Vehicle emergency call method and device, vehicle and storage medium

Abstract

The embodiment of the invention relates to a vehicle emergency call method and device, a vehicle and a storage medium, and the method comprises the steps: collecting the respiratory rate data of a target object in real time under the condition that the vehicle is powered on and started, and collecting the vehicle body acceleration data and wheel rotation speed data of the vehicle; judging whether the vehicle is in an accident triggering state according to the vehicle body acceleration data and the wheel rotating speed data; if the judgment result is that the target object is not in the accident triggering state, inputting the respiratory rate data into the trained neural network model, so that the neural network model outputs the physical sign state of the target object; and triggering an emergency call process when the physical sign state is determined to be abnormal or the judgment result is that the emergency call is in the accident triggering state. Therefore, the sign state of the driver can be continuously sensed and analyzed in combination with the current accident condition of the vehicle, the limitation of an existing emergency call system is made up, and the triggering timeliness and rescue response efficiency of the emergency call system and the reliability of safety guarantee are improved.

Description

Technical Field

[0001] This invention relates to the field of vehicle control technology, and in particular to a vehicle emergency call method, device, vehicle, and storage medium. Background Technology

[0002] With the development of intelligent connected vehicles, emergency call systems, as a vehicle safety feature, have been widely used in post-accident rescue scenarios. Existing emergency call systems typically operate based on vehicle collision detection mechanisms. When a collision occurs, an emergency call process is triggered, establishing communication with a remote rescue platform to provide rescue services to the occupants, or the call process can be manually triggered by the user.

[0003] Existing emergency call systems have relatively simple triggering conditions and generally lack the ability to continuously perceive the driver's state. They are unable to identify and judge sudden physical abnormalities that may occur during driving, especially in autonomous driving or long-distance driving scenarios, where existing systems are unable to respond in a timely manner when the driver experiences a sudden medical condition.

[0004] Furthermore, existing emergency call systems mostly rely on fixed thresholds or rule-based triggering methods, lacking the ability to comprehensively analyze complex scenarios. This makes it difficult to identify and warn of potential risks in advance, resulting in delayed emergency call triggering and missed critical rescue time. Therefore, there is an urgent need for a technical solution that can effectively perceive the driver's state even when no collision has occurred, and automatically trigger an emergency call response when an emergency risk arises, in order to improve the intelligence level and timeliness of vehicle emergency rescue systems. Summary of the Invention

[0005] In view of this, in order to solve the above-mentioned technical problems or some of the technical problems, the present invention provides a vehicle emergency call method, device, vehicle and storage medium.

[0006] In a first aspect, embodiments of the present invention provide a vehicle emergency call method, comprising: When the vehicle is powered on and started, the respiratory rate data of the target object, as well as the vehicle's body acceleration data and wheel speed data are collected in real time. Determine whether the vehicle is in an accident-triggered state based on the vehicle body acceleration data and the wheel speed data; If the judgment result is that the accident is not triggered, the breathing rate data is input into the trained neural network model so that the neural network model outputs the vital signs of the target object. If the vital signs are determined to be abnormal, or if the determination result indicates that the situation is in the accident-triggered state, an emergency call process is triggered.

[0007] In one possible implementation, the neural network model is trained in the following manner: A training dataset is constructed based on the historical respiratory frequency data of the target object collected in history, and the training dataset is preprocessed to form time-series feature data; The time-series feature data is input into an initial model with a preset structure for training. The model parameters are iteratively optimized so that the model can output a prediction result that characterizes whether the physical condition of the target object is abnormal. After the model training is completed and the preset performance indicators are met, the model training is determined to be complete. The trained neural network model is then deployed to the vehicle's onboard controller, and a correspondence between the target object and the neural network model is generated.

[0008] In one possible implementation, the triggering of the emergency call process includes: Control the vehicle to generate an emergency call event; Based on the emergency call event, the vehicle's identity information, location information, current operating status information, and the type of the emergency call event are obtained and sent to the target platform to request the establishment of a communication connection. The type of the emergency call event includes at least accident triggering or abnormal vital signs. After the communication connection is established, control the in-vehicle microphone and in-vehicle speaker to enter the call state.

[0009] In one possible implementation, the method further includes: If the judgment result indicates that the accident is in the triggered state, the emergency call process is triggered, and the respiratory rate data is input into the trained neural network model so that the neural network model outputs the vital signs of the target object. During the execution of the emergency call process, the vital signs status is sent to the target platform.

[0010] In one possible implementation, the method further includes: When multiple target objects exist within the vehicle and the vehicle is not in the accident-triggered state, if any target object is detected to have an abnormal physical condition, the identity information of the abnormal target object is identified. If the identity information indicates the passenger, an alarm event is triggered inside the vehicle.

[0011] In one possible implementation, the method further includes: When the vital signs are determined to be abnormal and the vehicle is in the accident-triggered state, the accident currently triggered by the vehicle is classified according to the vehicle body acceleration data and the wheel rotation speed data to obtain the accident level, and the health risk of the target object is classified according to the vital signs to obtain the health risk level. The current response level of the vehicle is determined based on the accident level and the health risk level. The vehicle is controlled to trigger an emergency call process according to the control strategy corresponding to the response level.

[0012] In one possible implementation, after the neural network model outputs the vital signs of the target object, the method further includes: Based on a preset time window, time series analysis is performed on the vehicle acceleration data, the wheel speed data, and the breathing frequency data to generate the vehicle operation stability change trend and the vital signs change trend. The vehicle's operational stability change trend and the vital signs change trend are input into the trained evaluation model so that the evaluation model outputs the risk prediction result of the vehicle. The risk prediction result represents the first probability that the vehicle will enter an accident-triggered state in the future time period, and / or the second probability that the vital signs are abnormal. If the first probability is greater than the first threshold, or the second probability is greater than the second threshold, the vehicle is controlled to trigger a risk warning process.

[0013] In a second aspect, embodiments of the present invention provide a vehicle emergency call device, comprising: The data acquisition module is used to acquire the breathing rate data of the target object in real time, as well as the vehicle body acceleration data and wheel speed data, when the vehicle is powered on and started. The judgment module is used to determine whether the vehicle is in an accident-triggered state based on the vehicle body acceleration data and the wheel speed data. The processing module is used to input the respiratory rate data into the trained neural network model if the judgment result is that the target object is not in the accident-triggered state, so that the neural network model outputs the vital signs of the target object. The control module is used to trigger an emergency call process when the vital signs are determined to be abnormal, or when the determination result indicates that the accident is triggered.

[0014] Thirdly, embodiments of the present invention provide a vehicle, including: a processor and a memory, wherein the processor is configured to execute a vehicle emergency call program stored in the memory to implement the vehicle emergency call method described in any one of the first aspects above.

[0015] Fourthly, embodiments of the present invention provide a storage medium storing one or more programs, which can be executed by one or more processors to implement the vehicle emergency call method described in any of the first aspects above.

[0016] The vehicle emergency call solution provided in this invention collects the respiratory rate data of the target vehicle, as well as the vehicle's acceleration and wheel speed data in real time when the vehicle is powered on and started. Based on the acceleration and wheel speed data, it determines whether the vehicle is in an accident-triggered state. If the determination is that the vehicle is not in an accident-triggered state, the respiratory rate data is input into a trained neural network model, causing the model to output the target vehicle's vital signs. If the vital signs are determined to be abnormal, or if the determination indicates an accident-triggered state, the emergency call process is triggered. Therefore, by continuously sensing and analyzing the driver's vital signs in conjunction with the vehicle's current accident situation, it achieves proactive identification and automatic response to potential emergency risks, overcoming the limitations of existing emergency call systems. This improves the timeliness of emergency call system triggering and rescue response efficiency, and enhances the reliability of driver safety protection in complex driving scenarios. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a vehicle emergency call method provided in an embodiment of the present invention; Figure 2 A flowchart illustrating another vehicle emergency call method provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a vehicle emergency call device provided in an embodiment of the present invention; Figure 4 This is a structural schematic diagram of a vehicle provided in an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of the present invention.

[0020] Figure 1This is a flowchart illustrating a vehicle emergency call method provided in an embodiment of the present invention, as shown below. Figure 1 As shown, the method specifically includes: S11. When the vehicle is powered on and started, collect the target's breathing rate data in real time, as well as the vehicle's body acceleration data and wheel speed data.

[0021] The vehicle emergency call method provided in this invention is applied to vehicles, including intelligent connected new energy vehicles and fuel vehicles. It is suitable for scenarios involving continuous monitoring of the driver's vital signs during normal driving, autonomous driving, or assisted driving, as well as emergency response scenarios when a vehicle collision occurs or the driver faces a sudden medical risk. By analyzing the driver's vital signs and vehicle accident status in real time during vehicle operation and automatically triggering an emergency call service when necessary, proactive protection of user safety is achieved. The specific execution entity is the control equipment in the vehicle, preferably an onboard controller integrated into the vehicle communication and control system. This includes, but is not limited to, an onboard telematics-BOX (T-BOX) controller, a central control unit, and a continuous damping control (CDC) controller. These controllers have built-in processors, memory, and communication interfaces for executing vital sign monitoring, data analysis, and emergency call control logic.

[0022] In this embodiment, after the vehicle is powered on and started, the vehicle control equipment enters the operating state and initializes and activates various sensors that are communicated with it. The control equipment establishes data acquisition channels with the liveness detection unit, the vehicle body acceleration sensor, and the wheel speed sensor through the vehicle bus (CAN bus).

[0023] The vehicle control equipment establishes communication connections with various sensors through the vehicle's internal communication network (e.g., using the CAN communication protocol). The vehicle's internal communication network is preferably a vehicle controller area network, with a bus structure used for transmitting data packets between functional units. The liveness detection unit, body acceleration sensor, and wheel speed sensor act as data acquisition nodes, periodically or event-triggeredly sending sensor data packets to the control equipment via the vehicle's internal communication network. Each data packet includes at least the acquired data from the corresponding sensor and the corresponding time information. The control equipment, acting as a receiving node on the bus, listens for and receives data packets from different sensors, and parses different types of data according to the packet identifier to obtain breathing frequency data, body acceleration data, and wheel speed data respectively.

[0024] In some implementations, the liveness detection unit can first preprocess the original detection signal through its internal control module, and then send the preprocessed data to the control device through the vehicle's internal communication network to reduce bus load and improve the real-time performance of data transmission.

[0025] The liveness detection unit automatically enters working mode after the vehicle is powered on, continuously monitors the location of the target object, and obtains the respiratory frequency signal of the target object based on the detected minute displacement changes. The respiratory frequency signal is collected in real time at a preset sampling period to form respiratory frequency data.

[0026] Meanwhile, the vehicle body acceleration sensor is used to continuously collect longitudinal, lateral, or multi-axis acceleration changes during vehicle operation to obtain vehicle body acceleration data that reflects the dynamic state of the vehicle; the wheel speed sensor is used to collect the rotational speed information of each wheel in real time to obtain wheel speed data during vehicle operation.

[0027] Breathing frequency data, vehicle acceleration data, and wheel speed data are synchronously transmitted to the control equipment via the vehicle communication bus. The control equipment performs unified reception, time alignment, and buffering processing, providing a data foundation for subsequent status judgment, anomaly detection, and control strategy execution.

[0028] S12. Determine whether the vehicle is in an accident-triggered state based on the vehicle acceleration data and wheel speed data.

[0029] In this embodiment, the accident-triggered state refers to a state in which the vehicle's dynamic characteristics meet preset accident judgment conditions during operation, indicating that the vehicle may have already experienced a collision, instability, rollover, or has a high risk of an accident. Under the accident-triggered state, the vehicle's safety risk increases significantly, requiring the emergency response mechanism to be activated first.

[0030] After acquiring vehicle acceleration data and wheel speed data, the control equipment performs joint analysis on the vehicle acceleration data and wheel speed data to determine whether the vehicle is in an accident-triggered state.

[0031] Specifically, the control equipment first continuously samples and filters the vehicle acceleration data to eliminate instantaneous noise interference, and calculates the amplitude and duration of the vehicle acceleration change based on a preset time window. When the vehicle acceleration experiences a sudden change exceeding a preset threshold within the time window, and the duration of this change reaches a preset condition, the control equipment marks the acceleration change as an abnormal acceleration event. Simultaneously, the control equipment monitors wheel speed data to determine the relationship between the speeds of each wheel and the trend of speed changes. When a sudden drop, significant inconsistency, or mismatch with the vehicle's driving state is detected in the wheel speed within a short period, the control equipment marks the speed change as an abnormal speed event.

[0032] Based on this, the control equipment makes a comprehensive judgment on abnormal acceleration events and abnormal speed events: when abnormal acceleration events and abnormal speed events are detected simultaneously, or when the above two types of abnormal events are detected successively within a preset time correlation range, it is determined that the vehicle is in an accident-triggered state.

[0033] In one possible implementation, to avoid false triggering, the control device can also combine the vehicle's current driving status, historical data, or sensor reliability information to confirm the judgment result, thereby improving the accuracy and stability of the accident trigger judgment.

[0034] Specifically, after initially determining that the vehicle is in an accident-triggered state based on vehicle acceleration and wheel speed data, the control equipment does not directly confirm the accident. Instead, it further verifies the judgment to reduce the probability of false triggering and improve the reliability of the judgment. The control equipment first acquires the vehicle's current driving status information, including whether the vehicle is in motion, accelerating, decelerating, braking, steering, or at a low speed. If the vehicle is in normal driving or smooth braking and no continuous abnormal dynamic changes are detected, the initial judgment is suppressed. If the vehicle is in a high-speed driving, rapid deceleration, or abnormal posture change state, the confidence level of the accident-triggered state is increased. Simultaneously, the control equipment retrieves historical operating data corresponding to the current moment and compares and analyzes the trends of vehicle acceleration and wheel speed data over time. If the current abnormal event has a significant abrupt change compared to historical data and does not conform to the vehicle's previous normal operating patterns, the accident-triggered judgment is further confirmed; conversely, if the abnormal change can be found in historical data with a similar pattern, the priority of the accident-triggered judgment is reduced.

[0035] In addition, the control device evaluates the reliability information of each sensor, which characterizes the stability and reliability of the current sensor data. When the abnormal information output by the vehicle acceleration sensor and the wheel speed sensor is consistent in time and the sensors are operating normally, the control device confirms the accident triggering state; when only a single sensor malfunctions or unstable sensor data is detected, the control device delays or verifies the determination of the accident triggering state.

[0036] S13. If the judgment result is that the object is not in an accident-triggered state, the breathing rate data will be input into the trained neural network model so that the neural network model can output the vital signs of the target object.

[0037] In this embodiment, after confirming that the vehicle is not in an accident-triggered state, the control device reads the real-time collected respiratory rate data from the cache and organizes the respiratory rate data according to a preset time window to form a time-series data sequence for model input. Before inputting the respiratory rate data into the neural network model, the control device preprocesses the respiratory rate data, including data smoothing, outlier removal, and time alignment, to ensure the stability and consistency of the input data.

[0038] Subsequently, the control device inputs the pre-processed respiratory rate data into a pre-trained neural network model deployed within the control device. This neural network model is trained based on historical vital sign samples (historical respiratory rate data) and is used to identify and determine the respiratory rate characteristics and status of the target object.

[0039] The neural network model performs inference calculations on the input respiratory rate data and outputs the vital signs status results corresponding to the current respiratory characteristics. The vital signs status is used to characterize whether the target object is currently in a normal respiratory state, an abnormal respiratory state, or an intermediate state with potential risks.

[0040] The control equipment can also combine the confidence information output by the neural network model to confirm the results of the vital signs status, and use the vital signs status as the basis for subsequent early warning judgment and emergency call triggering.

[0041] In one possible implementation, the neural network model is trained in the following way: A training dataset is constructed based on historical respiratory frequency data of the target object collected in history. The training dataset is preprocessed to form time-series feature data. The time-series feature data is input into an initial model with a preset structure for training. The model parameters are iteratively optimized so that the model can output a prediction result that characterizes whether the target object's vital signs are abnormal. After the model training is completed and the preset performance indicators are met, the model training is determined to be complete. The trained neural network model is deployed to the vehicle's on-board controller, and the correspondence between the target object and the neural network model is generated.

[0042] In this embodiment, based on the historical respiratory frequency data of the target object collected in the past, the control device constructs a training dataset for training the neural network model. The historical respiratory frequency data is continuously collected by the liveness detection unit during vehicle operation and is associated and stored according to the target object to form a historical respiratory frequency data set corresponding to the target object.

[0043] After constructing the training dataset, the control device preprocesses the training dataset to form temporal feature data for model training. The preprocessing operations include: time alignment, noise suppression, and outlier removal of historical respiratory frequency data, and segmenting the processed respiratory frequency data according to a preset time window to generate temporal feature data reflecting the respiratory change trend of the target object.

[0044] The time-series feature data is input into an initial model with a pre-defined structure for training. The initial model is a neural network model with a pre-defined structure, used for feature extraction and state discrimination of respiratory rate changes in the target object. During model training, the control device calculates the difference between the model output and the actual vital signs based on the time-series feature data and the corresponding vital sign state annotations. It then iteratively optimizes and updates the model parameters, gradually enabling the neural network model to output predictive results characterizing whether the target object's vital signs are abnormal.

[0045] During model training, the control device continuously evaluates the training effect of the neural network model. When the model's prediction accuracy, stability, or other preset performance indicators on the validation dataset meet the preset requirements, the neural network model training is considered complete.

[0046] After the model training is complete, the control equipment deploys the trained neural network model to the vehicle's onboard controller (also known as the control device). This allows the onboard controller to call the neural network model to analyze and process the real-time respiratory rate data of the target object during vehicle operation. Simultaneously, the control equipment generates and stores the correspondence between the target object's identity information and the neural network model. This correspondence is used in subsequent operations to call the appropriate neural network model to determine the target object's vital signs. Specifically, the target object's identity information can be determined through facial feature recognition. Based on the identity information, the corresponding neural network model is determined from the correspondence. This neural network model is pre-trained for the target object with that specific identity information and is fully adapted to it.

[0047] As an example, during vehicle operation, the T-BOX or CDC controller monitors and collects driver breathing frequency data reported by the liveness detection unit in real time via the CAN bus. The controller first performs preprocessing operations on the monitored and collected data, including noise reduction, time alignment, and outlier filtering, and extracts a large amount of training data (t) over continuous time periods for subsequent neural network model training.

[0048] After data preparation, the controller uses a convolutional neural network (CNN) built based on the deep reinforcement concept as the initial model to process the time-series data model(t) of respiratory rate formation. The CNN can automatically extract the features of respiratory rate changes over time, thereby establishing a correspondence between respiratory patterns and the driver's vital signs.

[0049] During model training, the controller defines a loss function, loss_fn, to measure the difference between the model's predictions and the actual vital signs. This loss function guides the adjustment of model parameters to reflect the current model's accuracy in identifying abnormal respiratory states.

[0050] Meanwhile, the controller selects stochastic gradient descent as the optimizer to update the weights and biases in the neural network model. The optimizer gradually adjusts the model parameters based on the calculated loss_fn.

[0051] Subsequently, the controller uses the training data (t) collected and preprocessed in the first step to train the CNN model. During training, the controller outputs the prediction result based on model(t), calculates the corresponding loss_fn, and updates the model parameters through the optimizer, thereby continuously improving the model's ability to identify abnormal respiratory states.

[0052] After model training is complete, the controller evaluates the model performance using data not used in training. Evaluation metrics include at least accuracy, precision, and recall, used to verify the model's predictive effectiveness and stability under different physiological states. When the model evaluation results do not meet preset requirements, the controller adjusts the hyperparameters used in the model training process, such as adjusting the learning rate and batch size, and re-executes the model training process to improve model performance and convergence speed.

[0053] After model training and optimization, the controller deploys the trained neural network model to the edge-side task application of T-BOX or CDC, and converts the model into executable program code. Simultaneously, the controller creates a corresponding API interface for the neural network model, allowing the onboard control program to call the model in real time during vehicle operation to analyze the driver's current respiratory rate data and output predicted vital signs.

[0054] S14. If the vital signs are determined to be abnormal, or if the judgment result indicates that the situation is in an accident-triggered state, the emergency call process is triggered.

[0055] In this embodiment, if the target object's vital signs are determined to be abnormal, or if the vehicle is determined to be in an accident-triggered state, the control device initiates an emergency call process to achieve timely rescue of the target object.

[0056] Specifically, when the neural network model outputs a vital sign status indicating that the target object has abnormal breathing, apnea, or other preset high-risk conditions, the control device marks the vital sign status as a vital sign abnormality trigger condition to trigger the emergency call process. Alternatively, when the joint judgment result based on vehicle acceleration data and wheel speed data indicates that the vehicle is in an accident-triggered state, the control device marks the judgment result as an accident-triggered condition to trigger the emergency call process.

[0057] Upon detecting that any triggering condition is met, the control device sends an emergency call command to the emergency call unit. The emergency call command includes at least a trigger type identifier, indicating whether the trigger was due to abnormal vital signs or an accident, vehicle identification information, and the current time. Upon receiving the emergency call command, the emergency call unit automatically establishes a communication connection with the remote emergency rescue platform.

[0058] During the communication establishment process, the control equipment simultaneously sends the vehicle's basic information, current location information, and emergency call instructions to the remote emergency rescue platform, so that the rescue platform can know the risk status of the vehicle and the target object in advance.

[0059] Once communication is established, the emergency call unit automatically activates the in-vehicle audio acquisition function, using the in-vehicle microphone to collect voice information about the target's environment and transmit the voice information to the remote emergency rescue platform in real time, so as to support rescue personnel to conduct voice interaction or status confirmation with the target.

[0060] In some implementations, after the emergency call process is initiated, the control device can also continuously collect the target object's vital signs data and vehicle operation status data, and periodically or through event-triggered methods send updated data to the remote emergency rescue platform to assist rescue personnel in dynamically assessing the target object's current status.

[0061] In one possible implementation, triggering the emergency call process includes: Control the vehicle to generate an emergency call event; based on the emergency call event, obtain the vehicle's identity information, location information, current operating status information, and the type of emergency call event, and send them to the target platform to request the establishment of a communication connection. The type of emergency call event must include at least accident triggering or abnormal vital signs; after the communication connection is established, control the in-vehicle microphone and in-vehicle speaker to enter the call state.

[0062] In this embodiment, when an accident-triggered state is detected in the vehicle or an abnormal condition of the target object, the control device sends an emergency call trigger command to the vehicle control system to control the vehicle to generate an emergency call event. The emergency call event is used to characterize an emergency situation requiring external rescue intervention.

[0063] After an emergency call event is generated, the control equipment acquires basic and status information related to the vehicle based on the event. The basic information includes at least the vehicle's identification information, used to uniquely identify the vehicle; the status information includes at least the vehicle's location information and current operating status information. Specifically, the vehicle identification information is pre-set at the factory and stored in the onboard control system, the location information is acquired in real-time by the vehicle positioning module, and the current operating status information is generated by the onboard control system based on vehicle sensor data. The control equipment types the emergency call event to determine its type. The types of emergency call events include at least accident triggering types (indicating a vehicle accident) and abnormal vital sign types (indicating a target object is in danger), used to distinguish different emergency scenarios and provide corresponding event background information to the target platform.

[0064] After completing the above information collection, the control device packages the vehicle identity information, location information, current operating status information, and emergency call event type into an emergency call request message, and sends it to the target platform through the vehicle communication unit to request the establishment of a communication connection with the target platform.

[0065] Upon receiving the emergency call request message, the target platform establishes a communication connection with the vehicle. Once the communication connection is successfully established, the control equipment activates the in-vehicle microphone and speakers to enter call mode, thereby enabling in-vehicle audio acquisition and playback functions. This allows remote personnel on the target platform to conduct two-way voice communication with people inside the vehicle, facilitating status confirmation and rescue communication in emergency situations.

[0066] In some implementations, during the call, the control device may periodically send updated vehicle location information or operating status information to the target platform to help the target platform dynamically grasp the current status of the vehicle and the target object.

[0067] In one possible implementation, the method further includes: If the judgment result indicates that an accident has been triggered, the emergency call process is initiated, and the respiratory rate data is input into the trained neural network model so that the neural network model outputs the vital signs status of the target object; during the execution of the emergency call process, the vital signs status is sent to the target platform.

[0068] In this embodiment, when not in an accident-triggered state, respiratory rate data is continuously monitored. The emergency call process is only triggered when the respiratory rate data indicates that the target's vital signs are abnormal. If an accident-triggered state is in effect, the emergency call process is triggered directly. Simultaneously, the respiratory rate data of the current target is acquired, and the vital signs status is output for synchronous transmission to the target platform.

[0069] Specifically, when the control equipment determines that the vehicle is in an accident-triggered state based on vehicle acceleration data and wheel speed data, it immediately issues an emergency call trigger command to the vehicle control system to control the vehicle to generate an emergency call event and request to establish a communication connection with the target platform. Simultaneously with the emergency call process initiation, the control equipment acquires the respiratory rate data of the target object from the liveness detection unit and processes and preprocesses the data according to a preset time window. The preprocessed respiratory rate data is then input into a trained neural network model deployed in the onboard controller, enabling the model to analyze the respiratory characteristics of the target object at the time of the accident and output model predictions characterizing the target object's current vital signs.

[0070] During the execution of the emergency call process, once the control device obtains the vital signs status output by the neural network model, it associates the vital signs status with the emergency call event and sends it to the target platform as supplementary information. This allows the target platform to synchronously obtain the vital signs status of the target object during the establishment of a communication connection or the call. The control device can also periodically send updated vital signs status to the target platform during the duration of the emergency call process, or when the vital signs status changes, to assist the target platform in continuously assessing the current health status of the target object and making rescue decisions.

[0071] In one possible implementation, when multiple target objects are present in the vehicle and the vehicle is not in an accident-triggered state, if any target object is detected to have an abnormal physical condition, the identity information of the abnormal target object is identified; if the identity information is that of a passenger, an alarm event is triggered in the vehicle.

[0072] In this embodiment, the control device collects respiratory rate data for multiple target objects within the vehicle using liveness detection sensors, cabin perception sensors, or occupant monitoring modules deployed inside the vehicle. Each target object is assigned a unique object identifier to distinguish the source of its vital signs data. Based on the object identifier, the control device inputs the collected respiratory rate data into a trained neural network model to obtain the output results of the vital signs status for each target object. When the control device detects that the vital signs status of at least one target object is abnormal, it extracts its object identifier from the vital signs data link corresponding to the abnormal target object and further obtains the identity information of the abnormal target object based on the object identifier. The identity information characterizes the target object's role type within the vehicle, including at least the driver or passenger.

[0073] During the identity verification process, the control device can combine the spatial relationship between the target object and the driver's seat, the seat occupancy status, the seat belt status, or historical identity binding information to confirm whether the abnormal target object is a passenger. When the identity information of the abnormal target object is confirmed to be a passenger, the control device does not trigger the emergency call procedure, but instead triggers an alarm event inside the vehicle, including: controlling the in-vehicle speakers to output voice alarm information, controlling the central control display to display abnormal prompt information, or controlling the in-vehicle ambient lighting and seat vibration module to execute alarm actions, so as to remind other target objects in the vehicle or the driver to pay attention to the passenger's abnormal physical condition.

[0074] In one possible implementation, the method further includes: When the vital signs are determined to be abnormal and the vehicle is in an accident-triggered state, the accident currently triggered by the vehicle is classified into levels based on the vehicle body acceleration data and wheel speed data to obtain the accident level, and the health risk of the target object is classified into levels based on the vital signs to obtain the health risk level; the current response level of the vehicle is determined based on the accident level and the health risk level; and the vehicle is controlled to trigger the emergency call process according to the control strategy corresponding to the response level.

[0075] In this embodiment, when it is determined that the target object's vital signs are abnormal and the vehicle is in an accident-triggered state, the control device comprehensively assesses the severity of the current accident and the health risk level of the target object to determine the vehicle's response level and execute the corresponding emergency call process control strategy accordingly.

[0076] Specifically, the control equipment first classifies the currently triggered accident based on the collected vehicle acceleration and wheel speed data. The control equipment performs amplitude and duration analysis on the vehicle acceleration data to characterize the impact intensity during a collision or instability; simultaneously, it performs rate of change and abnormal state analysis on the wheel speed data to characterize whether the vehicle exhibits accident characteristics such as sudden deceleration, skidding, rollover, or overturning. The control equipment then matches the combined characteristics of the vehicle acceleration and wheel speed data with preset accident classification rules to determine the vehicle's current accident level.

[0077] While obtaining the accident level, the control equipment classifies the health risk of the target object based on the abnormal vital signs. Specifically, the control equipment quantifies the health risk of the target object based on the degree of abnormality, duration of abnormality, and trend of abnormality output by the neural network model, and determines the corresponding health risk level of the target object according to the preset health risk level classification rules.

[0078] After obtaining the accident level and health risk level respectively, the control equipment uses the accident level and health risk level as joint decision inputs, and determines the vehicle's current response level based on a preset response level mapping relationship. The response level is used to comprehensively characterize the combined impact of the accident severity and the health risk to the target object; the higher the level, the more urgent the emergency response measures that the vehicle needs to take.

[0079] After determining the vehicle's current response level, the control equipment controls the vehicle to trigger the emergency call process according to the control strategy corresponding to the response level. Specifically, the control equipment selects the corresponding emergency call process mode based on the response level and controls the emergency call unit to send emergency call event information containing the accident level, health risk level, and response level to the target platform to request the establishment of a communication connection; after the communication connection is established, the control equipment controls the in-vehicle microphone and in-vehicle speaker to enter the call state to support real-time interaction of emergency rescue services.

[0080] By employing the above methods, when abnormal vital signs and accident-triggered conditions occur simultaneously, a joint assessment of the severity of the accident and the degree of health risk is achieved. Based on the assessment results, the response level is dynamically determined, enabling the emergency call process to be triggered and executed in a differentiated manner according to the actual risk level, thereby improving the intelligence and response accuracy of the vehicle emergency call system.

[0081] As an example, accident levels can be categorized as follows: Accident Level 1: Minor Accident Level. This level indicates that the vehicle's acceleration data shows short-term fluctuations, but the peak value is below the first acceleration threshold; wheel speed data shows a momentary drop, but does not remain at zero or change in the opposite direction; the ESP does not output rollover or severe instability signals. This can include scenarios such as low-speed rear-end collisions, minor scrapes, and emergency braking without structural collisions. In this case, the vehicle's structural risk is low, and the direct impact of the accident on the driver is relatively limited.

[0082] Accident Level Two: Moderate Accident Level. This level is characterized by vehicle acceleration exceeding the first acceleration threshold for a duration exceeding the first time threshold; a sharp drop in wheel speed or a significant difference in wheel speed between the left and right wheels; and the ESP outputting a vehicle instability signal, but without detecting rollover. This can include scenarios such as moderate-speed frontal or side collisions, or vehicles exhibiting significant loss of control but remaining upright. It indicates a clear safety risk to the vehicle, and the driver may experience impact or a stress response.

[0083] Accident Level 3: Severe Accident Level. This level indicates that the vehicle's acceleration exceeds the second acceleration threshold; wheel speed rapidly drops to zero or exhibits abnormal reverse changes; and the ESP outputs rollover, severe tilt, or multiple instability signals. This can include scenarios such as high-speed collisions, rollover accidents, and multiple consecutive impacts. It signifies that the vehicle and driver are in a high-risk state, requiring immediate implementation of the highest level of emergency response.

[0084] A strategy for classifying health risk levels can be based on the vital signs output by control devices using a neural network model, thereby grading the health risk of a target individual. This can include the following levels: Health Risk Level 1: Mild Risk, including: vital signs: respiratory rate deviates from the normal range but remains continuous; respiratory rhythm shows an irregular trend but is not interrupted; neural network model outputs low-confidence abnormal results. This characterizes the target subject's physical signs such as stress, short-term discomfort, and fluctuating respiratory rhythm.

[0085] Health Risk Level Two: Moderate Risk, including: Significantly abnormal respiratory rate, weakened respiratory amplitude, or intermittent pauses; the neural network model continuously outputs abnormal prediction results. This indicates respiratory distress, a strong stress response, and potential health risks in the target subject.

[0086] Health Risk Level 3: Severe Risk, including: vital signs: extremely low respiratory rate or detected apnea; prolonged absence of respiratory signals; abnormally high confidence output of neural network model. Characterized by coma, sudden illness, or significantly abnormal vital signs in the target subject.

[0087] The strategy for determining the response level (accident level × health risk level) involves the control equipment making a fusion decision based on the accident level and health risk level to determine the vehicle's current response level, including: Response Level 1: Alert-type response. Example trigger combinations: Incident Level 1 + Health Risk Level 1; Incident Level 1 + Health Risk Level 2. Response characteristics: Risk is controllable, primarily focused on alerts and monitoring.

[0088] Response Level Two: Enhanced Response. Triggering combinations include: Incident Level Two + Health Risk Level One, Incident Level Two + Health Risk Level Two, and Incident Level One + Health Risk Level Three. Response characteristics: Significant risk, requiring proactive intervention and manual confirmation.

[0089] Response Level 3: Emergency Response. Triggering combination examples: Accident Level 3 + any health risk level, or any accident level + health risk level 3. Response characteristics: Highly dangerous, requiring immediate emergency response.

[0090] Examples of control strategies corresponding to different response levels: Response Level 1: Trigger an alarm event inside the vehicle; continuously collect vital signs data and increase the sampling frequency; do not trigger the remote emergency call process. Response Level 2: Automatically generate an emergency call event; send vehicle identity information, location information, accident level, and health risk level to the target platform; after establishing a communication connection, prioritize requesting manual confirmation; control the in-vehicle microphone and speakers to enter call mode. Response Level 3: Immediately trigger the emergency call process; send complete accident level, health risk level, and response level information to the target platform; automatically enter an emergency rescue mode without confirmation; continuously report changes in vital signs to the target platform.

[0091] In one possible implementation, based on a preset time window, time series analysis is performed on vehicle acceleration data, wheel speed data, and breathing frequency data to generate trends in vehicle operational stability and vital signs. These trends are then input into a trained evaluation model, which outputs a risk prediction result for the vehicle. The risk prediction result represents a first probability that the vehicle will enter an accident-triggered state in the future time period, and / or a second probability that its vital signs are abnormal. If the first probability is greater than a first threshold, or the second probability is greater than a second threshold, the vehicle is controlled to trigger a risk warning process.

[0092] In this embodiment, during vehicle power-on operation, the control device continuously receives vehicle acceleration data, wheel speed data, and breathing frequency data. To avoid interference from instantaneous noise or occasional fluctuations in the judgment results, the control device uses a preset time window as the basic analysis unit to perform time series analysis on the above data. The preset time window is a continuous time interval, such as several seconds to tens of seconds; the time window is updated in a sliding manner so that the analysis results can reflect the dynamic changes in the data. The vehicle acceleration data, wheel speed data, and breathing frequency data are timestamped; missing data is interpolated for compensation, and abnormal peak data is smoothed; the processed data are then arranged in chronological order to form a multidimensional time series.

[0093] The control equipment extracts temporal features that characterize the vehicle's operating state based on the vehicle's acceleration data and wheel speed data within a preset time window, and generates a trend of vehicle operating stability changes.

[0094] Specifically, this includes: Vehicle acceleration data analysis: analyzing the magnitude, rate of change, and directional consistency of acceleration amplitude; determining whether there are unstable characteristics such as continuously increasing fluctuations or frequent abrupt changes. Wheel speed data analysis: analyzing the synchronicity and changing trends of the speeds of each wheel; determining whether there are sudden drops in speed or a continuous widening of the speed difference between the left and right wheels. Trend generation: integrating the above analysis results to form a trend curve reflecting the change of vehicle stability over time; the trend of vehicle operational stability is used to characterize the process of the vehicle evolving from a stable state to an unstable state.

[0095] The method for generating the trend of changes in vital signs is as follows: the control device performs time-series analysis on the changes in vital signs of the target object based on respiratory rate data within a preset time window, and generates the trend of changes in vital signs.

[0096] Specifically, this includes: Temporal analysis of respiratory rate: calculating the mean, fluctuation amplitude, and direction of change of respiratory rate; analyzing whether the respiratory rhythm slows down, becomes disordered, or is intermittently interrupted. Trend modeling: constructing a trend of changes in vital signs based on changes in respiratory rate over a continuous period; this trend is used to characterize the risk level of the target subject's vital signs evolving from normal to abnormal.

[0097] After generating trends in vehicle operational stability and vital sign status, the control equipment uses both as joint inputs to feed into the trained evaluation model. The input data includes trends in vehicle operational stability and vital sign status; the input data is time-series data, reflecting the future direction of evolution. Based on historically learned correlations, the evaluation model outputs at least one of the following predictions: a first probability representing the vehicle entering an accident-triggered state in the future, and a second probability representing the target object's vital sign status being abnormal in the future. The first probability is used to assess vehicle operational risks in advance, and the second probability is used to assess the target object's health risks in advance.

[0098] The control equipment performs threshold judgments on the risk prediction results output by the assessment model to determine whether early intervention is necessary. When the first probability is greater than a preset first threshold, it is determined that the vehicle has a high probability of entering an accident-triggered state; when the second probability is greater than a preset second threshold, it is determined that the target object has a high probability of exhibiting abnormal vital signs. If the first probability is greater than the first threshold, or the second probability is greater than the second threshold, the control equipment controls the vehicle to trigger a risk warning process. This risk warning process is used to intervene in advance on the vehicle or target object or to provide in-vehicle warnings before an accident or abnormal vital signs occur.

[0099] The vehicle emergency call system provided in this invention integrates driver vital sign monitoring with vehicle accident detection by introducing a liveness detection unit into the vehicle and combining it with a neural network algorithm. This enables multi-stage proactive perception and intelligent judgment before, during, and after a collision. Through collaborative analysis of multi-source data such as vehicle acceleration, wheel speed, and respiratory rate, it can identify driver health abnormalities and vehicle risk states earlier and more accurately. It automatically triggers corresponding emergency call and rescue procedures based on different scenarios, significantly improving the coverage and timeliness of the vehicle emergency call system for medical emergencies, reducing the risk of rescue delays due to sudden health risks or accidents, and enhancing the overall vehicle safety level.

[0100] This invention also provides a vehicle emergency call system, which, by adding a liveness detection unit and introducing a neural network intelligent algorithm into the main control program, enables proactive monitoring of the driver's vital signs and provides early warning and emergency response when the driver's breathing is abnormal. The system includes at least: a vehicle acceleration sensor, wheel speed sensors, a liveness detection unit, a T-BOX or CDC controller, an emergency call unit, an in-vehicle microphone, and a neural network intelligent algorithm model.

[0101] Among them, the T-BOX or CDC controller serves as the main controller, embedding a neural network intelligent algorithm model, while the remaining units serve as data acquisition units or execution units, working in conjunction with the main controller.

[0102] The vehicle acceleration sensor, wheel speed sensor, liveness detection unit, T-BOX or CDC controller, emergency call unit, and neural network intelligent algorithm model all communicate via the vehicle's CAN bus to achieve data transmission and control command interaction. Data collected by each sensor is sent to the T-BOX or CDC controller via the CAN bus, where the main control program receives, processes, and responds with control measures.

[0103] Vehicle acceleration sensors are used to collect real-time data on vehicle acceleration in all directions. When a vehicle is involved in a frontal, side, or rear-end collision, its acceleration changes significantly, providing fundamental data for accident assessment.

[0104] Wheel speed sensors are used to collect the rotational speed data of each wheel. When a collision occurs, or when the ESP detects that the vehicle is tilting or rolling over significantly, the wheel speed sensors send a vehicle instability signal to the emergency call system to help improve the accuracy of accident assessment.

[0105] The liveness detection unit, based on the Doppler frequency shift principle, performs non-contact detection of the driver inside the vehicle, identifies and calculates the driver's breathing rate in real time, and reflects the driver's current vital signs.

[0106] The T-BOX or CDC controller is used to collect data signals from the vehicle acceleration sensor, wheel speed sensor and liveness detection unit, and input the collected data into the neural network intelligent algorithm model for automated detection and analysis.

[0107] After obtaining the analysis results, the T-BOX or CDC controller outputs control commands to the corresponding execution unit based on the calculation results.

[0108] The neural network intelligent algorithm model is an algorithm model that has been trained and optimized with a large amount of prototype vehicle data. It is used to monitor the respiratory rate signal samples collected by the liveness detection unit in real time and automatically identify abnormal scenarios such as abnormal driver breathing, thereby achieving early warning.

[0109] The emergency call unit is used to initiate remote emergency rescue call service when the warning conditions of the liveness detection unit are triggered or the neural network intelligent algorithm model determines that the driver has a risk of abnormal breathing. The T-BOX or CDC controller sends a trigger command to the emergency call unit via the CAN bus.

[0110] The in-vehicle microphone, once the emergency call unit is activated, enters working mode to collect ambient audio information around the driver, helping remote emergency rescue personnel to communicate with the driver via audio and further confirm the driver's medical condition.

[0111] Figure 2 The diagram shown is a flowchart of another vehicle emergency call method provided by an embodiment of the present invention.

[0112] When the vehicle's high-voltage power system is powered on, the vehicle acceleration sensor, wheel speed sensors, and liveness detection unit start simultaneously. The vehicle acceleration sensor and wheel speed sensors collect information on acceleration changes and wheel speeds during vehicle movement, respectively, while the liveness detection unit continuously collects vital signs signals such as the driver's breathing rate or respiratory count. The signals collected by each sensor are input to the T-BOX controller's main control program via the vehicle communication link.

[0113] Inside the T-BOX controller, the main control program runs a collision triggering algorithm and a neural network intelligent algorithm model in parallel. The collision triggering algorithm determines whether the vehicle is in an accident-triggered state based on vehicle acceleration and wheel speed data; the neural network intelligent algorithm model analyzes the breathing signals collected by the liveness detection unit to identify whether the driver has health risks such as abnormal breathing. When either algorithm meets the preset triggering conditions, the main control program sends a trigger command to the emergency call unit.

[0114] Once the emergency call unit is triggered, the emergency call process is automatically initiated, and the in-vehicle microphone is activated to collect in-vehicle audio information. A communication connection is established with platforms such as human customer service or medical assistance to confirm the driver's current condition and generate a corresponding medical emergency rescue plan. Through this process, this solution achieves joint perception and intelligent judgment of the vehicle accident status and the driver's vital signs, improving the emergency call system's response capability in medical risk scenarios.

[0115] Figure 3 The diagram shown is a structural schematic of a vehicle emergency call device provided in an embodiment of the present invention. Figure 3 As shown, the device specifically includes: The acquisition module 31 is used to acquire the breathing rate data of the target object in real time, as well as the vehicle body acceleration data and wheel speed data, when the vehicle is powered on and started. The judgment module 32 is used to determine whether the vehicle is in an accident-triggered state based on the vehicle body acceleration data and the wheel speed data; Processing module 33 is used to input the respiratory rate data into the trained neural network model if the judgment result is that the target object is not in the accident triggering state, so that the neural network model outputs the vital signs status of the target object. Control module 34 is used to trigger an emergency call process when the vital signs are determined to be abnormal, or when the determination result indicates that the accident is triggered.

[0116] In one possible implementation, the processing module is further configured to construct a training dataset based on historical respiratory frequency data of the target object collected in the past, and to preprocess the training dataset to form time-series feature data. The time-series feature data is input into an initial model with a preset structure for training. The model parameters are iteratively optimized so that the model can output a prediction result that characterizes whether the physical condition of the target object is abnormal. After the model training is completed and the preset performance indicators are met, the model training is determined to be complete. The trained neural network model is then deployed to the vehicle's onboard controller, and a correspondence between the target object and the neural network model is generated.

[0117] In one possible implementation, the control module is specifically configured to control the vehicle to generate an emergency call event; Based on the emergency call event, the vehicle's identity information, location information, current operating status information, and the type of the emergency call event are obtained and sent to the target platform to request the establishment of a communication connection. The type of the emergency call event includes at least accident triggering or abnormal vital signs. After the communication connection is established, control the in-vehicle microphone and in-vehicle speaker to enter the call state.

[0118] In one possible implementation, the control module is further configured to, if the determination result indicates that the accident is in the triggered state, simultaneously trigger the emergency call process and input the respiratory rate data into the trained neural network model, so that the neural network model outputs the vital signs of the target object. During the execution of the emergency call process, the vital signs status is sent to the target platform.

[0119] In one possible implementation, the control module is further configured to identify the identity information of the abnormal target object if the vital signs of any target object are detected to be abnormal when multiple target objects are present in the vehicle and the vehicle is not in the accident-triggered state. If the identity information indicates the passenger, an alarm event is triggered inside the vehicle.

[0120] In one possible implementation, the control module is further configured to, when determining that the vital signs are abnormal and the vehicle is in the accident-triggered state, classify the accident currently triggered by the vehicle according to the vehicle body acceleration data and the wheel rotation speed data to obtain an accident level, and classify the health risk of the target object according to the vital signs to obtain a health risk level. The current response level of the vehicle is determined based on the accident level and the health risk level. The vehicle is controlled to trigger an emergency call process according to the control strategy corresponding to the response level.

[0121] In one possible implementation, the control module is further configured to perform time series analysis on the vehicle acceleration data, the wheel rotation speed data, and the breathing frequency data based on a preset time window, and generate trends in vehicle operating stability and vital signs. The vehicle's operational stability change trend and the vital signs change trend are input into the trained evaluation model so that the evaluation model outputs the risk prediction result of the vehicle. The risk prediction result represents the first probability that the vehicle will enter an accident-triggered state in the future time period, and / or the second probability that the vital signs are abnormal. If the first probability is greater than the first threshold, or the second probability is greater than the second threshold, the vehicle is controlled to trigger a risk warning process.

[0122] The vehicle emergency call device provided in this embodiment can be as follows: Figure 3 The apparatus shown can perform, as Figure 1-2 All steps of the vehicle emergency call method, thereby achieving Figure 1-2 For details on the technical effects of the vehicle emergency call method shown, please refer to [link / reference]. Figure 1-2 The relevant descriptions are presented concisely and will not be elaborated upon here.

[0123] Figure 4 This is a schematic diagram of the structure of a vehicle provided in an embodiment of the present invention. Figure 4 The vehicle 400 shown includes at least one processor 401, a memory 402, at least one network interface 404, and other user interfaces 403. Various components in the vehicle 400 are coupled together via a bus system 405. It is understood that the bus system 405 is used to implement communication between these components. In addition to a data bus, the bus system 405 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in… Figure 4 The general designated all buses as Bus System 405.

[0124] The user interface 403 may include a display, keyboard, or clicking device (e.g., mouse, trackball, touchpad, or touchscreen).

[0125] It is understood that the memory 402 in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate Synchronous DRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memory 402 described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0126] In some implementations, memory 402 stores elements, executable units or data structures, or subsets thereof, or extended sets thereof: operating system 4021 and application program 4022.

[0127] The operating system 4021 includes various system programs, such as the framework layer, core library layer, and driver layer, used to implement various basic business functions and handle hardware-based tasks. The application program 4022 includes various applications, such as a media player and a browser, used to implement various application functions. The program implementing the method of this embodiment can be included in the application program 4022.

[0128] In this embodiment of the invention, by calling the program or instructions stored in the memory 402, specifically the program or instructions stored in the application program 4022, the processor 401 executes the method steps provided in each method embodiment, including, for example: When the vehicle is powered on and started, the respiratory rate data of the target object, as well as the vehicle's body acceleration data and wheel speed data are collected in real time. Determine whether the vehicle is in an accident-triggered state based on the vehicle body acceleration data and the wheel speed data; If the judgment result is that the accident is not triggered, the breathing rate data is input into the trained neural network model so that the neural network model outputs the vital signs of the target object. If the vital signs are determined to be abnormal, or if the determination result indicates that the situation is in the accident-triggered state, an emergency call process is triggered.

[0129] The methods disclosed in the above embodiments of the present invention can be applied to processor 401, or implemented by processor 401. Processor 401 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 401 or by instructions in the form of software. The processor 401 may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of the present invention can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software units in the decoding processor. The software units may be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 402. Processor 401 reads the information in memory 402 and, in conjunction with its hardware, completes the steps of the above method.

[0130] It is understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described herein, or combinations thereof.

[0131] For software implementation, the techniques described herein can be implemented by units that perform the functions described herein. The software code can be stored in memory and executed by a processor. The memory can be implemented in the processor or external to the processor.

[0132] The vehicle provided in this embodiment can be as follows: Figure 4 The vehicle shown can perform the following actions: Figure 1-2 All steps of the vehicle emergency call method, thereby achieving Figure 1-2 For details on the technical effects of the vehicle emergency call method shown, please refer to [link / reference]. Figure 1-2 The relevant descriptions are presented concisely and will not be elaborated upon here.

[0133] This invention also provides a storage medium (computer-readable storage medium). This storage medium stores one or more programs. The storage medium may include volatile memory, such as random access memory; it may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive; and it may also include combinations of the above types of memory.

[0134] One or more programs in the storage medium can be executed by one or more processors to implement the vehicle emergency call method described above that is executed on the device side.

[0135] The processor is used to execute a vehicle emergency call program stored in the memory to implement the following steps of a vehicle emergency call method executed on the device side: When the vehicle is powered on and started, the respiratory rate data of the target object, as well as the vehicle's body acceleration data and wheel speed data are collected in real time. Determine whether the vehicle is in an accident-triggered state based on the vehicle body acceleration data and the wheel speed data; If the judgment result is that the accident is not triggered, the breathing rate data is input into the trained neural network model so that the neural network model outputs the vital signs of the target object. If the vital signs are determined to be abnormal, or if the determination result indicates that the situation is in the accident-triggered state, an emergency call process is triggered.

[0136] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0137] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented in hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0138] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A vehicle emergency call method characterized by, The method comprises the following steps: In the case of starting the vehicle, the breathing frequency data of the target object is collected in real time, and the body acceleration data and wheel speed data of the vehicle are collected; According to the body acceleration data and the wheel speed data, it is determined whether the vehicle is in an accident triggering state; If the determination result is that the vehicle is not in the accident triggering state, the breathing frequency data is input into the trained neural network model, so that the neural network model outputs the vital sign state of the target object; In the case where it is determined that the vital sign state is abnormal, or the determination result is that the vehicle is in the accident triggering state, an emergency call process is triggered.

2. The method of claim 1, wherein, The neural network model is trained in the following way: Based on the historical breathing frequency data of the target object collected in the past, a training data set is constructed, and the training data set is preprocessed to form time series feature data; The time series feature data is input into an initial model with a preset structure for training, and the model parameters are iteratively optimized so that the model can output a prediction result for indicating whether the vital sign state of the target object is abnormal; After the model training is completed and the preset performance indicators are met, it is determined that the model training is completed, the trained neural network model is deployed to the vehicle-mounted controller of the vehicle, and a corresponding relationship between the target object and the neural network model is generated.

3. The method of claim 1, wherein, The emergency call process comprises the following steps: Controlling the vehicle to generate an emergency call event; Based on the emergency call event, obtaining the vehicle identity information, location information, current running state information and type of the emergency call event of the vehicle, and sending them to a target platform to request to establish a communication connection, wherein the type of the emergency call event at least includes accident triggering or abnormal vital sign state; After the communication connection is established, the in-vehicle microphone and in-vehicle loudspeaker are controlled to enter a call state.

4. The method of claim 1, wherein, The method further comprises the following steps: If the determination result is that the vehicle is in the accident triggering state, the breathing frequency data is input into the trained neural network model while the emergency call process is triggered, so that the neural network model outputs the vital sign state of the target object; During the execution of the emergency call process, the vital sign state is sent to the target platform.

5. The method of claim 1, wherein, The method further comprises the following steps: When there are multiple target objects in the vehicle and the vehicle is not in the accident triggering state, if the vital sign state of any target object is detected to be abnormal, the identity information of the abnormal target object is identified; If the identity information is a passenger, an alarm event is triggered in the vehicle.

6. The method of claim 4, wherein, After the neural network model outputs the vital sign state of the target object, the method further comprises the following steps: When it is determined that the vital sign state is abnormal and the vehicle is in the accident triggering state, the current accident triggered by the vehicle is classified according to the body acceleration data and the wheel speed data to obtain an accident level, and the health risk of the target object is classified according to the vital sign state to obtain a health risk level; According to the accident level and the health risk level, the response level of the vehicle at present is determined; According to the control strategy corresponding to the response level, an emergency call process of the vehicle is triggered.

7. The method of claim 1, wherein, The method further includes: Based on a preset time window, time series analysis is performed on the vehicle body acceleration data, the wheel speed data and the breathing frequency data to generate a vehicle operation stability change trend and a vital sign state change trend; The vehicle operation stability change trend and the vital sign state change trend are input into the trained evaluation model to cause the evaluation model to output a risk prediction result of the vehicle, the risk prediction result representing a first probability of the vehicle entering an accident triggering state in a future time period and / or a second probability of a vital sign state being abnormal; In a case where the first probability is greater than a first threshold or the second probability is greater than a second threshold, a vehicle risk warning process is triggered.

8. A vehicle emergency call apparatus characterized by comprising: The method includes: The acquisition module is configured to, in a case where the vehicle is powered on, acquire breathing frequency data of a target object in real time, and acquire vehicle body acceleration data and wheel speed data of the vehicle; The judgment module is configured to determine, according to the vehicle body acceleration data and the wheel speed data, whether the vehicle is in an accident triggering state; The processing module is configured to, in a case where the judgment result is that the vehicle is not in the accident triggering state, input the breathing frequency data into a trained neural network model to cause the neural network model to output a vital sign state of the target object; The control module is configured to, in a case where it is determined that the vital sign state is abnormal or the judgment result is that the vehicle is in the accident triggering state, trigger an emergency call process.

9. A vehicle characterized by comprising: The processor is configured to execute a vehicle emergency call program stored in the memory to implement the vehicle emergency call method in any one of claims 1-7. The storage medium stores one or more programs, and the one or more programs are executable by one or more processors to implement the vehicle emergency call method in any one of claims 1-7.

10. A storage medium, characterized by ​

Images