent of the initial threshold judgment and is entirely based on evidence of dynamic deterioration in the current state. Even if the first control strategy has been executed, if the environmental risk is not mitigated, the system will still prioritize occupant safety and proactively upgrade to the highest level of protection. This process ensures that the airbag intervenes at the most critical moment, compensating for the insufficient protection capabilities of pre-protection measures in the face of high-intensity, continuous rollovers, and achieving a seamless transition from "preparatory protection" to "ultimate interception."

[0190] In this embodiment, the dynamic upgrade mechanism described above continuously assesses the risk evolution trend after the execution of the first control strategy, endowing the system with adaptive response capabilities to complex accident scenarios. This solves the protection lag problem caused by the "wait after pre-protection" in traditional graded systems, enabling airbag detonation to be dynamically triggered based on the actual risk progression rather than being limited by the static judgment of the initial threshold, significantly improving the protection effectiveness in highly dynamic composite events such as high-speed rollovers, multiple airborne events, and continuous impacts.

[0191] The technical solutions of the embodiments of this application will be illustrated below with reference to preferred embodiments.

[0192] While the incidence of vehicle airborne accidents (such as flying over road shoulders or plunging down slopes) and rollover accidents has decreased with the improvement of vehicle performance and the increasing complexity of driving environments, their fatality rate remains high. Most existing ACUs rely on built-in micro-electro-mechanical systems (MEMS) sensors, mainly including multi-axis accelerometers and gyroscopes.

[0193] Existing technologies generally suffer from the following drawbacks: Response lag. Traditional algorithms primarily rely on vehicle roll angles or angular velocities exceeding thresholds to determine rollover. However, when a vehicle veers off the road or becomes airborne, the wheels often lose ground restraint before the vehicle body. Relying solely on vehicle attitude sensors cannot identify this in the initial stage of instability (the instant the wheels leave the ground), leading to a delay in airbag ignition timing. There is a trade-off between attitude calculation accuracy and robustness. For example, regarding drift, during continuous vehicle rollovers, the gyroscope's integration error accumulates rapidly over time, causing the calculated roll angle to deviate significantly from the true value, easily resulting in false or missed airbag deployments. Another example is dynamic interference. While introducing magnetometers and accelerometers can correct drift, during violent vehicle movements (such as rollovers or impacts), external linear acceleration can interfere with gravity field measurements, and metal guardrails or vehicle deformation can interfere with geomagnetic field measurements. Traditional fixed-parameter Kalman filtering cannot achieve a balance between "suppressing dynamic interference" and "correcting drift using observations," either leading to attitude divergence due to over-reliance on sensors or resulting in response lag due to over-filtering. The rigid system architecture, with traditional ACUs highly coupling sensors and computing units and fixed within the cockpit, hinders the utilization of high-performance computing resources and impedes the fusion of multiple sensors (such as smart chassis sensors) as electrical and electronic architectures (E / E architectures) evolve towards domain centralization.

[0194] In this embodiment, considering the above-mentioned situations, a vehicle airborne rollover recognition and airbag control method based on the fusion of nine-axis inertial information and wheel speed is proposed. Real-time acquisition of three-axis acceleration, three-axis angular velocity, and three-axis magnetic field strength data from the vehicle's nine-axis IMU, as well as four-wheel wheel speed sensor data, is performed. Based on the wheel speed signals, the rotational speed difference characteristics of the wheels are calculated to identify early signs of wheels losing vertical support from the ground, generating a first ground-leaning risk factor. An AKF attitude calculation model is constructed, and a dynamically adjusted noise covariance matrix is ​​used to resolve the contradiction in attitude observation under high dynamic motion and strong magnetic interference, accurately calculating the vehicle's roll and pitch angles, generating a second ground-leaning risk factor. The first and second ground-leaning risk factors are adaptively weighted and fused according to the vehicle's current operating conditions to calculate an airborne rollover risk index. Based on the risk index, a graded ignition threshold matrix is ​​matched to execute control strategies such as seatbelt pretensioning, side curtain airbag pre-inflation, or airbag deployment. The control unit in this embodiment has a flexible physical location and can be integrated into a traditional ACU or vehicle domain controller. The embodiments of this application effectively solve the technical problems of prediction lag and attitude drift of traditional single sensors, while possessing extremely high system architecture flexibility and functional safety.

[0195] The methods of the embodiments of this application will be further illustrated below.

[0196] This embodiment aims to solve the aforementioned technical problems by providing an airbag control scheme that combines wheel speed prediction, employs high-precision adaptive Kalman filter attitude calculation, and supports distributed deployment. The technical solution adopted is as follows:

[0197] A vehicle airbag control method based on nine-axis inertial information and wheel speed fusion for airbag recognition during vehicle takeoff and rollover includes: multi-source data acquisition, simultaneously acquiring nine-axis IMU data (acceleration, angular velocity, magnetic field) and four-wheel wheel speed data; wheel speed-assisted airbag prediction, analyzing the deviation between each wheel speed and the reference vehicle speed, combining steering angle information to eliminate wheel speed differences caused by normal steering, identifying the characteristics of instantaneous transfer or loss of wheel load, and generating a first takeoff risk factor; adaptive Kalman filter (AKF) attitude calculation (soft core), constructing an extended Kalman filter model, dynamically adjusting the noise covariance matrices Q and R in real time according to the vehicle's motion state (angular velocity, acceleration) and environmental interference (magnetic field strength). In the stable phase, observation noise is reduced, maximizing the use of accelerometers and magnetometers to correct gyroscope drift; in the stage of violent rollover or strong magnetic interference, observation noise is increased, mainly relying on high-bandwidth gyroscopes to maintain attitude, thereby providing high-precision and robust roll and pitch angles throughout the rollover process, generating a second takeoff risk factor. Adaptive fusion decision-making dynamically adjusts the weights of wheel speed and attitude signals based on information such as vehicle speed, steering state, and road surface roughness to calculate a comprehensive risk index. Tiered ignition control employs a pre-tensioning + pre-charging ignition strategy. Fault diagnosis and degradation processing monitors the validity of sensor signals in real time and executes a three-level degradation strategy.

[0198] Utilizing the technical solution described in this application, millisecond-level prediction is achieved. By leveraging the high sensitivity of wheel speed signals to wheel liftoff, the risk of airborne injury is identified tens of milliseconds earlier than traditional attitude methods. Achieving ultimate attitude accuracy and robustness, the adaptive Kalman filter algorithm solves the accuracy problem of traditional algorithms under dynamic disturbances. By adjusting the "trust level" in real time, the attitude angle remains accurate and reliable even under extreme conditions such as vehicle rollover, airborne injury, and passage over metal guardrails, keeping errors within a minimal range. The architecture is flexible and adaptable to future vehicle models. The control and computing units in this application are not bound to specific hardware locations. They can be integrated into traditional ACUs or deployed in domain controllers, utilizing high-performance automotive chips for more complex fusion calculations, adapting to the centralization trend of automotive electronic and electrical architectures.

[0199] Figure 3 This is a schematic diagram of a vehicle airborne and rollover recognition control system according to an embodiment of this application, as shown below. Figure 3As shown, the system may include a sensor layer 31, a communication network 32, a computing and control layer 33, and an actuator 34. The sensor layer 31 may include a nine-axis inertial measurement unit 311, four-wheel speed sensors 312, and other sensors (steering angle / camera, etc.) 313. The communication network 32 may include a CAN bus / FlexRay / Ethernet 321. The computing and control layer 33 may include a traditional ACU 331, a domain controller 332, a core algorithm module 333, a fault diagnosis and degradation module 334, and a non-volatile memory 335. The actuator 34 may include an ignition drive circuit 341, a seatbelt warning device 342, and side curtain airbags 343.

[0200] Figure 4 This is a flowchart of a vehicle airborne and rollover recognition and airbag control method based on the fusion of nine-axis inertial information and wheel speed, according to an embodiment of this application. Figure 4 As shown, the method may include the following steps.

[0201] Step S401: Initialize parameters and model.

[0202] In this embodiment, when the system is powered on or started, key parameters of the adaptive extended Kalman filter (EKF) are initialized, including the state vector (quaternion and gyroscope zero bias), the baseline value of the noise covariance matrix, the dynamic threshold of wheel speed, the graded ignition thresholds (Thresh1, Thresh2), the upper and lower limits of the hysteresis comparator, and the low-pass filter coefficients, so as to establish a stable algorithm foundation for subsequent real-time data processing.

[0203] Step S402: High-frequency acquisition of sensor data.

[0204] In this embodiment, raw data from a nine-axis IMU (three-axis acceleration, three-axis angular velocity, and three-axis magnetic field) and four-wheel speed sensors are simultaneously acquired at a sampling frequency of no less than 500Hz to ensure that the system has millisecond-level response capability and provides highly timely input for attitude calculation and ground clearance prediction.

[0205] Optionally, after acquiring sensor data, three stages can be executed in parallel: one stage is wheel speed liftoff feature extraction to generate the first liftoff risk factor, another stage is nine-axis adaptive attitude calculation to generate the second liftoff risk factor, and the third stage is energy and motion analysis to generate runaway factors.

[0206] In the stage of extracting wheel speed liftoff features to generate the first liftoff risk factor, steps S403 to S407 can be executed. In the stage of calculating the nine-axis adaptive attitude to generate the second liftoff risk factor, steps S408 to S413 can be executed. In the stage of generating runaway factors through energy and motion analysis, steps S414 to S416 can be executed.

[0207] Step S403: Calculate the reference vehicle speed.

[0208] In this embodiment, the reference vehicle speed is calculated. ,in, It can be used to indicate reference vehicle speed; It can be used to represent the wheel speed signal (rotation information) of the left non-drive wheel, that is, the instantaneous rotational speed value collected by the wheel speed sensor installed on the left non-drive axle of the vehicle (such as the rear wheel of a front-wheel drive vehicle and the front wheel of a rear-wheel drive vehicle). It can be used to represent the wheel speed signal (rotation information) of the right non-drive wheel, that is, the instantaneous rotational speed value collected by the wheel speed sensor installed on the right non-drive axle of the vehicle; It can be used to represent the operation of taking the minimum value between the wheel speed signal of the left non-driving wheel and the wheel speed signal of the right non-driving wheel.

[0209] Step S404: Obtain the steering angle.

[0210] In this embodiment, real-time steering angle signals are read from the steering angle sensor or the ESP system. This is used to correct the normal wheel speed difference caused by turning, and to avoid misjudging normal steering as the wheels leaving the ground.

[0211] Step S405: Calculate the deviation of each wheel.

[0212] In this embodiment, the steering angle is combined Compensation Theoretical Difference ,in, It can be used to represent the theoretical speed difference, which can be the theoretical difference in wheel speed that should exist in the left and right wheels when the vehicle is in a pure steering motion state, due to the different radii of the travel trajectories of the left and right wheels, calculated based on the vehicle's geometry and kinematics. It can be used to represent the steering angle of a vehicle; It can be used to indicate wheelbase.

[0213] Step S406: Does the deviation exceed the dynamic threshold?

[0214] In this embodiment, if the deviation exceeds the dynamic threshold, step S407 can be executed; otherwise, if the deviation does not exceed the dynamic threshold, step S408 can be executed.

[0215] Step S407: Output the first off-ground risk factor.

[0216] In this embodiment, if the corrected wheel speed deviation on a specific side... If the threshold is exceeded, the risk of ground-based accidents is considered to have increased.

[0217] Step S408: Keep monitoring.

[0218] In this embodiment, if monitoring is required, execution can be resumed from step S402.

[0219] Step S409: Detect motion state and interference, and perform low-pass filtering and hysteresis adjustment.

[0220] In this embodiment, the angular velocity magnitude |omega| is calculated in real time. |), Magnetic field magnitude| | And the linear acceleration magnitude, a first-order low-pass filter is applied to the linear acceleration to eliminate high-frequency vibration interference, and a hysteresis comparison mechanism is combined to determine the stable state, avoiding frequent switching caused by noise.

[0221] Step S410: Run the adaptive EKF.

[0222] In this embodiment, based on the quaternion state model, extended Kalman filtering is performed, attitude prediction is performed using gyroscope angular velocity, and state correction is performed by combining accelerometer and magnetometer observations; the system noise Q and observation noise R matrices are dynamically adjusted according to the motion and disturbance states detected in S409.

[0223] Step S411: Calculate the attitude angles.

[0224] In this embodiment, after calculating the attitude angle, step S412 can be executed, and the covariance and weights can be input to the multi-sensor adaptive fusion module.

[0225] Step S412: Calculate the change in angular rate.

[0226] In this embodiment, the roll angle and pitch angle are differentiated to obtain angular velocity and angular acceleration, which are used to assess the dynamic severity of the roll and prevent misjudgment or omission due to lag in a single angle.

[0227] Step S413: Output the second off-ground risk factor.

[0228] In this embodiment, a second ground-lift risk factor is constructed based on the uncertainty of attitude angle, angular velocity, angular acceleration and their covariance estimation. When the attitude changes drastically and the confidence level is high, the value of this factor is significantly improved to reflect the vehicle rollover trend.

[0229] Step S414: Is the vertical acceleration / total energy abnormal?

[0230] In this embodiment, if the vertical acceleration / total energy is abnormal, step S415 can be executed; otherwise, step S416 can be executed.

[0231] Step S415: Output off-ground / out-of-control confidence level.

[0232] In this embodiment, if the vertical acceleration is detected to be continuously below the threshold or the total energy is drastically reduced, it is determined that the vehicle may be in a state of airborne or out of control. The weight or value of the out-of-control factor is increased to enhance the system's sensitivity to complex risks.

[0233] Step S416: Maintain the baseline state.

[0234] In this embodiment, if both vertical acceleration and total energy are within the normal range, the runaway factor is maintained as the baseline value, and the risk fusion logic is not actively interfered with to avoid false triggering.

[0235] Step S417: Calculate the comprehensive risk index Risk.

[0236] In this embodiment, after the multi-sensor adaptive fusion module receives the first off-ground risk factor and the second off-ground risk factor, the comprehensive risk factor can be determined in the multi-sensor adaptive fusion module.

[0237] Optionally, considering the use of adaptive Kalman filtering (AKF) to calculate the attitude, the attitude angles output by the system include a covariance matrix. This matrix represents the uncertainty of the current attitude estimation. Embodiments of this application utilize... The fusion weights are adjusted in conjunction with operating condition information. A comprehensive risk index is defined. The calculation formula is as follows:

[0238]

[0239] in, The first ground clearance risk factor based on wheel speed characterizes abnormal wheel loads. Based on attitude angle (roll angle) Pitch angle The second risk factor for takeoff is angular velocity, which is considered in the calculation. and angular acceleration This prevents lag in single attitude angle integrals. The runaway factor, based on the change in the total energy of the vehicle body (translation + rotation), is used to assist in determining the collision energy level.

[0240] Optionally, the dynamic adjustment logic of the weighting coefficients may include attitude confidence adjustment and working condition adaptive adjustment.

[0241] Optionally, for attitude confidence adjustment Calculate the covariance matrix of the AKF output. Trajectory corresponding to attitude angle .when When the value is small (i.e., the filter is in a stable locked state), The higher value is chosen, primarily relying on the attitude angle for roll detection. When When the situation is relatively large (i.e., the vehicle is in a highly dynamic state, experiencing drastic changes in attitude, or is in the recovery period from strong magnetic interference), It automatically lowers its weight, reducing reliance on uncertain attitudes and instead increasing the weight of wheel speed signals.

[0242] Optionally, adaptive adjustment based on operating conditions. When the vehicle speed is lower than the preset value (e.g.) When wheel speed sensors fail to accurately reflect ground clearance characteristics (possibly due to slippage leading to misjudgment), forced... It relies on the IMU. When the vehicle speed is higher than the preset value and When it is large, increase The high sensitivity of wheel speed to wheel lift-off compensates for attitude uncertainty.

[0243] Step S418, hierarchical decision-making.

[0244] In this embodiment, if Risk≤Thresh1 is detected ( If Thresh1≤Risk≤Thresh2 is detected, step S421 can be executed. If Risk≥Thresh2 is detected, step S419 can be executed. If so, then step S422 can be executed.

[0245] Optionally, a multi-stage control process of pre-tightening + pre-charging + ignition can be adopted to deal with accidents at different stages of development.

[0246] Step S419: Execute Level 1 control.

[0247] In this embodiment, the seat belt pretensioning and side curtain airbag pre-charging are triggered to tighten the occupant restraint system in advance and prepare energy for airbag ignition, thereby shortening the response delay and improving protection efficiency.

[0248] Optionally, for Level 1 control (alert level), the triggering condition is: Action: Seatbelt pretensioning activates the ignition circuit, instantly tightening the seatbelt and pulling the occupant towards the seat, eliminating webbing looseness. Side curtain airbag pre-charge activates the preheating or capacitor charging circuit of the side airbag initiator (if equipped), reducing ignition delay from the normal... shortened to Within [a certain period], awaiting final judgment. Status maintained; system enters high-frequency scanning mode for monitoring. The changing trend.

[0249] Step S420: Has the trend worsened?

[0250] In this embodiment, if the trend worsens, step S422 can be executed. Conversely, step S421 can be executed.

[0251] Step S421: Maintain cyclic monitoring.

[0252] In this embodiment, when the risk does not reach the ignition threshold or the trend is stable, the system remains in the monitoring cycle, continuously collecting and fusing data, waiting for the next risk evolution or the end of the event.

[0253] Step S422: Execute secondary control.

[0254] In this embodiment, the side curtain airbags, side airbags, and other rollover protection airbags are detonated to complete the final occupant protection action.

[0255] Optionally, for secondary control (ignition stage), the triggering condition is: and duration (De-shaking). Action: Deploy the side curtain airbags, side airbags, and rollover protection airbags. During this phase, threshold According to and Fine-tune the current state (e.g., if wheel speed has a high weight, the threshold can be relaxed to allow for a fast response; if attitude has a high weight, the threshold must be strictly controlled to prevent false alarms).

[0256] Step S423: Continuously monitor the accident status.

[0257] In this embodiment, after the airbag is deployed, the system continues to monitor the IMU and wheel speed signals to determine whether it is a multi-stage accident (such as a secondary collision) and ensures that the ignition logic is triggered only once to prevent repeated detonation.

[0258] Step S424, Diagnosis and Fault Recording.

[0259] In this embodiment, in the post-accident stage, key data (such as original sensor values, risk index change curves, fault codes, and EKF covariance status) are automatically written into non-volatile memory for use in accident analysis, recall tracing, and subsequent algorithm optimization, thus meeting functional safety requirements.

[0260] Figure 5 This is a schematic diagram illustrating the principle of attitude calculation using adaptive Kalman filtering according to an embodiment of this application, as shown below. Figure 5As shown, this principle can include an update step 51, an adaptive adjustment core 52, and a prediction step 53. In the update step 51, the observation equation, Kalman gain, accelerometer observations, magnetometer observations, and attitude are determined. In the adaptive adjustment core 52, an adaptive noise regulator dynamically adjusts the observation noise and system noise based on angular velocity, linear acceleration, and magnetic field mode. The prediction step 53 uses the state equation f to determine the attitude and can also determine the attitude and gyroscope angular velocity from the previous moment.

[0261] Figure 6 This is a flowchart of an adaptive adjustment logic decision-making method according to an embodiment of this application, such as... Figure 6 As shown, the method may include the following steps.

[0262] Step S601: Calculate motion characteristics.

[0263] In this embodiment, the current three-axis angular velocity magnitude of the vehicle is calculated in real time during each control cycle. ω |Magnetic field strength modulus| B | and the linear acceleration magnitude after removing the gravitational component As the input feature quantity for subsequent adaptive noise adjustment, it is used to sense vehicle dynamics and environmental disturbance status.

[0264] Step S602: Perform a first-order low-pass filter on the linear acceleration to obtain |a_flt|.

[0265] In this embodiment, to eliminate the interference of high-frequency vibration noise caused by road bumps on acceleration judgment, the original linear acceleration magnitude is... A first-order low-pass filter is performed to obtain the smoothed linear acceleration characteristics. This improves the stability and noise resistance of status determination.

[0266] Step S603, |a_flt|>Th_bump_on.

[0267] In this embodiment, if the above formula is satisfied, step S604 can be executed; otherwise, step S605 can be executed.

[0268] Step S604: Mark the condition as bumpy and slightly increase R_acc.

[0269] In this embodiment, when the system determines that it is in a bumpy state, the accelerometer observation noise variance R_acc is appropriately increased. This reduces the Kalman filter's confidence in accelerometer observations and prevents false gravity direction misjudgments caused by bumps from affecting attitude calculation accuracy.

[0270] Step S605, |a_flt| < Th_bump_off.

[0271] In this embodiment, if the above conditions are met, step S606 can be executed; otherwise, step S607 can be executed.

[0272] Step S606: Mark as smooth state and restore R_acc to the baseline.

[0273] In this embodiment, when the vehicle is in a stable driving state, the accelerometer observation noise is... Reset to the preset baseline value to maximize the use of gravity reference information provided by the accelerometer, effectively correct long-term gyroscope drift, and improve attitude calculation accuracy.

[0274] Step S607: Keep the previous bump status marker, and keep R_acc unchanged.

[0275] In this embodiment, when When the state is between Thbump_off and Thbump_on, the current state flag is not changed, and the flag from the previous time step is maintained. This setting helps prevent frequent fluctuations in noise parameters caused by signal volatility around a threshold, thus improving system robustness.

[0276] Step S608, triaxial angular velocity > Th_rot_on.

[0277] In this embodiment, if the triaxial angular velocity is greater than Th_rot_on, step S610 can be executed; otherwise, step S609 can be executed.

[0278] Step S609, triaxial angular velocity > Th_rot_off.

[0279] In this embodiment, if the triaxial angular velocity is greater than Th_rot_off, step S611 can be executed; otherwise, step S612 can be executed.

[0280] Step S610: Mark the state as high dynamic roll and significantly increase R_acc.

[0281] In this embodiment, once the vehicle is confirmed to be in a high-dynamic rollover state, the accelerometer will observe the noise. By significantly increasing the value (e.g., several to ten times the reference value), the filter is forced to ignore accelerometer data contaminated by centrifugal force and instead rely on gyroscope integration to maintain attitude continuity.

[0282] Step S611: Mark as a stable state and restore R_acc to the baseline.

[0283] In this embodiment, when the angular velocity drops below Thrt_off and the duration meets the stability condition, the vehicle is determined to have regained stability and immediately resumes normal operation. Once the value reaches the standard, the filter re-relies on the accelerometer for attitude correction, eliminating gyroscope drift.

[0284] Step S612: Keep the previous roll state flag unchanged, and keep R_acc unchanged.

[0285] In this embodiment, when the angular velocity is between Throt_off and Throt_on, the current state flag is maintained. The hysteresis characteristic of state switching remains unchanged, preventing frequent filter switching caused by small angular velocity oscillations and improving system stability.

[0286] Step S613, |B| deviation > threshold.

[0287] In this embodiment, if the above formula is satisfied, step S614 can be executed; otherwise, step S616 can be executed.

[0288] Step S614: Mark the state as strong magnetic interference and set R_mag=C_max.

[0289] In this embodiment, once strong magnetic interference is detected, the magnetometer observation noise variance R_mag ( Setting it to the maximum value C_max (such as 106106 times the reference value) is equivalent to softly disabling the magnetometer correction function in the filter to prevent erroneous magnetic field data from skewing the attitude calculation results.

[0290] Step S615: Reset the stability calculator T_stable.

[0291] In this embodiment, when magnetic interference is detected, the magnetic field stability timer T_stable is immediately cleared to record the cumulative time for the magnetic field strength to return to normal. This ensures that the magnetometer correction is only allowed to be restored after the interference is completely eliminated, thus avoiding erroneous restoration.

[0292] Step S616, T_stable ≥ T_stable_min.

[0293] In this embodiment, if the above formula is satisfied, step S617 can be executed; otherwise, step S618 can be executed.

[0294] Step S617: Softly restore to baseline by exponential decay R_mag.

[0295] In this embodiment, once the magnetic field stability condition is met, it decays exponentially (e.g.) ←γ , γ∈[0.9,0.99] γ Gradually reduce the magnetometer observation noise in the range [0.9, 0.99] to achieve a smooth transition and avoid sudden attitude changes caused by the sudden activation of the magnetometer.

[0296] Step S618: Keep R_mag unchanged.

[0297] In this embodiment, if the magnetic field returns to normal but does not reach the stability threshold, or if interference is detected again, the current state is maintained. The magnetometer correction remains unchanged to ensure the stability and safety of the system during the interference recovery period.

[0298] Step S619: Perform a comprehensive calculation of R_mag.

[0299] In this embodiment, all magnetometer interference judgment logics are comprehensively considered to output the final magnetometer observation noise value. This is a component of the reconstructed observation noise matrix.

[0300] Step S620: Perform a comprehensive calculation of R_acc.

[0301] In this embodiment, all accelerometer disturbance judgment logics are comprehensively considered to output the final accelerometer observation noise value. This ensures that it accurately reflects the current vehicle's dynamics and road conditions.

[0302] Step S621: Reconstruct the observation noise matrix.

[0303] In this embodiment, the updated and Substitute the diagonal observation noise covariance matrix This allows for the dynamic configuration of the Kalman filter observation model, enabling it to adaptively match the sensor reliability under the current operating conditions at every moment.

[0304] Step S622, output to AKF step.

[0305] In this embodiment, the reconstructed observation noise matrix is... The data is passed to the Adaptive Extended Kalman Filter (AKF) attitude calculation module for Kalman gain calculation and state update in the next cycle, realizing closed-loop adaptive control of perception-decision-correction, and ensuring high accuracy and robustness of attitude calculation under all operating conditions.

[0306] This application describes the mathematical construction of an adaptive extended Kalman filter (AKF) attitude calculation model, and the adaptive noise adjustment logic for solving the attitude divergence problem under high dynamic and strong interference conditions. This model dynamically adjusts the trust weights within the system by observing the vehicle's motion state and environmental interference in real time.

[0307] Quaternions are used as the attitude description method to avoid gimbal lock issues and achieve continuous description of all attitude angles. The system's state vector is defined. A 7-dimensional column vector:

[0308]

[0309] in, A unit quaternion describing the attitude of the vehicle body coordinate system relative to the navigation coordinate system (NED or ENU system), satisfying normalization constraints. ; This is a random zero-bias estimate for a three-axis gyroscope, used for online real-time compensation of the gyroscope's constant drift.

[0310] Angular velocity measured using a gyroscope Establish nonlinear state differential equations.

[0311] The measured angular velocity is compensated using the zero-bias estimate to obtain an estimate of the true angular velocity. .

[0312]

[0313] in, This represents quaternion multiplication.

[0314] Assume the zero bias is a random walk process:

[0315]

[0316] After discretization, the one-step predicted state is obtained. And one-step prediction of covariance matrix :

[0317]

[0318]

[0319] in, The state transition matrix (Jacobi matrix) Discretization) The system process noise covariance matrix describes the uncertainty of the system model.

[0320] Accelerometers and magnetometers are introduced as observation sources to correct the drift caused by gyroscope integration.

[0321] Observation vector A 6-dimensional column vector:

[0322]

[0323] in, The specific force measured by the accelerometer (equal to the gravitational vector in an ideal static and horizontal state). ); The geomagnetic field vector measured by the magnetometer. This is the local geomagnetic field reference vector; Quaternion The corresponding rotation matrix (usually defined as from the Body frame to the Nav frame). Used here. In order to make the reference vector (gravity) in the navigation frame and geomagnetism Transform to the vehicle body coordinate system to match the sensor measurements. (In Body system) Perform residual calculation directly. The statistical properties of observation noise are derived from the observation noise covariance matrix. describe.

[0324] According to Kalman gain Update status:

[0325]

[0326]

[0327] Key correction step: Since state updates violate quaternion normalization constraints, quaternion normalization must be explicitly performed after the update calculation.

[0328]

[0329] The observation noise covariance matrix is ​​reconstructed in real time using the following adaptive adjustment logic. This enables the filter to dynamically switch between trusted sensors (to correct drift) and trusted gyroscopes (to maintain continuity), and introduces low-pass filtering and hysteresis logic to improve robustness and stability.

[0330] Define the observation noise matrix For diagonal matrices:

[0331]

[0332] in, and These are the dynamic variances of the accelerometer and the magnetometer, respectively.

[0333] In each control cycle (e.g., 2ms), the system calculates the following characteristic quantities based on the raw sensor data.

[0334] Triaxial angular velocity magnitude: ;

[0335] Triaxial magnetic field strength magnitude: ;

[0336] The linear acceleration magnitude after removing gravity: .

[0337] To avoid misjudgment of bump detection due to high-frequency noise, a first-order low-pass filter is applied to the linear acceleration magnitude to obtain smoothed linear acceleration features. :

[0338]

[0339] in These are the filter coefficients, which can be determined based on the sampling period. and expected time constant approximate Typical value range is approximately .

[0340] This application provides the following strategy one, which is based on high dynamic adjustment (with hysteresis) of angular velocity magnitude.

[0341] Physical principle: When When the magnitude is large, the vehicle is in a state of violent rollover, and the centrifugal acceleration caused by the IMU installation position being off-center from the center of gravity will seriously interfere with the accelerometer's observation of the direction of gravity.

[0342] Hysteresis detection: Set an upper threshold With lower threshold ,satisfy For example, it is advisable and .

[0343] like If so, then the current moment is marked as a high-dynamic roll state;

[0344] like If so, then the current time is marked as a stable state;

[0345] like If the state falls between these two extremes, then the state marker from the previous time step remains unchanged.

[0346] Noise adjustment: When in a high-dynamic tumbling state, increase the noise observed by the accelerometer. For example, adjust according to the following formula:

[0347]

[0348] in, This is the adjustment coefficient. When in a stable state, the recovery... To the benchmark value Maintain the current state within the hysteresis interval. Unchanged, thus avoiding The noise parameters frequently jump around when fluctuating slightly around the threshold.

[0349] This application provides the following strategy two, based on the interference detection and adjustment of the magnetic field modulus (including a stabilization timer and soft recovery).

[0350] Interference determination: Assume the local standard geomagnetic field strength is... The allowable deviation range is (For example, 30%).

[0351] like Exceeding the range If so, it is determined to be strong magnetic interference.

[0352] Soft disable: When determined to be strong magnetic interference, set the magnetometer observation noise to the maximum value. (e.g., reference noise) (times), equivalent to soft disabling magnetic field correction.

[0353] Stability Timer and Soft Recovery: Introducing a Stability Timer It is used to record the length of time that the magnetic field magnitude remains within the normal range.

[0354] Reset immediately upon detection of interference. ;when Once it returns to the normal range, the sum will be accumulated. ;when The interference is considered to have truly disappeared only after 500ms (for example), and then the process begins. Perform exponentially decaying soft recovery:

[0355]

[0356] in For example, the attenuation coefficient can be taken as... If in Not achieved If interference is detected again, it will be reclassified as strong magnetic interference and the timer will be reset.

[0357] This application provides the following strategy three: road bump adjustment based on linear acceleration (low-pass filtering + hysteresis).

[0358] Physical principle: In scenarios where there is no rollover but the road surface is rough, the high-frequency vibration of the vehicle body causes the linear acceleration magnitude to fluctuate greatly. Although the mean is zero, the variance is large. If it is used directly for noise adjustment, it will cause Kalman gain jitter.

[0359] Hysteresis detection: for low-pass filtered... A dual-threshold hysteresis comparison is used: upper threshold lower threshold ,and ;like If so, then mark the current time as a turbulent state; if If so, then mark the current time as a smooth state; if If the state falls between these two extremes, then the state marker from the previous time step remains unchanged.

[0360] Noise adjustment: When in a bumpy environment, appropriately increase the accelerometer noise. For example, by slightly enlarging it using the following formula:

[0361]

[0362] in, For smaller positive constants, for example, can be taken as When in a smooth state, restore To the baseline value; maintain within the hysteresis interval Unchanged, to avoid due to Small fluctuations around the threshold cause the noise parameters to switch frequently.

[0363] In this embodiment, combining the above three strategies, the system adaptively adjusts the observation noise matrix according to the following steps in each control cycle. Initialize the baseline noise variance , and various thresholds and timers; calculation , , And on Perform low-pass filtering to obtain The current accelerometer observation noise is determined jointly by Strategy 1 (high dynamic hysteresis) and Strategy 3 (bump hysteresis). The current magnetometer observation noise is determined based on Strategy Two (magnetic field stabilization timer and soft recovery). ;Will , Substitute into the diagonal matrix And use the updated version in subsequent EKF iterations. Calculate Kalman gain and status update By introducing the adaptive adjustment mechanism of low-pass filtering and hysteresis logic, the AKF model of this application embodiment can make full use of the observation information of accelerometer and magnetometer to eliminate gyroscope drift when the vehicle is driving smoothly. Under complex conditions such as high dynamic roll, strong magnetic interference or road bumps, it can suppress instantaneous interference and jitter near the threshold by increasing the observation noise and maintaining state hysteresis, thereby ensuring high accuracy and high robustness of attitude calculation in the entire working range.

[0364] Optionally, the above three strategies form a complete adaptive adjustment mechanism: initialization, setting the baseline noise variance. and Real-time calculation, performed every control cycle (e.g., 2ms). Noise reconstruction: Priority judgment; if magnetic interference is detected (Strategy 2), force... Dynamic judgment, combined with and Calculate the current EKF iteration, substituted with the updated version Calculate Kalman gain and status update .

[0365] By using the above-described method of real-time dynamic adjustment of observation noise, the AKF model of this application embodiment can maximize the use of absolute reference information from accelerometers and magnetometers when the vehicle is driving smoothly, eliminate gyroscope integral drift, and ensure long-term accuracy. When the vehicle rolls over, is airborne, or is subjected to strong magnetic interference, it can intelligently shield unreliable sensors and rely on high-bandwidth gyroscopes to maintain the continuity and robustness of attitude, thereby achieving high-precision roll recognition under all working conditions.

[0366] Figure 7 This is a schematic diagram of a graded ignition control state transition according to an embodiment of this application, such as... Figure 7 As shown, after successful initialization, the vehicle's status during operation can be monitored. If normal vehicle operation is detected, monitoring can continue. Conversely, if signs of airborne injury are detected, a pre-standby state can be entered. During the pre-standby phase, alert can be maintained, and seatbelt warnings will be issued. Conversely, if the signs of airborne injury disappear, monitoring of the vehicle's status can resume. If it is determined that an irreversible rollover is imminent, the ignition state can be entered, the airbags can be deployed, and the vehicle will enter post-collision status.

[0367] Figure 8 This is a flowchart of a sensor fault diagnosis and system degradation processing method according to an embodiment of this application, such as... Figure 8 As shown, the method may include the following steps.

[0368] Step S801: Check the nine-axis IMU.

[0369] In this embodiment, the fault diagnosis process is initiated. First, the integrity, validity, and rationality of the signals from the three-axis accelerometer, three-axis gyroscope, and three-axis magnetometer of the nine-axis inertial measurement unit (IMU) are comprehensively tested to determine whether there are any abnormalities such as signal loss, exceeding limits, deadlock, or drift, which serve as the basis for determining the degradation strategy.

[0370] Optionally, a rationality check can be continuously performed on the sensor signals during operation. This includes a signal range check: detecting whether acceleration, angular velocity, and magnetic field strength exceed physical limits (e.g., acceleration). or magnetic field strength Signal stationary check: Detects whether the signal remains unchanged for an extended period (dead code). Signal drift check: Detects whether the zero bias exceeds the calibration range while stationary. Interpolation check: Compares the same physical quantity calculated by different sensors (e.g., vehicle speed calculated from wheel speed integral vs. vehicle speed calculated from acceleration integral); if the deviation is too large, the sensor is marked as suspicious.

[0371] Optionally, when a sensor malfunction is detected, the system immediately switches to the corresponding degraded mode and forcibly modifies the parameters to safe values ​​via a software interface.

[0372] Step S802: Magnetometer signal is normal.

[0373] In this embodiment, if the magnetometer signal is normal, step S803 can be executed; otherwise, if the magnetometer signal is not normal, step S804 can be executed.

[0374] Step S803: Gyroscope / accelerometer is normal.

[0375] In this embodiment, if the gyroscope / accelerometer is working properly, step S805 can be executed; otherwise, step S806 can be executed.

[0376] Step S804: Enter the first downgrade mode.

[0377] In this embodiment, when the magnetometer signal is abnormal, the system enters the first degrade mode: the magnetometer correction function is disabled, and attitude calculation is performed only by the accelerometer and gyroscope. At the same time, the roll recognition threshold is increased to compensate for the risk of heading drift caused by the lack of geomagnetic reference and maintain basic roll recognition capability.

[0378] Optionally, the first degradation mode can be triggered by magnetometer failure or strong magnetic interference. Triggering conditions: magnetometer signal loss, error reporting, or the soft recovery logic in Example 3 failing to recover the observation noise for a continuous period of time. Return to normal levels. Degradation strategy: parameter locking, forcibly reducing the AKF observation noise matrix. Magnetometer component in Locked to infinity (or system maximum), disabling magnetic field correction. Drift compensation relies solely on accelerometer correction for gyroscope drift. Since the accelerometer cannot correct for heading, the system allows for greater attitude uncertainty over longer time windows. Threshold adjustment reduces the dependence of roll ignition on heading and improves the effect based on roll angle. Improve ignition sensitivity, or appropriately increase it. To compensate for potential long-term heading drift errors.

[0379] Step S805: The number of wheels with normal wheel speed sensor readings is ≥2.

[0380] In this embodiment, if the number of wheels with normal speed sensor is ≥2, the test can proceed to step S808; otherwise, step S807 can be performed.

[0381] Step S806: Enter the third downgrade mode.

[0382] In this embodiment, when either the gyroscope or the accelerometer malfunctions (such as signal loss or continuous saturation), the system enters a third degradation mode: completely disabling the attitude-based rollover recognition algorithm and retaining only the most basic collision detection function (such as direct acceleration threshold judgment) to ensure that the airbag can still respond in a collision, but loses its rollover protection capability.

[0383] Optionally, the third degradation mode is for core inertial sensor failures (gyroscope / accelerometer). Triggering conditions: gyroscope or accelerometer hardware failure, severe signal saturation, or complete signal loss. Degradation strategy: Disable the roll algorithm; due to the AKF model's inability to run (state prediction equation failure), the system forcibly disables the attitude calculation module and fusion logic. Basic collision protection is retained, preserving accelerometer-based direct collision detection algorithms (such as integral window algorithms) for triggering frontal / side collision airbags. If the roll function fails, the roll ignition path for all side curtain airbags is shut down to prevent accidental detonation due to erroneous attitude data. An alarm is triggered, immediately illuminating the Supplemental Restraint System (SRS) fault indicator and storing a Diagnostic Trouble Code (DTC).

[0384] Step S807: Enter the second downgrade mode.

[0385] In this embodiment, when the number of normal wheel speed sensors is less than two (e.g., only one or all are malfunctioning), but the IMU is still normal, the system enters the second degrade mode: the first ground-lift risk factor based on wheel speed is disabled, the wheel-lift prediction function is abandoned, and the roll judgment is made entirely by relying on the attitude calculation results of the nine-axis IMU, thereby reducing the system's dependence on wheel speed.

[0386] Optionally, the second degradation mode can be configured for wheel speed sensor failure. Triggering condition: Abnormal wheel speed signals from one or more wheels (e.g., ...). However, if the vehicle speed is high, or the signal is lost, the degradation strategy is to adjust the fusion weights and force them to be set in the fusion formula. Disable the first takeoff risk factor based on wheel speed. Some functions are retained; if at least one non-drive wheel (reference wheel) is functioning correctly, the system can continue to calculate the reference vehicle speed. However, it does not perform ground clearance determination; it is used for collision energy estimation. If all wheel speeds fail, attitude recognition relies entirely on the nine-axis IMU.

[0387] Step S808: Calculate V_ref using normal wheel speed.

[0388] In this embodiment, assuming at least two wheel speed sensors are functioning normally, the system selects the minimum effective signal value from the non-drive wheels as the reference vehicle speed V_ref. This is used for subsequent wheel speed deviation calculation and takeoff risk factor generation, ensuring that high reliability of takeoff prediction capability can still be maintained even with the failure of some sensors.

[0389] Step S809: Output to the fusion and ignition module.

[0390] In this embodiment, information such as the current degraded mode state of the system, the available sensor combinations, the adjusted fusion weights, and the safety thresholds are transmitted in real time to the upper-layer multi-sensor adaptive fusion module and the graded ignition decision module to ensure that subsequent risk calculations and airbag control strategies always match the current sensor health status and guarantee functional safety.

[0391] In this embodiment, the above-described fault diagnosis and degradation logic ensures that the adaptive algorithm and fusion decision are closely monitored, and the failure of any single sensor will not lead to unpredictable system behavior, thereby guaranteeing the functional safety of the system at the hardware level.

[0392] Figure 9(a) is a schematic diagram of an integrated architecture according to an embodiment of this application. As shown in Figure 9(a), the integrated architecture may include a nine-axis IMU (located on the internal PCB of the ACU) 91, an ACU core MCU (containing the AKF algorithm) 92, and an ignition drive circuit 93. The nine-axis IMU 91 and the ACU core MCU 92 are directly connected via SPI / I2C. The ACU core MCU 92 and the ignition drive circuit 93 are connected via an internal bus. The computing unit is directly integrated on the main PCB of the ACU. The nine-axis IMU is also soldered inside the ACU and communicates directly with the microcontroller unit (MCU) via a Serial Peripheral Interface (SPI) or I2C interface. This architecture features the fastest response time (<1ms), requires no in-vehicle network bandwidth, and has the lowest cost. Note that since the IMU is fixed at the ACU, if the ACU is installed off-center from the vehicle's center of gravity, the algorithm needs to introduce "lever arm compensation" logic to calculate the centrifugal acceleration using wheel speed and steering angle to eliminate measurement errors caused by IMU offset.

[0393] Figure 9(b) is a schematic diagram of a distributed architecture according to an embodiment of this application. As shown in Figure 9(b), the architecture may include: a nine-axis IMU (located at the vehicle's center of gravity) 94, wheel speed sensors 95, an in-vehicle Ethernet / CAN-FD 96, a central domain controller (running the AKF algorithm) 97, and a traditional ACU (only performing ignition) 98. The nine-axis IMU, as an independent intelligent sensor, is installed near the vehicle's center of gravity (to obtain optimal rigid body motion data). IMU data is sent to the central domain controller via in-vehicle Ethernet or CAN-FD. The System on a Chip (SoC) within the domain controller runs the AKF algorithm, and the calculated ignition command is sent to the ACU for execution via network or hardwired connection. Due to physical separation, the IEEE 1588 (PTP) protocol must be used to achieve nanosecond-level time synchronization between the IMU and the domain controller. The IMU adds a hardware timestamp at the acquisition time, and the domain controller aligns the data to the correct time for calculation based on network latency; otherwise, network jitter will cause attitude integration errors. It boasts powerful computing capabilities, supports over-the-air (OTA) upgrades, easily integrates intelligent driving / chassis data, and adapts to new electrical / electronic (E / E) architectures.

[0394] Figure 9(c) is a schematic diagram of a hybrid architecture according to an embodiment of this application. As shown in Figure 9(c), the architecture may include: a nine-axis IMU 99, an independent fusion module (running the AKF algorithm) 910, and a traditional ACU (performing ignition) 911. The computing unit is encapsulated in an independent fusion module. This independent fusion module receives IMU and wheel speed data via CAN / FlexPay and outputs attitude angles and risk values ​​to the traditional ACU. This approach balances increased computing power with compatibility with the existing supply chain.

[0395] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0396] According to another aspect of the embodiments of this application, and the above... Figure 2 Corresponding to the embodiments of the vehicle control method shown, this specification also provides a vehicle control system.

[0397] Figure 10 This is a schematic diagram of a vehicle control system according to an embodiment of this application, such as... Figure 10 As shown, the vehicle control system 100 may include: a sensing module 1002 for acquiring vehicle operating status information; a control and calculation unit 1004 for determining at least one risk information of the vehicle based on the operating status information, wherein the risk information is used to characterize the risk level of the vehicle's wheels losing external support and / or the vehicle rolling over; determining the vehicle's safety performance index based on at least one risk information, wherein the safety performance index is used to characterize the vehicle's safety level under the condition that the wheels lose external support and / or the vehicle rolls over; and an execution module 1006 for controlling the vehicle's operation according to a control strategy matched with the safety performance index, wherein the safety level of the vehicle controlled according to the control strategy is greater than the safety level of the vehicle to be controlled.

[0398] The embodiments of this application will be described in detail below with reference to the steps described above.

[0399] As an optional implementation, the system further includes: a fault diagnosis module, used to trigger a degradation control strategy in response to detecting that the operating state of the sensing module is in a fault state, wherein the degradation control strategy is used to represent the rules for maintaining the safe operating state of the system.

[0400] According to another aspect of the embodiments of this application, and the above... Figure 2Corresponding to the embodiments of the vehicle control method shown, this specification also provides a vehicle control device.

[0401] Figure 11 This is a schematic diagram of a vehicle control device according to an embodiment of this application, such as... Figure 11 As shown, the vehicle control device 110 may include: an acquisition module 1102, a first determination module 1104, a second determination module 1106, and a control module 1108. The acquisition module 1102 is used to acquire the vehicle's operating status information; the first determination module 1104 is used to determine at least one risk information of the vehicle based on the operating status information, wherein the risk information characterizes the degree of risk of the vehicle's wheels losing external support and / or the vehicle rolling over; the second determination module 1106 is used to determine the vehicle's safety performance index based on at least one risk information, wherein the safety performance index characterizes the degree of safety of the vehicle under the condition that the wheels lose external support and / or the vehicle rolls over; the control module 1108 is used to control the vehicle's operation according to a control strategy matched to the safety performance index, wherein the safety degree of the vehicle controlled according to the control strategy is greater than the safety degree of the vehicle to be controlled.

[0402] Embodiments of this application also provide a vehicle, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods described in various embodiments of this application when it runs.

[0403] Figure 12 This is a schematic diagram of an electronic device according to an embodiment of this application, such as... Figure 12 As shown, an electronic device 120 is also provided, including: a memory 1201 storing an executable program; and a processor 1202 for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0404] Embodiments of this application also provide a computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0405] Embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0406] Embodiments of this application also provide a computer program product, including a non-volatile computer-readable storage medium for storing a computer program that, when executed by a processor, implements the methods in various embodiments of this application.

[0407] Embodiments of this application also provide a computer program that, when executed by a processor, implements the methods described in the various embodiments of this application.

[0408] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0409] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection can be through some interfaces; the indirect coupling or communication connection between units or modules can be electrical or other forms.

[0410] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0411] Furthermore, 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. The integrated unit can be implemented in hardware or as a software functional unit.

[0412] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a 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 all or part 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, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0413] The above description is the preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for controlling a vehicle, characterized in that, include: Obtain the vehicle's operating status information; Based on the operating status information, at least one risk information of the vehicle is determined, wherein the risk information is used to characterize the degree of risk of the vehicle's wheels losing external support and / or the risk of the vehicle rolling over. Based on the at least one risk information, a safety performance index of the vehicle is determined, wherein the safety performance index is used to characterize the safety level of the vehicle in the event that the wheels lose the external support force and / or the vehicle is at risk of rollover; The vehicle is controlled to operate according to a control strategy that matches the safety performance indicators, wherein the safety level of the vehicle controlled according to the control strategy is greater than the safety level of the vehicle to be controlled.

2. The method according to claim 1, characterized in that, The determination of at least one risk information of the vehicle based on the operational status information includes: Based on the wheel rotation speed information in the operating status information, a first risk information is determined, wherein the first risk information is used to indicate the degree of risk of the vehicle losing the external support force; And / or, Based on the three-axis information in the operating status information, second risk information is determined, wherein the three-axis information is used to represent the vehicle's motion posture in three-dimensional space and / or the interaction state between the vehicle and the magnetic field of the three-dimensional space, and the second risk information is used to represent the degree of risk of the vehicle rolling over when the wheels lose the external support force.

3. The method according to claim 2, characterized in that, The determination of the first risk information based on the wheel rotation speed information in the operating status information includes: Based on the rotational speed information, the rotational speed difference characteristics of the wheel are determined, wherein the rotational speed difference characteristics are used to represent the degree of deviation of the wheel's rotational speed abnormality caused by load transfer or displacement; Based on the aforementioned speed difference characteristics, the first risk information is determined.

4. The method according to claim 3, characterized in that, The step of determining the rotational speed difference characteristics of the wheels based on the rotational speed information includes: Based on the rotational speed information of the non-driving wheels in the wheel set of the vehicle, the reference speed of the vehicle is determined; Based on the reference vehicle speed, the vehicle's steering angle and wheelbase in the operating status information, the theoretical speed difference of the wheels is determined; Using the theoretical speed difference, the difference between the speed information and the reference vehicle speed is corrected to obtain the corrected difference; The corrected difference is used to determine the speed difference feature, wherein the speed difference feature and the first risk information are positively correlated. And / or, The determination of the first risk information based on the speed difference characteristics includes: In response to the speed difference feature exceeding the speed difference feature threshold, the first risk information is output.

5. The method according to claim 2, characterized in that, The determination of the second risk information based on the three-axis information in the operational status information includes: The three-axis information is input into the vehicle's attitude calculation model. Using the attitude calculation model, the roll angle and pitch angle of the vehicle are calculated from the three-axis information. The second risk information is determined based on the roll angle and the pitch angle.

6. The method according to claim 5, characterized in that, The step of using the attitude calculation model to calculate the roll angle and pitch angle of the vehicle from the three-axis information includes: After inputting the triaxial information into the attitude calculation model, the current attitude of the vehicle is predicted by using the angular velocity measured by the gyroscope in the attitude calculation model through a quaternion integration strategy, and the angular velocity is compensated by using the zero bias estimate of the gyroscope to obtain the actual angular velocity. By combining the gravity component measured by the accelerometer and the geomagnetic field component measured by the magnetometer in the attitude calculation model, the current attitude is corrected to obtain the corrected current attitude. During the process of correcting the current attitude, the changes in the three-axis information are monitored, wherein the changes are used to characterize whether the vehicle is in a complex working condition. Based on the changed state, the corrected current posture is adjusted to obtain the adjusted current posture; The roll angle and pitch angle are calculated from the adjusted current attitude.

7. The method according to claim 5, characterized in that, The determination of the second risk information based on the roll angle and the pitch angle includes: Determine the angular rate of change of the roll angle and the pitch angle, wherein the angular rate of change is used to characterize the severity of the risk of rollover or pitch of the vehicle. The second risk information is determined based on the change in angular rate.

8. The method according to any one of claims 1 to 7, characterized in that, The method further includes: Based on the operating status information, the vehicle's runaway information is determined, wherein the runaway information is used to represent the degree of energy mutation of the vehicle; as well as, Based on the operational status information and the ground status information, the driving conditions of the vehicle are determined; Based on the driving conditions, determine the weights corresponding to the loss of control information and the at least one risk information, respectively. Determining the vehicle's safety performance indicators based on the at least one piece of risk information includes: The safety performance index is obtained by using the weights corresponding to the out-of-control information and the at least one risk information, respectively.

9. The method according to any one of claims 1 to 7, characterized in that, The method further includes: In response to the safety performance index being greater than or equal to a first safety performance index threshold and the safety performance index being less than a second safety performance index threshold, the control strategy matching the safety performance index is determined to be a first control strategy, wherein the first control strategy is used to represent the rules for controlling the seat belts and / or the side curtain airbags in the vehicle. In response to the safety performance index being greater than or equal to the second safety performance index threshold, the control strategy matching the safety performance index is determined as the second control strategy, wherein the safety level of the vehicle to be controlled according to the second control strategy is less than the safety level of the vehicle to be controlled according to the first control strategy, and the second control strategy is used to represent the rules for controlling the airbags in the vehicle. In response to the security performance index being less than the first security performance index threshold, the process returns to start executing from the following steps: Obtain the aforementioned operating status information.

10. The method according to claim 9, characterized in that, The control strategy, which matches the safety performance indicators, controls the vehicle's operation, including: In response to the control strategy being the first control strategy, the seat belt is tightened, and / or the side curtain airbags are inflated; In response to the control strategy being the second control strategy, the airbag is detonated.

11. The method according to claim 10, characterized in that, The method further includes: If the safety level of the vehicle operating according to the first control strategy decreases after the tightening operation and / or the inflation operation are completed, the airbag is detonated according to the second control strategy.

12. A vehicle control system, characterized in that, include: A sensing module is used to acquire the vehicle's operating status information; A control and calculation unit is configured to determine at least one risk information of the vehicle based on the operating status information, wherein the risk information is used to characterize the risk level of the vehicle's wheels losing external support and / or the vehicle experiencing a rollover risk; and to determine the vehicle's safety performance index based on the at least one risk information, wherein the safety performance index is used to characterize the safety level of the vehicle in the event that the wheels lose the external support and / or the vehicle experiences a rollover risk. An execution module is used to control the operation of the vehicle according to a control strategy matched with the safety performance indicators, wherein the safety level of the vehicle controlled according to the control strategy is greater than the safety level of the vehicle to be controlled.

13. The system according to claim 12, characterized in that, The system also includes: The fault diagnosis module is used to trigger a degradation control strategy in response to the detection that the operating state of the sensing module is in a fault state. The degradation control strategy is used to represent the rules for maintaining the safe operating state of the system.

14. A vehicle control device, characterized in that, include: The acquisition module is used to acquire the operating status information of the vehicle; The first determining module is used to determine at least one risk information of the vehicle based on the operating status information, wherein the risk information is used to characterize the degree of risk of the vehicle's wheels losing external support and / or the risk of the vehicle rolling over. The second determining module is used to determine the safety performance index of the vehicle based on the at least one risk information, wherein the safety performance index is used to characterize the safety level of the vehicle in the event that the wheels lose the external support force and / or the vehicle is at risk of rollover. The control module is used to control the operation of the vehicle according to a control strategy matched with the safety performance indicators, wherein the safety level of the vehicle controlled according to the control strategy is greater than the safety level of the vehicle to be controlled.

15. An electronic device, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the method according to any one of claims 1 to 11.

16. A vehicle, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the method according to any one of claims 1 to 11.

Images