Active and passive safety integrated object control method, device and computer equipment

Abstract

The application relates to an active and passive safety fusion object control method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring an image data set and a point cloud data set outside a vehicle, and determining a to-be-collision object and an object feature set of the to-be-collision object based on the image data set and the point cloud data set; performing prediction processing on the object feature set and a control signal set of the vehicle according to a collision working condition prediction model to obtain a collision working condition feature set; acquiring passenger feature sets of passengers in the vehicle, and determining the ignition time of each airbag based on the collision working condition feature set and the passenger feature sets. The method can improve the accuracy of the object control method.

Description

Technical Field

[0001] This application relates to the field of automotive technology, and in particular to an object control method, device, computer equipment, computer-readable storage medium, and computer program product that integrates active and passive safety. Background Technology

[0002] Airbags are a crucial component of a vehicle's passive safety system. During normal driving, airbag deployment is controlled via object control methods to protect the safety of passengers and improve their survival rate in collisions.

[0003] Current airbag deployment methods acquire vehicle acceleration data from various sensors within the vehicle and determine whether each acceleration meets preset airbag deployment conditions. If the acceleration meets the deployment conditions, an ignition signal is sent to the airbag system to rapidly generate gas to inflate the airbag (object), thereby causing the airbag to deploy.

[0004] However, current object control methods rely solely on the acceleration after a collision to determine whether the airbag should be detonated. This results in a single determining factor for detonation and inaccurate detonation timing, leading to low accuracy in current object control methods. Summary of the Invention

[0005] Therefore, it is necessary to provide an object control method, device, computer equipment, computer-readable storage medium, and computer program product that integrates active and passive security to address the above-mentioned technical problems.

[0006] Firstly, this application provides an object control method that integrates active and passive security, including:

[0007] Acquire image datasets and point cloud datasets of the vehicle exterior, and determine the object to be collided with and the object feature set of the object to be collided with based on the image datasets and the point cloud datasets;

[0008] The collision condition feature set is obtained by performing prediction processing on the object feature set and the vehicle control signal set according to the collision condition prediction model.

[0009] Obtain the passenger feature set of each passenger in the vehicle, and determine the deployment time of each airbag based on the collision condition feature set and each passenger feature set.

[0010] In one embodiment, determining the object to be collided and the object feature set of the object to be collided based on the image dataset and the point cloud dataset includes:

[0011] Based on the image recognition model, image recognition is performed on each image data in the image dataset to obtain the object to be collided with and the initial object feature set of the object to be collided with.

[0012] The velocity and volume of the object to be collided are determined based on the point cloud dataset and the image dataset.

[0013] The initial object feature set is updated based on the velocity and volume of the object to be collided with, thus obtaining the object feature set.

[0014] In one embodiment, the step of performing image recognition on each image data in the image dataset based on an image recognition model to obtain the object to be collided with and the initial object feature set of the object to be collided with includes:

[0015] Based on the image recognition model, image recognition is performed on each image data in the image dataset to obtain each initial collision object and the category of each initial collision object;

[0016] According to the preset collision filtering conditions, select collision objects from each of the initial collision objects;

[0017] Based on the category of the object to be collided, the quality of the object to be collided is queried in the database, and an initial object feature set of the object to be collided is constructed according to the category and quality of the object to be collided.

[0018] In one embodiment, obtaining the passenger feature set of each passenger in the vehicle includes:

[0019] The attribute feature set of each passenger is obtained through the seat sensors inside the vehicle;

[0020] The posture feature set of each passenger is obtained based on image acquisition equipment and / or infrared sensors;

[0021] Construct passenger feature sets for each passenger based on each attribute feature set and each posture feature set.

[0022] In one embodiment, determining the deployment time of each airbag based on the collision condition feature set and each of the passenger feature sets includes:

[0023] Based on the vehicle's speed, acceleration, and a first distance between the vehicle and the object to be collided with, the first collision time between the vehicle and the object to be collided with is determined.

[0024] The initial collision acceleration is determined based on the preset collision process time, the speed difference in the collision condition feature set, the object feature set of the object to be collided with, and the attribute information of the vehicle.

[0025] For each passenger, a second collision time is determined based on the passenger feature set, the initial collision acceleration, and the collision pose in the collision condition feature set; the second collision time is the time from the passenger's head and neck to the vehicle interior components.

[0026] The deployment time of the airbag corresponding to the passenger is determined based on the first collision time and the second collision time.

[0027] In one embodiment, determining the second collision time for each passenger based on the passenger feature set, the initial collision acceleration, and the collision pose in the collision condition feature set includes:

[0028] For each passenger, a second distance between the passenger's head and neck and the device in the vehicle is determined based on the posture parameters and position vector in the passenger feature set of the passenger.

[0029] The initial collision acceleration is corrected based on the seat belt restraint state and basic attributes in the passenger feature set and the collision pose in the collision condition feature set to obtain the collision acceleration.

[0030] According to the collision time algorithm, the collision acceleration and the second distance are processed to obtain the second collision time.

[0031] In one embodiment, after determining the deployment time of each airbag based on the collision condition feature set and each of the passenger feature sets, the method further includes:

[0032] Determine whether the airbags corresponding to each passenger need to be deployed based on the passenger feature set and the collision condition feature set;

[0033] When a target airbag for a target passenger needs to be detonated, a target ignition signal is generated based on the target detonation mode and detonation time corresponding to the target passenger.

[0034] The target ignition signal is sent to the airbag device, instructing the target airbag corresponding to the target passenger to deploy.

[0035] Secondly, this application also provides an object control device that integrates active and passive safety features, comprising:

[0036] The acquisition module is used to acquire image datasets and point cloud datasets of the vehicle exterior, and to determine the object to be collided with and the object feature set of the object to be collided with based on the image datasets and the point cloud datasets;

[0037] The prediction module is used to perform prediction processing on the object feature set and the vehicle control signal set according to the collision condition prediction model to obtain the collision condition feature set.

[0038] The determination module is used to acquire the passenger feature set of each passenger in the vehicle, and determine the deployment time of each airbag based on the collision condition feature set and each passenger feature set.

[0039] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0040] Acquire image datasets and point cloud datasets of the vehicle exterior, and determine the object to be collided with and the object feature set of the object to be collided with based on the image datasets and the point cloud datasets;

[0041] The collision condition feature set is obtained by performing prediction processing on the object feature set and the vehicle control signal set according to the collision condition prediction model.

[0042] Obtain the passenger feature set of each passenger in the vehicle, and determine the deployment time of each airbag based on the collision condition feature set and each passenger feature set.

[0043] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0044] Acquire image datasets and point cloud datasets of the vehicle exterior, and determine the object to be collided with and the object feature set of the object to be collided with based on the image datasets and the point cloud datasets;

[0045] The collision condition feature set is obtained by performing prediction processing on the object feature set and the vehicle control signal set according to the collision condition prediction model.

[0046] Obtain the passenger feature set of each passenger in the vehicle, and determine the deployment time of each airbag based on the collision condition feature set and each passenger feature set.

[0047] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0048] Acquire image datasets and point cloud datasets of the vehicle exterior, and determine the object to be collided with and the object feature set of the object to be collided with based on the image datasets and the point cloud datasets;

[0049] The collision condition feature set is obtained by performing prediction processing on the object feature set and the vehicle control signal set according to the collision condition prediction model.

[0050] Obtain the passenger feature set of each passenger in the vehicle, and determine the deployment time of each airbag based on the collision condition feature set and each passenger feature set.

[0051] The aforementioned active and passive safety integrated object control method, device, computer equipment, computer-readable storage medium, and computer program product acquire image datasets and point cloud datasets of the vehicle's exterior, and determine the object to be collided with and its object feature set based on the image datasets and point cloud datasets; perform predictive processing on the object feature set and the vehicle's control signal set according to a collision condition prediction model to obtain a collision condition feature set; acquire passenger feature sets of each passenger inside the vehicle, and determine the deployment time of each airbag based on the collision condition feature set and each passenger feature set. Using this method, the collision condition feature set is determined through the object feature set of the object to be collided with, and the deployment time of the airbag corresponding to each passenger is determined based on the collision condition feature set and passenger feature sets. Compared to directly deploying airbags based on acceleration, this method considers three influencing factors: the object to be collided with, the collision state of the vehicle, and the passenger state, resulting in an accurate deployment time. Therefore, the airbags are deployed based on the accurate deployment time, improving the accuracy of the object control method. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a schematic diagram of the control process of a conventional object control method in an exemplary embodiment;

[0054] Figure 2 This is a schematic diagram illustrating the airbag action time under a conventional object control method in an exemplary embodiment.

[0055] Figure 3 This is a flowchart illustrating an object control method that integrates active and passive security in one embodiment.

[0056] Figure 4 This is a structural diagram of a collision condition prediction model in an exemplary embodiment;

[0057] Figure 5 This is a schematic diagram illustrating the operation of a collision condition prediction model in an exemplary embodiment.

[0058] Figure 6 This is a schematic diagram of the input and output of a collision condition model in an exemplary embodiment;

[0059] Figure 7 This is a flowchart illustrating the process of determining the object to be collided with and the object feature set of the object to be collided with in one embodiment;

[0060] Figure 8 This is a schematic diagram of the input and output of a multimodal recognition model in an exemplary embodiment;

[0061] Figure 9 This is a schematic diagram of the process for determining the object to be collided in one embodiment;

[0062] Figure 10 This is a schematic diagram illustrating the operation flow of an image recognition model in an exemplary embodiment;

[0063] Figure 11 This is a schematic diagram of the process for obtaining a passenger feature set in one embodiment;

[0064] Figure 12 This is a flowchart illustrating the process of determining the detonation time in one embodiment;

[0065] Figure 13 This is a flowchart illustrating the process of determining the second collision time in one embodiment;

[0066] Figure 14 This is a schematic diagram of the process for generating a target ignition signal in one embodiment;

[0067] Figure 15 This is a schematic diagram illustrating the input and output of an airbag detonation model in an exemplary embodiment.

[0068] Figure 16 This is a schematic diagram of the architecture of a control system in an exemplary embodiment;

[0069] Figure 17 This is a structural block diagram of an object control device that integrates active and passive safety in one embodiment;

[0070] Figure 18 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0071] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0072] Every year, numerous road traffic accidents occur globally, causing significant loss of life and property. Vehicle collision safety and participant protection are cornerstones of research aimed at improving road safety. Airbags are a crucial component of a vehicle's passive safety systems. Airbags can improve the survival rate of occupants in collisions. Therefore, during routine vehicle operation, object control methods are needed to manage airbag deployment to protect occupants and enhance their survival rate in collisions.

[0073] Current airbag deployment methods acquire vehicle acceleration data from various sensors within the vehicle and determine whether each acceleration meets preset airbag deployment conditions. If the acceleration meets the deployment conditions, an ignition signal is sent to the airbag system to rapidly generate gas to inflate the airbag (object), thereby causing the airbag to deploy.

[0074] Specifically, Figure 1 This is a schematic diagram of the control process of a conventional object control method in an exemplary embodiment. For example... Figure 1 As shown, the vehicle is equipped with multiple airbag sensors to monitor collision conditions. These sensors include the airbag's own safety sensor, the front airbag sensor, and the center airbag sensor. Figure 1 The system also includes an Electronic Control Unit (ECU). When a collision occurs, airbag sensors detect changes in acceleration and then determine the severity of the collision based on the acceleration. When the sensors detect a sufficiently strong collision signal, i.e., the acceleration reaches the airbag deployment condition, the control unit (ACU, Airbag Control Unit) activates the airbag system, ignites the gas generator, and rapidly inflates the airbag. The gas generator typically uses a chemical reaction or electric ignition to produce gas, which rapidly fills the airbag. The gas is usually nitrogen or other inert gases. After inflation, the airbag rapidly expands and fills the vehicle's interior space, forming a buffer zone to slow the occupants' forward momentum and reduce the force of the direct impact. After the collision, the airbag rapidly deflates through a small hole or valve, allowing the passenger to safely exit the airbag and return to its normal position.

[0075] However, current object control methods rely solely on the acceleration after a collision to determine whether the airbag should be detonated. This results in a single determining factor for detonation and inaccurate detonation timing, leading to low accuracy in current object control methods.

[0076] Specifically, current airbag deployment methods rely on the instantaneous acceleration after a collision to determine the timing and force of airbag deployment, which often has a lag. Especially in the moments before a collision, since the entire collision process lasts only 200 ms, it is impossible to adjust the airbag deployment strategy in real time after the collision. The effective protection time of an airbag is only tens of milliseconds; if the occupant comes into contact with the airbag too early or too late, they will not receive sufficient protection. Figure 2 This is a schematic diagram illustrating the airbag's effective time in an exemplary embodiment. Figure 2 As shown, the gas generator is activated within 0-15 milliseconds or 0-20 milliseconds. The airbag deploys at approximately 30 milliseconds. Between 40 and 60 milliseconds, the airbag is fully inflated and makes contact with the passenger. Between 60 and 120 milliseconds, the gas inside the airbag is gradually deflated.

[0077] Furthermore, high-speed airbag deployment can easily cause secondary injuries to passengers. Specifically, at airbag deployment speeds of up to 300 km / h, the airbag acts like a rapidly inflating "mini-bomb" inside the vehicle, generating tremendous impact force. Occupants should not come into contact with the airbag before it has fully deployed. If passengers lean excessively forward or approach the airbag's deployment path before it has fully deployed, they are prone to secondary injuries. This is because a collision occurs within approximately 200 ms, and the airbag must deploy within a very short time (around 20 ms) after the collision to be effective. Therefore, it usually requires high-explosive propellant for ignition. If the airbag's ignition time can be advanced before the collision, the airbag deployment speed and the resulting additional hazards can be effectively reduced. Moreover, the point of contact between the airbag and the occupant will differ depending on the size of the passenger. Smaller passengers will be closer to the airbag. Although airbags are originally designed to contact the chest of adults, for children or even smaller adults, the airbag may impact their faces. Before the airbags deploy, a collision can alter the passenger's position along the airbag's deployment path, potentially causing the airbag to miss its intended contact with the passenger. For example, braking before the collision, a smaller distance between the passenger and the airbag before deployment, vehicle detachment from the road, or a prolonged collision can delay airbag deployment, or the passenger may not be properly wearing their seatbelt.

[0078] Furthermore, current object control methods suffer from insufficient accuracy in collision detection. Specifically, traditional object control methods typically rely on accelerometers and pressure sensors for collision detection and airbag deployment decisions, making the system less tolerant of sensor failures or data errors. In complex traffic environments or when irregular collisions occur, a single sensor may fail to provide accurate collision information, leading to inaccurate airbag deployment.

[0079] Furthermore, current airbag deployment methods suffer from a lack of flexibility in deployment modes. Specifically, most existing airbags can only select a fixed initiation time and a single deployment mode (single-stage deployment), failing to dynamically adjust according to different collision types and severity. For example, the airbag deployment requirements for minor and severe collisions are completely different. In a minor collision, the airbag initiation time is later and the inflation volume is smaller, while in a severe collision, the airbag initiation time is earlier and the inflation volume is larger to better prevent passengers from being pushed forward into the interior (vehicle interior components) and collided.

[0080] Current object control methods also neglect passenger status. Specifically, existing methods do not consider key factors such as occupant body characteristics, posture, and whether seatbelts are fastened. Passenger status has a significant impact on airbag protection effectiveness. For example, unbelted occupants or those in an dislocated position (such as a reclining position) may come into contact with the airbag too early or too late during a collision, thus failing to achieve the airbag's protective effect and even suffering airbag injury. Furthermore, individuals with significantly different body shapes cannot receive personalized protection under the same deployment criteria.

[0081] In response to current object control methods, this application provides an object control method that integrates active and passive safety. It determines the collision condition feature set by the object feature set of the object to be collided with, and determines the airbag deployment time for each passenger based on the collision condition feature set and the passenger feature set. Compared with directly deploying the airbag through acceleration, this method considers three influencing factors: the object to be collided with, the collision state of the vehicle, and the passenger state, thus obtaining an accurate deployment time. Based on the accurate deployment time, the airbag is deployed, improving the accuracy of the object control method.

[0082] In one embodiment, such as Figure 3 As shown, a method for object control that integrates active and passive safety is provided. This application uses an example of this method applied to a vehicle control system. This application does not limit the execution device of the method for integrating active and passive safety, and includes the following steps 302 to 306:

[0083] Step 302: Obtain the image dataset and point cloud dataset of the vehicle exterior, and determine the object to be collided with and the object feature set of the object to be collided with based on the image dataset and point cloud dataset.

[0084] The object feature set contains the attribute and motion features of the object to be collided with. The point cloud dataset is data collected by radar equipment.

[0085] In practice, the vehicle is equipped with image acquisition equipment and radar equipment. The image acquisition equipment collects image data from the vehicle's exterior in real time, obtaining an image dataset, and transmits this dataset to the control system. Simultaneously, the radar equipment collects point cloud data from the vehicle's exterior in real time, obtaining a point cloud dataset, and transmits this dataset to the control system. The control system receives both the image dataset and the point cloud dataset. Then, based on an image recognition model, the control system processes the image dataset to obtain the object to be collided with and its initial feature set. The control system updates the initial feature set based on the point cloud dataset and the image dataset, obtaining the final object feature set.

[0086] Specifically, the radar device is a lidar system. The image acquisition device collects image data from the vehicle's exterior in real time, obtaining an image dataset. Simultaneously, the lidar also collects point cloud data from the vehicle's exterior in real time, obtaining a point cloud dataset. Then, the image acquisition device transmits the image dataset to the control system via a transmission channel inside the vehicle. At the same time, the radar device transmits the image dataset to the control system via the same transmission channel. The control system performs image recognition on the image dataset based on an image recognition model, obtaining the target object and its initial feature set. The control system processes the point cloud dataset and image dataset to obtain the velocity and volume of the target object. Then, the control system updates the initial feature set based on the velocity and volume of the target object, obtaining the final object feature set.

[0087] In one optional embodiment, the radar device can be a lidar, a millimeter-wave radar, or a combination of lidar and millimeter-wave radar. Image acquisition devices include, but are not limited to, visual cameras. Image datasets are acquired via the visual camera, and deep learning methods are used for collision detection and prediction. Furthermore, high-precision 3D modeling is provided by lidar to obtain a point cloud dataset. Alternatively, different devices can be used to acquire the image dataset and the point cloud dataset.

[0088] In one alternative embodiment, image datasets and point cloud datasets are acquired based on different sensor combinations. For example, radar and ultrasonic sensors alone can be used to determine the object to be collided with, and an inertial measurement unit (IMU) can be used to assist in collision prediction. Single-sensor solutions are low-cost and relatively simple in technology, suitable for use in low-cost vehicles or simple environments, but their accuracy and robustness are poor in complex environments (such as low visibility, severe weather, etc.). Sensor combination solutions offer improved reliability but increase system complexity and cost.

[0089] Step 304: Based on the collision condition prediction model, perform prediction processing on the object feature set and the vehicle control signal set to obtain the collision condition feature set.

[0090] The vehicle's control signal set represents its operating state. The collision condition feature set represents the state when the vehicle collides with the object to be collided with.

[0091] In implementation, the control system incorporates a collision condition prediction model. The control system acquires the vehicle's speed, acceleration, and intervention signal set, and constructs a control signal set based on these parameters. The control system inputs the object feature set and control signal set into the collision condition prediction model, which then performs predictive processing to obtain the collision condition feature set. This feature set characterizes the severity of the collision between the vehicle and the target vehicle. The collision condition feature set includes at least the speed difference and collision pose. The speed difference represents the change in speed before and after the collision, while the collision pose includes the collision direction and the vehicle's position after the collision. For example, the collision direction includes forward and side collisions. The vehicle's position after the collision indicates whether the vehicle has rolled over, overturned, or otherwise overturned.

[0092] Specifically, the control system has a pre-set initial collision condition prediction model. The control system obtains training datasets from various databases and trains the initial collision condition prediction model based on these datasets to obtain the collision condition prediction model. The control system also obtains the vehicle's static prior information from the database, including the vehicle's speed, acceleration, target position, and intervention signal set. The intervention signal set consists of action signals from the vehicle's active safety control module to actively mitigate or avoid collisions before or in the early stages of a collision. Examples include braking signals, steering signals, and other intervention action signals. The control system constructs a signal set based on the speed, acceleration, target position, and intervention signal set. Then, the control system inputs the vehicle's static prior information, the object feature set, and the control signal set into the collision condition prediction model. The collision condition prediction model then performs predictive processing on the object feature set and the control signal set to obtain the collision condition feature set.

[0093] In an exemplary embodiment, the control system obtains the vehicle structure model and collision response characteristic information from a database, and combines the vehicle structure model and collision response characteristic information to obtain the vehicle's static prior information. The control system also obtains the vehicle's speed, acceleration, target position, and intervention signal set from the database. The control system constructs a signal set based on the speed, acceleration, target position, and intervention signal set. The collision condition prediction model includes a Transformer Encoder (the core part of the Transformer model, which is a deep learning model), a Vision Transformer (containing a Visual Transformer and a Sliding Window Transformer, abbreviated as ViT and Swing Window Transformer as Swing), an enhanced MLP (Multi-Layer Perceptron, residual & self-attention), a multi-head cross-attention gating fusion module, and a shared fully connected layer. The control system inputs the object feature set and static prior information into the enhanced MLP in the collision condition prediction model, and inputs the image dataset into the Vision Transformer in the collision condition prediction model. Simultaneously, the control system inputs the control signal set into the Transformer Encoder in the collision condition prediction model. The TransformerEncoder, Vision Transformer, and enhanced MLP encode the input static prior information, image dataset, object feature set, and control signal set. Then, a multi-head cross-attention gating fusion module fuses the initial features obtained from the feature encoding to obtain an initial feature set. Finally, the initial feature set is processed by a shared fully connected layer in the collision condition prediction model to obtain the collision condition feature set.

[0094] Specifically, Figure 4 This is a structural diagram of a collision condition prediction model in an exemplary embodiment. For example... Figure 4As shown, the collision prediction model includes a temporal data branch, an image or point cloud branch, a structural prior branch, an enhanced MLP, a multi-head cross-attention gating fusion module, and a shared fully connected layer. The shared fully connected layer includes a collision type classification module (Sofmax + Focal Loss, where Sofmax is normalization and Focal Loss is focus loss), a collision severity regression module, and a collision attitude regression module. The temporal data branch uses a Transformer Encoder-based temporal encoding module to process temporal signals such as vehicle speed, acceleration, and steering wheel angle. Compared to traditional RNN (Recurrent Neural Network) / LSTM (Long Short-Term Memory) / GRU (Gated Recurrent Unit) models, the Transformer Encoder utilizes a self-attention mechanism to more efficiently capture long-range dependencies and complex nonlinear features, thus significantly improving the expressive power of temporal information. The image / point cloud branch employs a visual information encoding module based on Vision Transformer (including implementations such as ViT or Swin Transformer) to extract global semantic information and local detail features from camera images or point cloud data. This module overcomes the limitations of traditional CNNs (Convolutional Neural Networks) or 3D-CNNs (3D Convolutional Neural Networks) in capturing global relationships, enhancing the model's ability to perceive complex environmental information. The structural prior branch is designed for vehicle structural parameters (such as stiffness, mass, and dimensions) and employs an enhanced multilayer perceptron (MLP) module. This MLP module introduces residual connections and self-attention mechanisms on top of traditional fully connected layers to fully explore the intrinsic relationships between various structural parameters and form prior features with higher expressive power.

[0095] Each branch encodes the input data to obtain initial features. Then, a multi-head cross-attention and gating fusion module performs deep fusion (feature fusion) on the outputs of the intermediate layers of each branch to obtain an initial feature set, achieving adaptive weighting and dynamic interaction of information from different modalities. The fused comfort feature set is then passed to a shared fully connected layer, which employs residual connections and layer normalization strategies to ensure the stability and generalization performance of feature propagation. Finally, the comprehensive features processed by the shared fully connected layer are fed into three task-specific output heads. This shared fully connected layer includes: a collision type classification head, a collision severity regression head, and a collision attitude regression module. The collision severity regression module is used to predict collision severity indices such as Δv and acceleration waveforms; the collision attitude regression module is used to predict collision attitude parameters such as roll angle and yaw angle.

[0096] Before inputting the object feature set, control signal set, and static prior information into the collision prediction model, preprocessing is required. Specifically, the control system fuses the control signal set, object feature set, and static prior information to obtain a target feature parameter set (target mass m, size (l, w, h), velocity v, acceleration a, and overall stiffness coefficient k) usable by the backend collision prediction model. The object feature set is the collision target features obtained through multi-sensor fusion (vision, lidar, millimeter-wave radar). The object feature set is used as a first-level input (first-level input 2) into the collision prediction model. Optionally, the control system can also construct a high-dimensional feature vector based on the control signal set, object feature set, and static prior information. Specifically, this high-dimensional feature vector includes dynamic parameters (relative velocity Δv, relative position, collision angle prediction), static parameters (target vehicle body structural stiffness, front crumple zone characteristics), and environmental parameters (road conditions, weather, presence of other obstacles). The control system quickly retrieves the most similar historical scenarios or simulation cases based on the "structural model" and "collision response characteristics" of the target vehicle or similar vehicles to obtain prior deformation coefficients, acceleration waveform references, etc. If a highly similar case cannot be found, it uses general stiffness and crumple zone reference data of the same type of vehicle as a second choice. The control system concatenates several frames (e.g., 15 frames before the collision, 30fps) of vehicle kinematic data and sensor data into a time-series tensor, and adds the structural stiffness coefficients, mass, and dimensions of the target vehicle body to merge them into the same feature vector. Then, the control system standardizes or performs Min-Max (minimum-maximum) normalization on variables such as velocity and acceleration to maintain the numerical stability of the input features.

[0097] A collision condition feature set is obtained by predicting collision accident characteristics strongly correlated with the severity of passenger injury using a collision condition prediction model. This feature set includes: 1. Δv (velocity difference): the change in velocity before and after the collision. 2. Collision pose: the spatial position and attitude (rollover, yaw) characteristics of the vehicle at the moment of collision. 3. Acceleration waveform: the acceleration time history during the collision, used for airbag control strategy optimization. 4. Structural deformation: the deformation state of the vehicle body after the collision, used to assess the vehicle's remaining safety margin.

[0098] Figure 5 This is a schematic diagram illustrating the operation of a collision condition prediction model in an exemplary embodiment. For example... Figure 5 As shown, after the vehicle is powered on, the control system loads the inference engine to execute the active and passive safety fusion object control method until the vehicle's journey ends. Various sensors and image acquisition devices on the vehicle collect image data and point cloud datasets. These datasets represent environmental perception data. Simultaneously, the sensors also collect vehicle state data, including speed and acceleration. The control system determines the target collision object and its object dataset based on the image data and point cloud dataset. The control system performs data cleaning and feature assembly on the object dataset, vehicle state data, and static prior information to obtain an initial target feature parameter set. The control system acquires an intervention signal set. This set includes active safety signals such as braking signals, steering signals, and ESC (Electronic Stability Control) status. The control system adds the intervention signal set to the initial target feature parameter set to obtain the target feature parameter set. The control system inputs the target feature parameter set into a collision condition prediction model, which then performs prediction processing on the target feature parameter set to obtain a collision condition feature set. This feature set includes speed differences. The system analyzes the collision pose (including collision type) and acceleration waveforms. The control system corrects the collision condition feature set based on confidence levels. If the confidence level of the collision condition is lower than the confidence threshold, the control system performs a secondary search or multi-model refinement. The control system determines the control strategy for each airbag based on the collision condition feature set and passenger feature set, including but not limited to adjusting the deployment timing and airbag deployment method. If the control system malfunctions or fails, such as sensor failure, it controls the airbags using traditional object control methods.

[0099] Building upon this foundation, a deep learning algorithm (collision condition prediction model) is used to perform multi-task learning of labels such as collision type (frontal, side, rear-end, rollover, etc.) and collision severity (Δv (speed difference), peak acceleration, deformation level). The vehicle's active safety modules (braking and steering intervention) activate before or in the early stages of a collision, potentially altering the collision trajectory and relative speed. During each iteration of inference, the latest braking or steering information (i.e., the intervention signal set) is incorporated into the model input or used as a correction term to adjust the output Δv and collision attitude. If the active safety action significantly decelerates or changes direction, the collision severity prediction is updated in real time.

[0100] Because the pre-collision prediction window is extremely short, and real-time vehicle safety control requires high precision, any errors in the prediction results can significantly impact passenger injury. If deep learning model prediction is time-consuming or resource-constrained, a nearest neighbor search can be performed directly on the object feature set and the control signal set (relative speed, angle, structure type, etc.) that does not include intervention signals, against a collision case library. This search can be conducted using methods such as KD-Tree (K-Dimensional Tree, a binary tree data structure) or LSH hashing (Locality-Sensitive Hashing, a technique for approximate nearest neighbor search in high-dimensional data) to find several historical scenes or simulation results with the highest similarity, thus approximating the characteristics of the collision accident. This search result can complement or validate the neural network output, improving prediction robustness.

[0101] In an optional embodiment, the control system trains a pre-set initial collision scenario prediction model. The control system collects training datasets from various data sources. Specifically, these data sources include a real accident database, a virtual risk generation database, a target vehicle collision simulation database, and a typical experimental calibration airbag scenario matrix. The control system collects, organizes, and labels collision scenario data from these data sources, and combines this data with sensor simulation data (the "simulated output" of LiDAR, vision, and millimeter-wave radar in the virtual scenario) and real road test data to form a training sample set. Data resources from the "typical experimental calibration airbag scenario matrix," "target vehicle collision simulation database," "virtual risk generation database," and "real accident database" provide prior information on different target types, collision scenarios, vehicle structure models, and historical real accidents, supporting collision prediction modeling and simulation. The typical experimental calibration airbag scenario matrix, through numerous actual collision tests and airbag deployment tests, calibrates the airbag system triggering and deployment characteristics under different collision types, vehicle speeds, and collision angles in the laboratory. This matrix contains standardized collision scenario parameters and airbag response mechanisms, providing reference data for the collision prediction model. The real-world accident database collects a large amount of real traffic accident data, including vehicle type, collision angle, speed difference, and collision consequences (vehicle damage, personal injury). This data provides real-world priors for the calibration and validation of collision prediction models. The virtual risk generation library uses simulation tools to generate a large amount of virtual pre-collision scenario data. By parametrically varying vehicle relative positions, speed differences, road conditions, and weather conditions, a comprehensive virtual risk scenario database is established, providing more scenario references beyond the scope of real-world testing for collision prediction. The target vehicle collision simulation database targets common vehicle target models (such as sedans, SUVs, trucks, etc.), using computer simulation to build a collision model library. Collision process prediction simulations are performed based on the input parameters using multibody dynamics models and finite element analysis or hybrid methods. Large-scale multibody dynamics is used to quickly predict the vehicle's pose changes and acceleration characteristics at the instant of collision. Small-scale finite element analysis can accurately model the deformation, stress distribution, and energy absorption characteristics of vehicle structures under higher computational costs. This database contains impact simulation results under different initial conditions (velocity, angle, position) from a real accident database and a virtual risk generation library, including deformation degree, collision energy absorption characteristics, and failure modes of the vehicle body load-bearing area. The acquisition frequency of the above data types is generally between 100 and 1000 Hz (Hz is Hertz, depending on sensor / test requirements); the key outputs include collision type labels (Front / Side / Rear / Multiple / Rollover, etc.), collision severity labels (Δv range, peak acceleration, vehicle body deformation level), and vehicle attitude and position information (rollover angle, yaw angle, etc.).

[0102] The training dataset contains subsets of training data, each representing a sample collision event. Each subset includes a training input data subset and a validation input data subset. The training input data subset includes at least the control signal set of the sample vehicle in the sample collision event, the static prior information of the sample vehicle, and the object feature set of the sample target object. The validation input data subset includes at least the sample collision condition feature set of the sample collision event. The control system feeds each training input data subset into the initial collision condition prediction model. The initial collision condition prediction model then performs prediction processing on each training input data subset to obtain the prediction results for each sample. Data processing is performed on each prediction result and each validation input data subset to obtain the loss value of the initial collision condition prediction model. The control system determines whether the loss value exceeds a preset loss threshold. If the loss value does not exceed the loss threshold, the control system confirms the initial collision condition prediction model as the collision condition prediction model. If the loss value exceeds the loss threshold, the control system continues to execute the step of feeding each training input data subset into the initial collision condition prediction model until the loss value does not exceed the loss threshold.

[0103] The trained collision scenario prediction model can estimate the severity of the collision between the vehicle and the target object 500 milliseconds before the collision, combining real-time and historical data, and predict the collision scenario feature set. This feature set characterizes the type, location, and direction of the collision. It includes the collision pose (forward collision, side collision) and collision severity (change in collision velocity, acceleration waveform, vehicle structural deformation). When an impending collision or a high-risk collision is detected, the control system calls the offline-trained deep learning model for rapid inference (typical cycle <10~20ms) to obtain the collision accident feature set (Δv, acceleration waveform, collision angle, vehicle deformation level, etc.).

[0104] In one exemplary embodiment, Figure 6 This is a schematic diagram of the input and output of a collision scenario model in an exemplary embodiment. For example... Figure 6As shown, the inputs to the collision scenario model include Level 1 Input 1, Level 2 Input 2, and Level 3 Input 3. Level 1 Input 1 is the input used to train the initial collision scenario model, obtaining the collision scenario model. Level 1 Input 1 contains a training dataset constructed from a typical experimental calibration airbag deployment scenario matrix, a target vehicle collision simulation database, a virtual hazard generation library, and a real accident database. Level 2 Input 2 is the object feature set of the object to be collided with, which is also the collision target feature. Level 3 Input 3 is the vehicle's control signal set. The collision scenario prediction model performs collision prediction processing on the Level 1 input data to obtain the collision scenario feature set. The collision scenario feature set (collision accident features) at least includes the collision pose and velocity difference. The collision scenario feature set may also include acceleration waveforms and structural deformation.

[0105] Optionally, the training input data subset may also include environmental parameters (weather data), etc. The training validation input data subset includes information such as casualty statistics. This application embodiment does not limit the content of the training input data subset and the training validation data subset. The more data types included in the training input data subset, the more accurate the collision condition prediction model will be.

[0106] Optionally, the collision condition feature set may also include acceleration waveforms and structural deformation. The acceleration waveform is the acceleration time history during a vehicle collision. This acceleration waveform is used to optimize the airbag control strategy. The structural deformation is the deformation state of the vehicle body after the collision, used to assess the vehicle's remaining safety margin. This application embodiment does not limit the collision condition features included in the collision feature set.

[0107] Step 306: Obtain the passenger feature set of each passenger in the vehicle, and determine the deployment time of each airbag based on the collision condition feature set and the passenger feature set.

[0108] The passenger feature set includes the passenger's attribute feature set and posture feature set.

[0109] In implementation, the control system acquires passenger feature sets for each passenger inside the vehicle using various sensors and image acquisition devices. Then, the control system determines the initial collision acceleration based on the collision condition feature set. This initial collision acceleration is the initial collision acceleration for each passenger. The control system determines the first collision time between the vehicle and the object to be collided with. Then, the control system determines the second time from each passenger to the components inside the vehicle, based on the initial collision acceleration, the passenger feature set, and the collision condition feature set. Finally, the control system determines the airbag deployment time for each passenger based on the first collision time and each of the second collision times.

[0110] In an optional embodiment, the control system calculates the airbag deployment speed and pressure control curve based on the collision condition feature set (acceleration waveform steepness, Δv magnitude) and the occupant position (whether close to the airbag module) in the occupant feature set, and determines the initiation time of each airbag to ensure that passengers are protected in a timely manner without causing excessive secondary injury to them.

[0111] In the above-mentioned active and passive safety integrated object control method, the collision condition feature set is determined by the object feature set of the object to be collided with, and the airbag deployment time corresponding to each passenger is determined based on the collision condition feature set and the passenger feature set. Compared with directly deploying the airbag by acceleration, this method considers three influencing factors: the object to be collided with, the collision state of the vehicle, and the state of the passengers, and obtains an accurate deployment time. Then, the airbag is deployed based on the accurate deployment time, which improves the accuracy of the object control method.

[0112] In one exemplary embodiment, such as Figure 7 As shown, the specific processing steps for determining the object to be collided and its object feature set based on the image dataset and point cloud dataset in step 302 include steps 702 to 706. Wherein:

[0113] Step 702: Based on the image recognition model, perform image recognition on each image data in the image dataset to obtain the object to be collided with and the initial object feature set of the object to be collided with.

[0114] In implementation, the control system inputs image datasets into an image recognition model. The model then performs image recognition on each dataset to obtain initial collision targets. The control system then filters these initial collision targets based on preset collision criteria and determines an initial object feature set. This initial object feature set includes the category and quality of the collision targets.

[0115] Specifically, the control system has pre-set collision screening conditions. Based on an image recognition model, the control system performs image recognition on each image in the image dataset to obtain each initial object to be collided with and its category. Then, the control system filters the initial objects to be collided with according to the collision screening conditions. The control system determines the quality of the objects to be collided with in the database and constructs an initial object feature set based on the quality and category of the objects to be collided with.

[0116] Step 704: Determine the velocity and volume of the object to be collided with based on the point cloud dataset and the image dataset.

[0117] In implementation, the radar data includes LiDAR (Light Detection and Ranging) 3D point cloud bounding boxes. The control system obtains the length, width, and height of the object to be collided with from the LiDAR 3D point cloud bounding boxes and calculates the volume of the object based on its length, height, and width. Then, based on the radial velocity data from the radar data, the LiDAR 3D data, and the visually varying coordinates, the control system calculates the target's longitudinal velocity, lateral velocity, and acceleration.

[0118] In one exemplary embodiment, the radar data includes a LiDAR (Light Detection and Ranging) 3D point cloud bounding box, LiDAR 3D data, and image velocity data. The control system obtains the length, width, and height of the object to be collided with from the LiDAR 3D point cloud bounding box, and calculates the volume of the object based on the length, height, and width. The control system determines the visual change coordinates of the object to be collided with based on each image data, and determines the target longitudinal velocity, target lateral velocity, and acceleration of the object to be collided with based on the visual change coordinates, radial velocity data, and LiDAR 3D data.

[0119] Step 706: Update the initial object feature set based on the velocity and volume of the object to be collided with, and obtain the object feature set.

[0120] In implementation, the control system adds the velocity and volume of the object to be collided with to the initial object feature set, resulting in an object feature set. Since the initial object feature set contains the category and mass of the object to be collided with, the updated object feature set contains the category, mass, volume, longitudinal velocity, lateral velocity, and acceleration of the object to be collided with.

[0121] In an optional embodiment, the control system queries the database for the stiffness of the object to be collided with based on the category of the target object, and adds the stiffness of the object to be collided with to the object feature set to obtain an updated object feature set.

[0122] In one exemplary embodiment, Figure 8 This is a schematic diagram illustrating the input and output of a multimodal recognition model in an exemplary embodiment. For example... Figure 8 As shown, the vehicle collects real-time data on the road and surrounding environment using multiple sensors (visual camera, LiDAR, millimeter-wave radar), resulting in image datasets and point cloud datasets. The image datasets and point cloud datasets are used to determine the object to be collided with, and to define the object's feature set. This feature set includes four physical properties strongly correlated with collision: volume, mass, velocity (vector), and stiffness.

[0123] Specifically, the control system performs image recognition on each image data based on an image recognition model, thereby obtaining the 2D (two-dimensional) bounding box of the initial collision target in the RGB (red, green, blue, a color standard) image data. It then uses instance segmentation to obtain the corresponding 3D (three-dimensional) information for preliminary estimation, thus identifying the collision target. Next, the control system distinguishes between soft targets (such as pedestrians and animals) and hard targets (such as vehicles and roadblocks) based on the identified target type. Finally, the control system determines the estimated mass of the collision target in the database based on the target category (such as car, truck, motorcycle, pedestrian).

[0124] The control system obtains the target's length, width, and height from the LiDAR 3D point cloud bounding box to estimate the volume of the object to be collided with. Then, based on the radial velocity data from the radar data, the LiDAR 3D data, and the visual coordinate changes, the control system calculates the target's longitudinal velocity, lateral velocity, and acceleration.

[0125] For raw data (image datasets and point cloud datasets) acquired from sources such as vision, LiDAR, and millimeter-wave radar, the control system performs temporal and spatial synchronization and registration of the raw data in a unified coordinate system, and constructs a rule-based classifier by integrating the registered raw information. Then, the control system estimates the mass and overall stiffness characteristics of the object to be collided with. The deep learning inference module of the control system is deployed on the automotive electronic control unit (ECU), GPU (Graphics Processing Unit), and dedicated AI (Artificial Intelligence) chip to achieve target detection, classification, velocity, and volume estimation.

[0126] The database includes typical mass ranges for various vehicles (e.g., 1 ton for small cars, 2.5 tons for SUVs, and over 3 tons for trucks), as well as stiffness information (classified by vehicle type and front structure). Soft targets (pedestrians, animals) and hard targets (bumpers, guardrails, etc.) are classified by stiffness; the stiffness coefficient for pedestrians / animals is much smaller than that for vehicles or road barriers.

[0127] As the vehicle and the object to be collided approach each other, the control system continuously updates the speed, position, and orientation. Once a potential collision risk is detected (e.g., TTC < 0.5s, TTC = Time to Collision), the latest output of the multimodal recognition module—including volume, mass, velocity (containing longitudinal and lateral components), and stiffness—is sent to the vehicle's collision prediction model.

[0128] In this embodiment, by integrating image datasets and point cloud datasets collected by multiple sensors such as image acquisition devices and radar devices, the vehicle's surrounding environment can be perceived more accurately. Based on the image datasets and point cloud datasets, the object to be collided with and its feature set are determined, enabling real-time assessment of potential collision risks. Compared to traditional single-sensor systems, this application significantly reduces sensor errors and incomplete information through the fusion of multi-source information (such as images, distance, and speed), providing a more comprehensive and accurate collision prediction capability.

[0129] In one exemplary embodiment, such as Figure 9 As shown, the specific processing procedure of step 702 includes steps 902 to 906. Wherein:

[0130] Step 902: Based on the image recognition model, perform image recognition on each image data in the image dataset to obtain each initial object to be collided with and the category of each initial object to be collided with.

[0131] In implementation, the control system inputs each image data from the image dataset into the image recognition model, and performs image recognition on each image data to obtain each initial object to be collided with and the category of each initial object to be collided with.

[0132] Step 904: Select collision objects from the initial collision objects according to the preset collision screening conditions.

[0133] In implementation, the control system is pre-set with collision screening conditions. For each initial object to be collided with, the control system determines whether the initial object meets the collision screening conditions based on the distance and angle between the initial object and the vehicle. If the initial object meets the collision screening conditions, it is designated as the object to be collided with.

[0134] In one exemplary embodiment, the collision screening criteria are initial collision targets that are at an angle of ±30 degrees (positive and negative 30 degrees) forward from the vehicle and are no more than 50 meters away from the vehicle. The control system selects the initial collision targets from among the initial collision targets as collision targets that are at an angle of ±30 degrees forward from the vehicle and are no more than 50 meters away from the vehicle.

[0135] Optionally, the collision screening criteria can be adjusted according to the vehicle type and speed. This application embodiment does not limit the collision screening criteria.

[0136] Step 906: Query the quality of the object to be collided in the database based on the category of the object to be collided, and construct an initial object feature set of the object to be collided based on the category and quality of the object to be collided.

[0137] In implementation, the control system determines the mass of the object to be collided with based on its category in the data. Then, the control system combines the category and mass of the object to be collided with to obtain an initial feature set of the object.

[0138] In an exemplary embodiment, the control system looks up an average mass or mass range from a table based on the category of the object to be collided with (such as a car, truck, motorcycle, or pedestrian), and then corrects the mass based on the volume estimated by the 3D bounding box.

[0139] In an alternative embodiment, Figure 10 This is a schematic diagram illustrating the operation flow of an image recognition model in an exemplary embodiment. When the vehicle starts, the control system continuously acquires image datasets and point cloud datasets based on the image recognition model and determines the object to be collided with until the vehicle finishes driving, the power is cut off, or the risk to the vehicle is eliminated. The control system acquires image datasets and point cloud datasets and performs time synchronization processing on them. Specifically, the control system aligns the visual frames, LiDAR point cloud frames, and millimeter-wave radar detection frames in the image dataset and point cloud dataset using a unified timestamp (GPS / IMU / ROS, where GPS is the Global Positioning System, IMU is the Inertial Measurement Unit, and ROS is the Robot Operating System). Spatial registration is achieved by calibrating the extrinsic parameters of the coordinate systems of the camera, LiDAR, and millimeter-wave radar during the offline calibration phase; during the online phase, coordinate transformation is performed according to the extrinsic parameter matrix to ensure that the same initial object to be collided with corresponds to the same position in different sensor data.

[0140] Then, the control system performs image recognition on the time-synchronized image dataset and point cloud dataset based on the image recognition model to obtain the initial collision objects and their types. The image recognition model uses object detection and instance segmentation networks (such as Faster R-CNN, YOLO, and Mask R-CNN; Faster R-CNN is a faster region convolutional neural network, YOLO is an image recognition model, and Mask R-CNN is a masked region convolutional neural network). The input format of the image recognition model is an RGB image, and the output is a 2D bounding box / instance segmentation mask plus the class probability of the initial collision objects.

[0141] When determining the collision target based on an image recognition model, the volume and velocity of the collision target can be determined simultaneously using a neural network. Specifically, the 3D object detection network of the LiDAR branch (such as PointNet++, VoxelNet, SECOND; PointNet++ is a deep learning model for processing point cloud data, VoxelNet is a voxel network, and SECOND is a single-shot multi-box detector) takes point cloud data as input and outputs 3D bounding boxes (x, y, z, length, width, height, orientation). The volume of the collision target can be determined using the 3D object detection network of the LiDAR branch. The radar branch takes point traces or sparse signals as input and processes the relative distance and relative velocity features using a Transformer-based sequence network to obtain the velocity of the collision target.

[0142] After determining the velocity, volume, and category of the target object, a multimodal fusion layer is used to match the target object, combining radar velocity information to output a unified target detection and velocity estimation result. Then, the control system performs attribute estimation (attribute regression) on the target object. Based on the fused features, the attribute regression head is trained to estimate the target volume, mass, and stiffness (classification or regression). Mass and stiffness are then "corrected" using an external database. The loss function for the detection part uses a combination of classification (focal loss) and bounding box regression (Smooth L1) loss, commonly used in target detection. The stiffness / mass classification loss function uses cross-entropy. When the initial mass estimate output by the network differs significantly from the prior database, online or offline calibration correction is performed. Similarly, stiffness estimation can be performed by interval mapping according to vehicle type or target material to correct the values ​​output by the neural network.

[0143] As the vehicle and the object to be collided approach each other, the control system continuously updates the target velocity, position, and orientation of the object to be collided with. Once a potential collision risk is detected (e.g., TTC < 0.5s, TTC = Time to Collision), the object's characteristics—volume, mass, velocity (including longitudinal and lateral components), and stiffness—are sent to the vehicle's collision prediction model.

[0144] In this embodiment, the objects to be collided with are determined in each image data through an image recognition model and collision screening conditions, thus obtaining the objects that will collide. By determining the initial object feature set of the objects to be collided with, the attribute features of the objects to be collided with are obtained, which facilitates subsequent prediction of the situation when the vehicle collides with the objects to be collided with.

[0145] In one exemplary embodiment, such as Figure 11As shown, the specific processing steps for obtaining the passenger feature sets of each passenger in the vehicle in step 306 include steps 1102 to 1106. Wherein:

[0146] Step 1102: Obtain the attribute feature set of each passenger through the seat sensors in the vehicle.

[0147] In practice, each seat in the vehicle is equipped with a seat sensor. The control system uses these seat sensors to acquire the attribute feature set of each passenger.

[0148] Specifically, since different passengers sit in different seats, and each passenger has different attributes and postures, different airbag control strategies are needed for each passenger. To formulate these airbag control strategies, passenger attribute feature sets need to be collected first using in-cabin sensors. These in-cabin sensors include seat pressure sensors, vision sensors for occupant height estimation, and seatbelt sensors. Seat pressure sensors and vision sensors are used to determine whether the passenger is an adult, child, or has a specific body shape. The control system acquires data transmitted from the seat pressure sensors, occupant weight, and vision sensors, and uses this data to determine the passenger's basic attributes. Seatbelt sensors transmit information about seatbelt usage and pretensioner tension and locking status. The control system determines each passenger's safety attributes based on the seatbelt sensors and constructs a passenger attribute dataset based on these safety attributes and basic attributes. Step 1104: Acquire the posture feature sets of each passenger using image acquisition equipment and / or infrared sensors.

[0149] In implementation, the control system acquires an initial posture feature set for each passenger through image acquisition equipment and / or infrared sensors. This posture feature set includes the passenger's head and neck position and body posture. Based on the initial posture feature set and historical experimental calibration, the control system performs prior classification of the passenger's basic physical characteristics (body size) and habitual sitting posture, and updates the classified sitting posture to the initial posture feature set, thus obtaining the posture feature set.

[0150] Optionally, but not limited to, identifying the initial posture feature set of each passenger using an infrared sensor, or acquiring a passenger image dataset using an image acquisition device and performing image recognition on the passenger image dataset to obtain the initial posture feature set of each passenger.

[0151] Step 1106: Construct passenger feature sets for each passenger based on each attribute feature set and each posture feature set.

[0152] In implementation, the control system performs data filtering, data cleaning, and coordinate alignment on the attribute feature set and posture feature set of each passenger to obtain the passenger feature set. This passenger feature set includes the passenger position vector P (relative to the in-vehicle coordinate system), posture parameters A (head and neck tilt angle, upper body forward tilt angle), seat belt restraint status S (fastened, unfastened, pretensioner status), and basic passenger attributes C (classified as child, adult, small female, medium-sized male, etc.).

[0153] In this embodiment, by collecting passenger feature sets for each passenger, the passenger's sitting posture and basic attributes are obtained, which facilitates subsequent optimization of the airbag detonation time based on the passenger's characteristics.

[0154] In one exemplary embodiment, such as Figure 12 As shown, the specific processing steps for determining the deployment time of each airbag based on the collision condition feature set and each passenger feature set in step 306 include steps 1202 to 1208. Wherein:

[0155] Step 1202: Determine the first collision time between the vehicle and the object to be collided based on the vehicle's speed, acceleration, and the first distance between the vehicle and the object to be collided.

[0156] In implementation, a first collision time algorithm is pre-set in the control system. The control system acquires the vehicle's acceleration and velocity, and determines the first distance between the vehicle and the object to be collided with from the latest image data. Then, the control system processes the vehicle's velocity, acceleration, and first distance according to the first collision time formula to obtain the first collision time between the vehicle and the object to be collided with.

[0157] Step 1204: Determine the initial collision acceleration based on the preset collision process time, the speed difference in the collision condition feature set, the object feature set of the object to be collided with, and the vehicle attribute information.

[0158] The collision condition feature set includes the speed difference. The object feature set of the object to be collided with includes the stiffness correction factor and mass of the object. The vehicle's attribute information includes the vehicle's mass and stiffness correction factor.

[0159] In implementation, the collision process time is pre-set in the control system. The control system processes the collision process time, speed difference, object feature set of the object to be collided, and vehicle attribute information according to the initial collision acceleration algorithm to obtain the initial collision acceleration.

[0160] The initial collision acceleration algorithm is shown in the following formula (1):

[0161] (1)

[0162] In the above formula (1), Δv is the velocity difference, t is the collision process time, and a is the initial collision acceleration. Vehicle stiffness correction factor This is the stiffness correction factor for the object to be collided with. For the quality of the vehicle, The mass of the object to be collided with.

[0163] Step 1206: For each passenger, determine the second collision time based on the passenger feature set, initial collision acceleration, and collision pose in the collision condition feature set.

[0164] The second collision time is the time it takes for the passenger's head and neck to come into contact with components inside the vehicle.

[0165] In implementation, the control system determines the distance from each passenger to components within the vehicle based on a passenger feature set. Then, the control system corrects the initial collision acceleration based on the passenger feature set and the collision condition feature set to obtain the final collision acceleration. The control system then determines the second collision force based on the collision acceleration and the second distance.

[0166] Step 1208: Determine the deployment time of the airbag corresponding to the passenger based on the first collision time and the second collision time.

[0167] In practice, the control system subtracts the first collision time from the corresponding second collision time for each passenger to obtain the airbag deployment time for that passenger.

[0168] In this embodiment, based on the real-time changes in the collision condition feature set, passenger feature set, and object feature set of the object to be collided with, the timing and force of airbag deployment for each passenger are dynamically optimized, improving the accuracy of airbag deployment time and thus enhancing the accuracy of the object control method. Furthermore, this adjustment not only considers the severity of the collision but also incorporates personalized configurations based on factors such as passenger body size, seating posture, and seatbelt usage, ensuring that the airbag system provides more precise and targeted protection.

[0169] In one exemplary embodiment, such as Figure 13 As shown, the specific processing steps for determining the second collision time of each airbag based on the collision condition feature set and each passenger feature set in step 1206 include steps 1302 to 1306. Wherein:

[0170] Step 1302: For each passenger, determine the second distance between the passenger's head and neck and the device in the vehicle based on the posture parameters and position vector in the passenger feature set.

[0171] Each passenger's passenger feature set contains the passenger's attitude parameters and position vector.

[0172] In implementation, the control system determines the head and neck coordinates of each passenger based on their posture parameters and position vector. Then, the control system determines the coordinates of the corresponding device within the vehicle for each passenger, and based on the head and neck coordinates and the device coordinates, determines a second distance between the passenger's head and neck and the device.

[0173] In one exemplary embodiment, the control system determines the head and neck coordinates of each passenger based on the passenger's posture parameters and positional mass. And determine the coordinates of the device in the vehicle corresponding to the passenger. This device is the one the passenger is most likely to hit. The control system calculates the coordinates of the head and neck and the device based on the second distance algorithm to obtain the second distance between the passenger's head and neck and the device. The second distance algorithm is shown in formula (2) below:

[0174] (2)

[0175] In the above formula (2), The second distance, For the head and neck coordinates, For device coordinates.

[0176] Optionally, the components in the vehicle may include, but are not limited to, the steering wheel, front seats, etc., as described in the embodiments of this application.

[0177] Step 1304: Correct the initial collision acceleration based on the seat belt constraint state and basic attributes in the passenger feature set and the collision pose in the collision condition feature set to obtain the collision acceleration.

[0178] The passenger feature set includes the passenger's seatbelt restraint status and basic attributes. The collision condition feature set includes the vehicle's collision pose.

[0179] In practice, the control system modifies the initial collision acceleration based on the seatbelt restraint status, basic data, and collision pose to obtain the final collision acceleration. For example, if the seatbelt restraint status is "fastened," the control system lowers the initial collision acceleration. If the passenger's basic attributes indicate that the passenger is female, the control system lowers the initial collision acceleration. If the collision pose indicates that the vehicle will roll over, the system increases the vehicle's lateral acceleration.

[0180] Step 1306: Based on the collision time algorithm, perform data processing on the collision acceleration and the second distance to obtain the second collision time.

[0181] In implementation, a collision time algorithm is pre-set in the control system. The control system processes the collision acceleration and the second distance to obtain the second collision time. The algorithm for the second collision time is shown in the following formula (3):

[0182] (3)

[0183] In the above formula (3), For the second collision time, The second distance, This refers to the collision acceleration.

[0184] In this embodiment, the second collision time when a passenger collides with a device inside the vehicle is determined by the collision condition feature set and the passenger feature set of each passenger, so as to facilitate the subsequent determination of the detonation time based on the second collision time.

[0185] In one exemplary embodiment, after determining the detonation time, it is also necessary to determine the gas chamber based on the detonation time. Therefore, as... Figure 14 As shown, after step 1206 is executed, the specific processing procedure of the airbag control method includes steps 1402 to 1406. Wherein:

[0186] Step 1402: Determine whether the airbags for each passenger need to be deployed based on the passenger feature set and collision condition feature set of each passenger.

[0187] In practice, the control system is pre-set with detonation conditions. For each passenger, the control system determines whether the airbag corresponding to that passenger needs to deploy based on that passenger's characteristic set and collision condition characteristic set, and obtains a determination result. If the determination result indicates that the passenger's airbag needs to deploy, the control system identifies that passenger as the target passenger and designates the airbag corresponding to the target passenger as the target airbag.

[0188] Specifically, the passenger feature set includes the passenger's seatbelt restraint status. The collision condition feature set includes speed difference and structural deformation. For each passenger, the control system determines the collision intensity within the vehicle based on the speed difference, structural deformation, and the passenger's seatbelt restraint status. The control system then determines whether the collision intensity reaches the preset detonation conditions. If the collision intensity reaches the detonation conditions, the control system identifies that passenger as the target passenger and designates the airbag corresponding to the target passenger as the target airbag.

[0189] Optionally, the detonation conditions are determined based on multiple collision experiments, but the embodiments of this application do not limit the detonation conditions.

[0190] Step 1404: If the target airbag of the target passenger needs to be detonated, generate a target ignition signal based on the target detonation mode and detonation time corresponding to the target passenger.

[0191] In practice, when a target airbag corresponding to a target passenger needs to be deployed, the control system determines the target deployment mode of the target airbag, which is also the target deployment mode corresponding to the target passenger, based on the passenger characteristic set and collision condition characteristic set of the target passenger. Then, the control system generates a target ignition signal based on the target deployment mode and the deployment time corresponding to the target passenger.

[0192] Specifically, the control system has pre-set detonation modes, including single-stage and multi-stage detonation modes. The control system dynamically selects the target detonation mode based on each passenger's characteristic set and the collision condition characteristic set. For minor collisions, the control system determines a low-intensity single-stage detonation mode as the target detonation mode; for severe collisions, the control system determines a high-intensity multi-stage detonation mode as the target detonation mode. Furthermore, the control system can adjust the detonation time of the target airbag in real time (delayed detonation or on-time detonation) based on the collision condition characteristics and the target passenger's passenger characteristic set.

[0193] When a target airbag corresponding to a target passenger needs to be deployed, the control system determines the target deployment mode of the target airbag, which is also the target deployment mode corresponding to the target passenger, based on the passenger feature set and the collision condition feature set of the target passenger. Each collision condition feature in the collision condition feature set will have a different impact on the target deployment mode and deployment intensity of the target airbag, as shown in Table 1 below:

[0194] Table 1

[0195]

[0196] Table 1 shows the relationship between the collision condition feature set and the detonation mode and detonation intensity of the target airbag. As shown in Table 1 above, the collision condition feature set includes collision pose, velocity difference, etc. Acceleration waveform and structural deformation. The collision pose includes the collision direction and collision type. Optionally, the control system can also determine the detonation mode and detonation intensity of the target airbag based on the intervention signal set.

[0197] Meanwhile, each passenger feature in the target passenger's passenger feature set will have a different impact on the target airbag's detonation mode and detonation intensity, as shown in Table 2 below:

[0198] Table 2

[0199]

[0200] Table 2 shows the influence of passenger characteristics in the passenger feature set on the target airbag's initiation mode and initiation intensity. The passenger feature set includes the passenger's seatbelt restraint conditions, the passenger's basic attributes (body type, height, weight), the passenger's sitting posture, and the distance between the passenger and vehicle interior components (steering wheel, dashboard).

[0201] For example, when the collision intensity reaches the detonation condition and the occupant is in a normal sitting position, seat belts are fastened, and the peak acceleration is moderate, the deployment of a standard single-stage airbag (single-stage detonation mode) can meet the safety requirements. When the collision is more severe (high Δv, severe structural deformation) or the occupant's posture is abnormal (not wearing a seat belt, excessive forward leaning), a two-stage or multi-stage detonation strategy, i.e., a multi-stage detonation mode, is activated.

[0202] It should be noted that the target airbag deployment mode is flexibly adjusted based on the collision situation represented by the collision condition feature set and the passenger status represented by the passenger feature set. The airbag deployment criteria used in this application to determine the target deployment mode require comprehensive consideration of collision-related factors and occupant-related factors. Through multi-sensor data fusion, real-time signal processing, and adaptive algorithms, the optimal deployment timing and intensity are calculated. Essentially, it leverages the coupling effect between collision dynamics and occupant status to ensure the protected object contacts the fully deployed airbag at the appropriate moment, thereby minimizing passenger injury risk during a collision while simultaneously considering the system's response speed and protective effect under different collision scenarios.

[0203] The control system references the detonation time corresponding to the target passenger and generates the target ignition signal corresponding to the target airbag according to the target detonation mode.

[0204] Step 1406: Send the target ignition signal to the airbag device to instruct the target airbag corresponding to the target passenger to deploy.

[0205] During implementation, the control system sends the target ignition signal to the airbag device, instructing the target airbag corresponding to the target passenger to deploy according to the target detonation mode in order to protect the target passenger.

[0206] In an optional embodiment, the control system includes an airbag detonation model to determine the airbag detonation strategy and detonation time. Figure 15 This is a schematic diagram illustrating the input and output of an airbag detonation model in an exemplary embodiment. For example... Figure 15 As shown, the airbag deployment model accepts two levels of input 1 (collision condition feature set) and two levels of input 2 (passenger feature set). The passenger feature set comes from the in-vehicle sensing system and internal state monitoring module, including seat belt signal, occupant position, posture, basic body characteristics and responses (such as whether they lean forward / backward, whether they lower their head).

[0207] The airbag deployment model utilizes a set of collision scenario features (such as acceleration waveforms and Δv) to rapidly estimate the possible kinematic responses of a passenger at the moment of impact within a very short time. Specifically, the airbag deployment model inputs the acceleration waveform into a simplified "human-seat-seatbelt" multibody dynamics model to predict the degree of forward inertial displacement of the occupant and the change in the relative distance between the head and neck and the steering wheel or dashboard at the moment of impact. If the occupant is not wearing a seatbelt or is in a forward-leaning posture, the head and chest displacement at the moment of impact will be greater and the duration shorter. If the occupant is in a backward-leaning posture, there is a possibility of the occupant diving relative to the seat at the moment of impact, resulting in poor seatbelt protection and excessive spinal curvature; these situations require special labeling for the airbag algorithm to make decisions.

[0208] After acquiring the collision condition feature set (secondary input 1) and the passenger feature set (secondary input 2), the airbag deployment model makes decisions based on a predefined deployment strategy to obtain the deployment time and target deployment mode of the target airbag. Specifically, this includes: 1. Determining the deployment time window: Airbag deployment requires a decision within a short millisecond timeframe, determining whether the airbag triggering conditions are met within 200ms before the collision begins. 2. Multi-stage deployment: When the collision intensity reaches a specific threshold and the occupant is in a normal sitting posture, seatbelt is fastened, and the peak acceleration is moderate, a standard single-stage airbag deployment is sufficient to meet safety requirements. Multi-stage deployment conditions: When the collision is more severe (high Δv, severe structural deformation) or the occupant's posture is abnormal (not wearing a seatbelt, excessive forward lean), a secondary or multi-stage deployment strategy is activated. 3. Deployment speed and deployment pattern: Based on the collision accident characteristics (acceleration waveform steepness, Δv magnitude) and the occupant's position (whether close to the airbag module), the airbag deployment speed and pressure control curve are calculated to ensure timely protection without causing excessive secondary injury to the occupant.

[0209] In practical applications, due to prediction errors, a reduced-order detonation strategy may be adopted, that is, only considering the dynamic detonation time. While ensuring that the original airbag detonation principle and deployment method are not changed, the contact between the airbag and the passenger can still be optimized under different collision conditions by changing the detonation time.

[0210] The deployment time is determined by taking into full account the passenger's actual posture, whether the seat belt is fastened, vehicle speed, and other occupant, collision dynamics, and vehicle status, so that the airbag can play the best protective role in the collision scenario, while avoiding secondary injury to the occupants.

[0211] No detonation: If the system analysis determines that the collision intensity is too light to cause serious damage, or that there is no actual collision (such as a minor scrape), the airbags will not be triggered to deploy.

[0212] Pre-deployment: If the system detects that the occupant is not wearing a seatbelt and is tilted forward, it advances the airbag deployment threshold time and shortens the decision-making time delay. Pre-deployment allows the airbag to initially deploy before the occupant is fully tilted forward into the danger zone, thus providing protection in a shorter time.

[0213] Timely Deployment: Under the traditional airbag control logic, once the thresholds such as acceleration, collision intensity, and relative velocity are met, the system will quickly trigger the airbag deployment at the standard design time point, avoiding any unnecessary delays.

[0214] Delayed activation: When the predicted collision intensity is low, or the occupant's posture is normal and the seat belt is well restrained, and the obstacle in front is not judged to be serious, the ignition timing of the airbag can be delayed, that is, after the traditional collision threshold is reached, a controllable, extremely short delay window is still maintained.

[0215] In this embodiment, by selecting the target airbags to be deployed and determining the corresponding deployment mode, different airbag protection is provided for each passenger, avoiding the "one-size-fits-all" fixed protection level of traditional airbag systems and reducing the risk of injury caused by over- or under-deployment of airbags. In the case of a minor collision, the risk of airbag-related injury to the occupant is reduced by delaying deployment or reducing the airbag deployment force; in the case of a severe collision, sufficient protection is provided by rapid and high-intensity airbag deployment.

[0216] In one exemplary embodiment, Figure 16 This is a schematic diagram of the architecture of a control system in an exemplary embodiment. For example... Figure 16 As shown, the control system includes a multimodal target recognition model, a collision condition prediction model, and an airbag deployment model. In the risk perception phase, the multimodal target recognition model identifies the target object and its feature set. In the pre-collision phase, the collision condition prediction model determines the collision condition feature set. The airbag deployment model determines the airbag deployment time and mode. In the active airbag deployment phase, the airbag is deployed according to the deployment time and mode. During the collision phase, the deployed airbag protects the passenger from injury. The airbag deployment is dynamically adjusted through these three models: multimodal target recognition, collision condition prediction, and airbag deployment.

[0217] This application utilizes a highly efficient collision prediction algorithm and a rapid data processing mechanism to predict collisions in real time before they occur. The control system can complete the entire process from collision prediction to airbag control in a very short time, improving the vehicle's reaction speed at the moment of collision and thus reducing the airbag deployment speed. This plays a crucial role in reducing the severity of occupant injuries and improving safety during a collision. It increases the vehicle's reaction time to emergencies, ensuring that the airbag system can deploy rapidly at critical moments, providing timely protection for occupants. In high-risk collision scenarios, through accurate judgment and rapid response, the probability of injury caused by the collision is effectively reduced.

[0218] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0219] Based on the same inventive concept, this application also provides an object control device for implementing the above-mentioned object control method for active-passive security fusion. The solution provided by this device is similar to the implementation described in the above method. Therefore, the specific limitations in one or more embodiments of the object control device for active-passive security fusion provided below can be found in the limitations of the object control method for active-passive security fusion described above, and will not be repeated here.

[0220] In one exemplary embodiment, such as Figure 17 As shown, a dynamic and passive safety fusion object control device 1700 is provided, comprising: an acquisition module 1701, a prediction module 1702, and a determination module 1703, wherein:

[0221] The acquisition module 1701 is used to acquire image datasets and point cloud datasets of the vehicle exterior, and to determine the object to be collided with and the object feature set of the object to be collided with based on the image datasets and point cloud datasets.

[0222] The prediction module 1702 is used to perform prediction processing on the object feature set and the vehicle control signal set according to the collision condition prediction model to obtain the collision condition feature set.

[0223] The determination module 1703 is used to acquire the passenger feature set of each passenger in the vehicle, and determine the deployment time of each airbag based on the collision condition feature set and the passenger feature set.

[0224] In one exemplary embodiment, the acquisition module 1701 includes a first acquisition submodule and a first determination submodule. The first determination submodule includes:

[0225] The first recognition submodule is used to perform image recognition on each image data in the image dataset based on the image recognition model, and obtain the object to be collided and the initial object feature set of the object to be collided;

[0226] The second determination submodule is used to determine the velocity and volume of the object to be collided with based on the point cloud dataset and the image dataset.

[0227] The first update submodule is used to update the initial object feature set based on the velocity and volume of the object to be collided with, thus obtaining the object feature set.

[0228] In one exemplary embodiment, the first identification submodule includes:

[0229] The second recognition submodule is used to perform image recognition on each image data in the image dataset based on the image recognition model, so as to obtain each initial object to be collided with and the category of each initial object to be collided with.

[0230] The first filtering submodule is used to filter objects to be collided from each initial object to be collided according to preset collision filtering conditions.

[0231] The first query submodule is used to query the quality of the object to be collided in the database based on the category of the object to be collided, and to construct an initial object feature set of the object to be collided based on the category and quality of the object to be collided.

[0232] In an exemplary embodiment, the determining module 1703 includes a second acquiring submodule and a third determining submodule. The third determining submodule includes:

[0233] The fourth determination submodule is used to determine the first collision time between the vehicle and the object to be collided with, based on the vehicle's speed, acceleration, and the first distance between the vehicle and the object to be collided with.

[0234] The fifth determination submodule is used to determine the initial collision acceleration based on the preset collision process time, the speed difference in the collision condition feature set, the object feature set of the object to be collided with, and the vehicle attribute information.

[0235] The sixth determination submodule is used to determine the second collision time for each passenger based on the passenger feature set, initial collision acceleration, and collision pose in the collision condition feature set; the second collision time is the time from the passenger's head and neck to the device inside the vehicle.

[0236] The seventh determination submodule is used to determine the deployment time of the airbag corresponding to the passenger based on the first collision time and the second collision time.

[0237] In one exemplary embodiment, the fourth determining submodule includes:

[0238] The eighth determination submodule is used to determine, for each passenger, a second distance between the passenger's head and neck and the device in the vehicle based on the posture parameters and position vector in the passenger feature set.

[0239] The correction submodule is used to correct the initial collision acceleration based on the seat belt constraint state and basic attributes in the passenger feature set and the collision pose in the collision condition feature set, so as to obtain the collision acceleration.

[0240] The first processing submodule is used to process the collision acceleration and the second distance according to the collision time algorithm to obtain the second collision time.

[0241] In one exemplary embodiment, the active-passive safety fusion object control device 1700 further includes:

[0242] The ninth determination submodule is used to determine whether the airbags corresponding to each passenger need to be deployed based on the passenger feature set and collision condition feature set of each passenger.

[0243] The generation submodule is used to generate a target ignition signal based on the target detonation mode and detonation time corresponding to the target passenger when the target airbag needs to be detonated in the presence of a target passenger.

[0244] The sending submodule is used to send the target ignition signal to the airbag device, instructing the target airbag corresponding to the target passenger to deploy.

[0245] The modules in the aforementioned integrated active and passive safety control system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0246] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 18As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements an object control method that integrates active and passive security. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0247] Those skilled in the art will understand that Figure 18 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0248] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0249] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0250] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0251] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0252] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0253] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method of object control for active and passive safety fusion, characterized by, The method comprises: acquiring an image dataset and a point cloud dataset outside a vehicle, and determining a to-be-collided object and an object feature set of the to-be-collided object based on the image dataset and the point cloud dataset; performing prediction processing on the object feature set and a control signal set of the vehicle according to a collision working condition prediction model to obtain a collision working condition feature set; acquiring a passenger feature set of each passenger in the vehicle, and determining a firing time of each airbag based on the collision working condition feature set and each passenger feature set; the determination of the firing time of each airbag based on the collision working condition feature set and each passenger feature set comprises: determining a first collision time between the vehicle and the to-be-collided object based on a speed, an acceleration of the vehicle, and a first distance between the vehicle and the to-be-collided object; determining an initial collision acceleration according to a preset collision process time, a speed difference in the collision working condition feature set, an object feature set of the to-be-collided object, and attribute information of the vehicle; for each passenger, determining a second collision time of the passenger based on a passenger feature set of the passenger, the initial collision acceleration, and a collision pose in the collision working condition feature set; the second collision time is a time when a head and neck of the passenger collide with an internal device in the vehicle; determining a firing time of an airbag corresponding to the passenger based on the first collision time and the second collision time.

2. The method of claim 1, wherein, the determination of the to-be-collided object and the object feature set of the to-be-collided object based on the image dataset and the point cloud dataset comprises: performing image recognition on each image data in the image dataset based on an image recognition model to obtain an initial object feature set of a to-be-collided object and the to-be-collided object; determining a speed and a volume of the to-be-collided object according to the point cloud dataset and the image dataset; updating the initial object feature set based on the speed and the volume of the to-be-collided object to obtain the object feature set.

3. The method of claim 2, wherein, the image recognition on each image data in the image dataset based on the image recognition model to obtain the initial object feature set of the to-be-collided object and the to-be-collided object comprises: performing image recognition on each image data in the image dataset based on an image recognition model to obtain each initial to-be-collided object and a category of each initial to-be-collided object; screening a to-be-collided object from each initial to-be-collided object according to a preset collision screening condition; querying a mass of the to-be-collided object in a database based on the category of the to-be-collided object, and constructing the initial object feature set of the to-be-collided object according to the category and the mass of the to-be-collided object.

4. The method of claim 1, wherein, the acquisition of the passenger feature set of each passenger in the vehicle comprises: acquiring an attribute feature set of each passenger through each seat sensor in the vehicle; acquiring a posture feature set of each passenger based on an image acquisition device and / or an infrared sensor; constructing a passenger feature set of each passenger according to each attribute feature set and each posture feature set.

5. The method of claim 1, wherein, the determination of the second collision time of the passenger based on the passenger feature set of the passenger, the initial collision acceleration, and the collision pose in the collision working condition feature set comprises: determine, for each of the passengers, a second distance between a head and neck of the passenger and a device in the vehicle based on a posture parameter and a position vector in a passenger feature set of the passenger; correct the initial collision acceleration according to a seat belt restraint state and a basic attribute in the passenger feature set and a collision pose in the collision condition feature set, to obtain a collision acceleration; perform data processing on the collision acceleration and the second distance according to a collision time algorithm, to obtain a second to-be-collision time.

6. The method of claim 1, wherein, The passenger feature set includes the posture parameter and the position vector of the passenger, and the second distance between the head and neck of the passenger and the device in the vehicle is determined based on the posture parameter and the position vector in the passenger feature set of the passenger, including: determining a head and neck coordinate of the passenger based on the posture parameter and the position vector; determining a device coordinate of the device corresponding to the passenger in the vehicle, and determining the second distance between the head and neck of the passenger and the device in the vehicle according to the head and neck coordinate and the device coordinate.

7. The method of claim 1, wherein, After the determination of the ignition time of each airbag based on the collision condition feature set and each passenger feature set, the method further includes: determining whether the airbag corresponding to each passenger needs to be ignited according to the passenger feature set of each passenger and the collision condition feature set; generating a target ignition signal based on a target ignition mode and an ignition time corresponding to the target passenger, in a case where the target airbag of the target passenger needs to be ignited; sending the target ignition signal to an airbag device, to instruct the target airbag corresponding to the target passenger to pop out.

8. An active and passive safety integrated object control device, characterized by, The device includes: an acquisition module configured to acquire an image data set and a point cloud data set outside a vehicle, and determine a to-be-collision object and an object feature set of the to-be-collision object based on the image data set and the point cloud data set; a prediction module configured to perform prediction processing on the object feature set and a control signal set of the vehicle according to a collision condition prediction model, to obtain a collision condition feature set; a determination module configured to acquire a passenger feature set of each passenger in the vehicle, and determine an ignition time of each airbag based on the collision condition feature set and each passenger feature set; The determination module includes a second acquisition submodule and a third determination submodule, and the third determination submodule includes: a fourth determination submodule configured to determine a first collision time between the vehicle and the to-be-collision object based on a speed, an acceleration of the vehicle, and a first distance between the vehicle and the to-be-collision object; a fifth determination submodule configured to determine an initial collision acceleration according to a preset collision process time, a speed difference in the collision condition feature set, an object feature set of the to-be-collision object, and attribute information of the vehicle; a sixth determination submodule configured to determine, for each of the passengers, a second collision time of the passenger based on the passenger feature set of the passenger, the initial collision acceleration, and a collision pose in the collision condition feature set; the second collision time is a time for a head and neck of the passenger to reach a device in the vehicle. A seventh determining sub-module is configured to determine the ignition time of the airbag corresponding to the passenger based on the first collision time and the second collision time. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 7.

Images