Vehicle takeover method and device and electronic equipment

Abstract

The invention discloses a vehicle takeover method and device and electronic equipment. The method comprises the steps that the driving state of a target object in a target vehicle is acquired, and the driving state is used for indicating the physiological state and the behavior state of the target object; the target object is evaluated based on the physiological state and the behavior state, an evaluation result is obtained, and the evaluation result is used for quantitatively representing the control ability of the target object to the target vehicle; and under the condition that the evaluation result indicates that the target vehicle meets the takeover triggering condition, the target vehicle is controlled to enter an emergency takeover mode. The technical problems that emergency takeover of an intelligent driving assistance system adopted in the related technology is passively triggered by complex manual operation of a driver, so that the intelligent driving emergency takeover function is difficult to start quickly, and the optimal handling opportunity is delayed are solved.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and more specifically, to a method, apparatus, and electronic device for vehicle takeover. Background Technology

[0002] Current intelligent driving assistance systems in new energy vehicles generally possess basic functions such as lane keeping, adaptive cruise control, and emergency braking. Some high-end models also feature driver status monitoring systems (DMS), which use visual cameras to identify obvious behavioral characteristics such as driver fatigue and distraction, and provide alerts with audible and visual signals. However, the emergency avoidance technologies employed by these systems are mostly passively triggered. The operational path for drivers to actively trigger intelligent driving emergency takeover is relatively cumbersome, making it difficult for drivers to quickly activate the intelligent driving emergency takeover function in emergency situations, thus delaying the best opportunity for response.

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

[0004] This application provides a method, apparatus, and electronic device for vehicle takeover, which at least solves the technical problem that the emergency takeover of the intelligent driving assistance system used in related technologies is passively triggered by the driver's complex manual operation, making it difficult to quickly activate the intelligent driving emergency takeover function and delaying the best handling time.

[0005] According to one aspect of the embodiments of this application, a method for vehicle takeover is provided, comprising: acquiring the driving state of a target object in a target vehicle, wherein the driving state is used to indicate the physiological state and behavioral state of the target object; evaluating the target object based on the physiological state and behavioral state to obtain an evaluation result, wherein the evaluation result is used to quantitatively represent the target object's control capability over the target vehicle; and controlling the target vehicle to enter an emergency takeover mode when the evaluation result indicates that the target vehicle meets the takeover triggering conditions.

[0006] In this embodiment, a multi-dimensional driving state information fusion approach is adopted. By collecting the physiological and behavioral states of the target object and comprehensively evaluating its control ability over the vehicle, the vehicle emergency takeover mode is automatically triggered when the evaluation results meet the preset takeover conditions. This achieves the goal of timely identification of abnormal driver states and rapid intervention in vehicle control, thereby realizing the technical effect of instant takeover of the intelligent driving system without relying on the driver's active operation. This solves the technical problem that the emergency takeover of the intelligent driving assistance system used in related technologies is passively triggered by the driver's complex manual operation, which makes it difficult to quickly activate the intelligent driving emergency takeover function and delays the best handling time. Attached Figure Description

[0007] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0008] Figure 1 This is a hardware structure block diagram of a computer terminal for a vehicle takeover method according to an embodiment of this application;

[0009] Figure 2 This is a flowchart of a vehicle takeover method according to an embodiment of this application;

[0010] Figure 3 This is an overall flowchart of a vehicle takeover method according to an embodiment of this application;

[0011] Figure 4 This is a flowchart of a vehicle takeover method according to an embodiment of this application;

[0012] Figure 5 This is an emergency parking flowchart of a vehicle takeover method according to an embodiment of this application;

[0013] Figure 6 This is a flowchart illustrating the early warning and rescue process of a vehicle takeover method according to an embodiment of this application;

[0014] Figure 7 This is a schematic diagram of a vehicle takeover device according to an embodiment of this application. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0016] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] To better understand the embodiments of this application, the technical terms involved in the embodiments of this application are explained below:

[0018] Emergency Take-over Mode: This refers to an automated driving mode in which the intelligent driving system automatically takes over steering, braking, and power control when the vehicle detects that the driver has lost or is about to lose control, and performs safe stopping and warning operations. In this embodiment, the emergency take-over mode is activated when the evaluation result indicates that the driver is in a state of "severe discomfort" or "loss of operational ability." It is used to complete a full range of autonomous operations, including smooth deceleration, lane changing, pulling over to the side of the road, activating hazard lights, and V2X warnings, ensuring that the vehicle stops safely without driver intervention. This achieves proactive response to sudden incapacitation scenarios and avoids traffic accidents caused by delayed human response.

[0019] Physiological state refers to the intrinsic physiological characteristics of a target object during driving, as reflected by biological signs, used to characterize whether its bodily functions are within a normal or abnormal range. In the embodiments of this application, physiological state can be quantified, for example, by continuous physiological parameters such as heart rate, blood pressure, and respiratory rate collected by seat physiological sensors. This serves as an objective basis for determining whether the driver has a sudden illness (such as myocardial infarction or cerebral infarction) or severe fatigue, and is used in conjunction with behavioral state to comprehensively assess its handling ability, thereby improving the early identification of hidden risks.

[0020] Behavioral State: This refers to the operational characteristics exhibited by a target object during driving through external actions and interactions, reflecting its level of attention and control intent towards the vehicle. In this embodiment, behavioral state can be characterized by non-physiological indicators such as eyelid closure and head posture obtained by an infrared camera, grip force magnitude and continuous changes obtained by a steering wheel grip force sensor, and abnormal voice responses detected by a voice recognition module. These indicators are used to determine whether the driver exhibits obvious abnormalities such as distraction, unconsciousness, or operational failure, complementing the physiological state and enhancing the comprehensiveness and accuracy of state assessment.

[0021] Take-over Trigger Condition: This refers to the set of logical criteria upon which the system determines whether to activate the emergency takeover mode. It consists of a combination of quantifiable and verifiable state parameters. In this embodiment, the takeover trigger condition includes two types: driver-triggered (e.g., one-button press or voice command) and automatic-triggered (e.g., abnormal physiological and behavioral states with no operation for 5 seconds). This ensures that the system can respond to the driver's active intentions and automatically initiate takeover based on multi-dimensional state fusion judgment in the absence of response. This overcomes the passive response delay problem of traditional manual operation while ensuring system reliability.

[0022] Current intelligent driving assistance systems (such as L2 and L3 levels) for new energy vehicles have basic lane keeping, adaptive cruise control, and emergency braking functions. Some high-end models are equipped with driver status monitoring systems (DMS), which can identify driver fatigue, distraction, and other states through cameras and provide reminders through sound and light. At the same time, existing emergency avoidance technologies are mostly "passively triggered," meaning that limited emergency stopping functions are only activated when the driver is detected to have completely lost the ability to operate the vehicle (such as not taking over the steering wheel for a long time, or the vehicle is about to deviate from the lane without any corrective action). After stopping, the vehicle can only send location information to a preset contact through the in-vehicle terminal. Some models support connection to roadside assistance platforms, but they lack proactive, tiered takeover warnings and full-process emergency response capabilities.

[0023] Furthermore, in related technologies, the operation path for drivers to actively trigger intelligent driving emergency takeover is relatively cumbersome (requiring multiple levels of menu operations or specific key combinations), and after active takeover, only basic parking can be achieved, without being able to optimize parking strategies according to road conditions (such as highways, urban main roads, and non-motorized vehicle lanes). At the same time, existing systems have problems such as unreasonable parking locations (such as parking in the middle of the fast lane), incomplete alarm information (only containing the location, without driver health status or vehicle fault information), and untimely rescue coordination.

[0024] Therefore, it can be seen that the relevant technologies mainly have the following problems:

[0025] (1) Low efficiency of driver takeover: The operation process is complicated, and it is difficult for the driver to quickly activate the intelligent driving emergency takeover function in an emergency, thus delaying the best time to deal with the situation.

[0026] (2) Passive takeover triggering conditions are singular: they rely heavily on obvious features such as driver operation failure or vehicle deviation from lane, and have low accuracy and delayed response in recognizing hidden discomfort conditions such as sudden illness of the driver (such as myocardial infarction or cerebral infarction).

[0027] (3) Emergency parking strategies lack scenario adaptability: the optimal parking location is not selected based on road type, traffic flow, and surrounding facilities (such as emergency lanes and bus stops), which can easily lead to secondary traffic accidents;

[0028] (4) The early warning and rescue link is not closed: after parking, only basic information can be pushed, and it is not possible to actively connect with the traffic management platform and emergency center. Furthermore, there is a lack of proactive early warning prompts for surrounding vehicles, which poses a safety hazard.

[0029] (5) Insufficient system fault tolerance: a single sensor failure can easily lead to the failure of the takeover function. There is no multi-sensor redundancy verification mechanism, and the reliability needs to be improved.

[0030] To address the aforementioned technical problems, this application provides corresponding solutions, which are detailed below.

[0031] The vehicle takeover method embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware block diagram of a computer terminal for implementing a vehicle takeover method is shown. Figure 1 As shown, the computer terminal 10 may include one or more processors (shown as 102a, 102b, ..., 102n in the figure) (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission module 106 for communication functions connected via wired and / or wireless networks. In addition, it may also include: a display, a keyboard, a cursor control device, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, and a BUS bus. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0032] It should be noted that the aforementioned one or more processors and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10. As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0033] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the vehicle takeover method in this embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the aforementioned vehicle takeover method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0034] The transmission module 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission module 106 includes a network interface controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission module 106 may be a radio frequency (RF) module, used for wireless communication with the Internet.

[0035] The display can be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10.

[0036] It should be noted here that, in some optional embodiments, the above... Figure 1 The computer terminal shown may include hardware elements (including circuitry), software elements (including computer code stored on a computer-readable medium), or a combination of both hardware and software elements. It should be noted that... Figure 1 This is only one instance of a specific particular instance, and is intended to illustrate the types of components that may exist in the aforementioned computer terminal.

[0037] In the above operating environment, this application provides a method embodiment for vehicle takeover. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0038] Figure 2 This is a flowchart of a vehicle takeover method according to an embodiment of this application, such as... Figure 2 As shown, the method includes the following steps:

[0039] Step S202: Obtain the driving status of the target object in the target vehicle, wherein the driving status is used to indicate the physiological and behavioral status of the target object.

[0040] In step S202 above, driving state refers to the combined representation of the physiological and behavioral states of the driver (i.e. the target object) during driving, rather than a simple monitoring of a single dimension. The physiological state is reflected in the biological parameters of the driver's body that can be objectively measured by sensors, such as increased heart rate, abnormal blood pressure, and disordered respiratory rate, reflecting whether the driver is experiencing a sudden illness or serious physical discomfort. The behavioral state is reflected in the external action characteristics that can be perceived during the interaction between the driver and the vehicle, such as prolonged eyelid closure time, abnormal head tilt, continuous decrease or disappearance of steering wheel grip strength, and lack of voice commands, reflecting that the driver's attention is distracted, consciousness is blurred, or the driver has lost control intention.

[0041] In some embodiments of this application, non-contact and contact data acquisition based on an in-vehicle multimodal sensor array can be employed. Specifically, this method continuously captures the driver's facial features using an infrared camera deployed in the cockpit, extracting behavioral parameters such as eyelid closure duration, head tilt angle, and facial muscle relaxation. Simultaneously, a grip force sensor integrated on the steering wheel measures the magnitude and trend of the force exerted by the driver's hands on the steering wheel in real time. Physiological sensors embedded in the seat continuously collect physiological parameters such as heart rate, blood pressure, and respiratory rate through bioelectrical sensing and pressure fluctuation analysis. For example, during vehicle operation, if the infrared camera detects that the driver's eyelids are closed for more than 3 seconds and the head is continuously tilted to one side, while the grip force sensor data drops below 10N and remains stable, the system determines that the behavioral state is abnormal. If the physiological sensor simultaneously detects that the heart rate exceeds the normal threshold, it confirms that the driving state has entered a dangerous zone.

[0042] Furthermore, anomaly command recognition and response loss monitoring based on a voice interaction module can also be employed. Specifically, this method continuously monitors the driver's voice content through an in-vehicle voice recognition system, establishing a baseline model of voice in normal driving contexts, such as routine conversations, navigation commands, and air conditioning adjustments. When the system detects a prolonged period without voice input, or identifies abnormal voice (such as groans, cries for help, or unclear shouts), it serves as supplementary evidence for judging the driver's behavioral state. For example, in the event of a sudden stroke, the driver may utter non-standard voices such as "Help!" or "I feel so dizzy." The system, through both semantic and voiceprint analysis, identifies the abnormal voice content and uses it as key evidence to assess the driver's state, forming a multi-dimensional chain of evidence together with physiological and behavioral data.

[0043] The intelligent driving assistance systems used in related technologies rely solely on single behavioral characteristics (such as releasing the steering wheel or lane departure) to determine the timing of takeover. Their ability to identify latent incapacity states, such as sudden driver illness, is weak, leading to delayed or even complete failure of takeover response, missing the optimal intervention window, and easily causing secondary accidents. This application's embodiment, by simultaneously collecting physiological and behavioral data, constructs a comprehensive assessment foundation from the inside out, enabling the system to identify early abnormalities 3 to 5 seconds in advance when the driver has not yet completely lost their ability to act, thereby achieving proactive and preemptive takeover intervention.

[0044] It should be noted that false anomalies in physiological parameters and behavioral characteristics under specific scenarios can lead to misjudgments by the assessment system, triggering unnecessary takeover operations, interfering with normal driving, and reducing user trust. For example, during long-distance driving, severe vibrations on bumpy roads can cause the seat's physiological sensors to falsely report abnormal heart rates; or normal actions such as making a phone call or reaching for an object may cause a temporary decrease in steering wheel grip or head posture shift, which could be mistaken for abnormal behavior. To address this issue, a time-series consistency verification mechanism for multimodal data can be introduced, combined with dynamic threshold adaptive adjustment logic, to achieve intelligent identification of abnormal signals.

[0045] Specifically, the system does not trigger a judgment simply when a single sample exceeds the threshold. Instead, it requires that within a continuous time window of more than 5 seconds, both physiological abnormalities (such as a heart rate consistently above 110 beats per minute) and behavioral abnormalities (such as grip strength consistently below 15N with no recovery trend) coexist and show a consistent trend before a true abnormality is identified. Simultaneously, the system dynamically adjusts the threshold based on the driving environment. For example, it relaxes the tolerance for brief fluctuations in grip strength at high speeds and increases the sensitivity to behavioral lags in congested traffic. Furthermore, the system considers the vehicle's operational status (such as whether it is changing lanes or experiencing emergency braking) during the judgment process to avoid misjudging normal driving actions as incapacity. This mechanism effectively filters out environmental interference and occasional disturbances, upgrading the driving status assessment from a single-point trigger to trend confirmation, significantly improving the system's accuracy and usability in real-world, complex scenarios.

[0046] Furthermore, when the system detects a change in the driving environment of the target vehicle, it will automatically identify multi-dimensional environmental factors such as current road type, traffic density, speed limit level, weather conditions, and light intensity based on high-precision maps and real-time environmental perception data. Combined with a preset environment-sensitivity mapping model, it will dynamically adjust the evaluation thresholds for physiological and behavioral states.

[0047] (1) The system uses high-precision maps and Beidou / GPS dual-mode positioning information to spatially match and accurately identify the road type where the vehicle is currently located (such as highways, urban expressways, tunnels, bridges, urban main roads, etc.). It also simultaneously obtains the legal speed limit, number of lanes and surrounding facility characteristics of the road section, providing a structural basis for subsequent environmental risk level determination.

[0048] (2) Based on the road attributes determined in the first step, the system initiates an environmental perception strategy that matches the road attributes. It calls on lidar, millimeter-wave radar and visual camera to collect the distribution density of traffic participants within 50 meters around the vehicle. It also combines the results of meteorological sensors or visual recognition to determine whether there are interference factors such as rain, snow, fog, and strong light. This generates a dynamic environmental risk profile that is highly correlated with the current road type, ensuring that the perception dimension and scene characteristics are accurately matched.

[0049] (3) Based on the environmental risk profile generated in the second step, the system automatically loads the combination of physiological and behavioral assessment thresholds corresponding to the road type and environmental conditions from the preset multidimensional environment-sensitivity mapping table. For example, it increases the sensitivity to eyelid closure time in high-speed and low-density environments, disables visual feature extraction and strengthens grip strength and physiological signal weights in low-light tunnel environments, and shortens the physiological abnormality judgment time window in rainy and foggy weather, so as to achieve strong coupling and strong correlation between threshold adjustment and environmental risk.

[0050] (4) The system injects the threshold parameters dynamically loaded in the third step into the driver state fusion judgment module in real time as the only benchmark for evaluating the driver's control ability in the current driving environment. This ensures that the comprehensive judgment logic of physiological state and behavioral state always keeps in sync with the risk level of the road ahead, traffic density and meteorological interference intensity, thereby constructing a closed loop of "environmental recognition - perception adaptation - threshold mapping - state assessment". This effectively eliminates the risk of misjudgment and omission of fixed thresholds in complex scenarios and comprehensively improves the robustness of judgment and reliability of decision-making under changing working conditions.

[0051] Step S204: Evaluate the target object based on its physiological and behavioral states to obtain evaluation results, wherein the evaluation results are used to quantitatively represent the target object's control capability over the target vehicle.

[0052] In step S204 above, the evaluation result refers to the comprehensive level value calculated by the system based on multi-source data of physiological and behavioral states through a fusion algorithm, which is used to quantitatively represent the target object's control capability over the target vehicle. This result is not a simple summation of a single indicator, but rather a multi-level state evaluation conclusion (such as "normal / mild discomfort / severe discomfort / loss of operational ability") output after joint analysis of static features and temporal dynamic features by a deep learning model (such as CNN+LSTM).

[0053] In some embodiments of this application, a time-series feature fusion evaluation mechanism based on multimodal sensor data can be employed. Specifically, the system first performs time alignment on heterogeneous data such as facial features from infrared cameras (e.g., eyelid closure time, head posture changes), grip force time-series curves from steering wheel grip force sensors, heart rate and blood pressure fluctuation curves from seat physiological sensors, and response missing records from the speech recognition module. Subsequently, a convolutional neural network is used to extract spatial static features (e.g., facial expression morphology) from each sensor data, and a long short-term memory network is used to model their behavioral patterns over time (e.g., a continuous decreasing trend in grip force, an increasing frequency of eyelid closure). Finally, static and dynamic features are integrated through a feature-level fusion layer, input into a classifier, and the output evaluation level is determined.

[0054] Furthermore, based on feature fusion, the system can also load corresponding evaluation weight factors from a pre-set environment-sensitivity mapping table, taking into account current driving environment parameters (such as high-speed low-density, tunnel low-light, rain and snow weather). For example, in a highway environment, the system assigns a higher weight to "eyelid closure time" because the risk of loss of attention is highest in this scenario; in tunnels or low-light conditions at night, the system automatically reduces the weight of visual features and instead strengthens the contribution of grip strength and physiological parameters; when the road surface is slippery due to rain or snow, the system shortens the judgment time window for abnormal physiological parameters to cope with the accelerated physiological reaction caused by driver tension. Finally, the system outputs the final evaluation result after weighted fusion, ensuring that the judgment logic always matches the current risk level and achieving personalized evaluation driven by environment perception.

[0055] In some embodiments of this application, the target object can be evaluated based on physiological and behavioral states to obtain evaluation results by: collecting physiological parameters of the target object, wherein the physiological parameters are used to reflect the physiological state of the target object; comparing the physiological parameters with a preset threshold to obtain a comparison result; when the comparison result indicates that the physiological parameters are greater than the preset threshold, collecting behavioral characteristics of the target object, wherein the behavioral characteristics are used to characterize the behavioral state of the target object in the process of driving the target vehicle in multiple dimensions; and determining the evaluation result based on the comparison result and the behavioral characteristics.

[0056] It should be noted that behavioral characteristics are used to characterize the driver's level of attention, the persistence of operational intentions, and the integrity of control ability in multiple dimensions (visual, tactile, and auditory), such as the duration of eyelid closure, head posture deviation angle, steering wheel grip force change trend, and duration of missing voice response.

[0057] Specifically, the physiological parameters of the target can be collected by a multi-channel physiological sensor array integrated into the driver's seat. The sensors are based on bioimpedance spectroscopy, photoplethysmography, and pressure sensing technology, and continuously sample the driver's heart rate, blood pressure, and respiratory rate at a sampling frequency of no less than 10 Hz to ensure the capture of transient characteristics of sudden physiological fluctuations.

[0058] Furthermore, the system adaptively adjusts the judgment criteria for physiological parameters based on the current driving environment (such as highways, tunnels, rain, and fog). For example, in a low-density free-flow environment on a highway, the system sets the "heart rate threshold" to 110 beats per minute, while in a rainy and slippery road environment, due to increased psychological stress, the threshold is automatically lowered to 95 beats per minute. The comparison result only outputs a binary judgment of "whether it exceeds the standard". If it does not exceed the standard, the evaluation process terminates, and the system maintains normal driving status; if it exceeds the standard, it enters the next level of behavioral feature collection process, realizing a low-power, high-efficiency hierarchical judgment strategy.

[0059] Furthermore, when physiological parameters exceed limits, the system activates the infrared camera, steering wheel grip force sensor, and voice recognition module to simultaneously collect behavioral characteristics such as eyelid closure duration, head posture angle, grip force change rate, and voice response absence time, forming a multi-dimensional behavioral profile. For example, if physiological parameters exceed limits, but the driver continues to lightly grip the steering wheel, blinks frequently, and responds normally to voice prompts, the behavioral characteristics are judged as "mild discomfort"; if the three characteristics of continuous eyelid closure, zero grip force, and no voice response are superimposed, the behavioral characteristics are confirmed as "severe disability".

[0060] Furthermore, the physiological comparison results (exceeding limits / not exceeding limits) and multi-channel data of behavioral characteristics are input into a deep learning network (CNN+LSTM) for joint spatiotemporal feature analysis. For example, the output is a four-level evaluation conclusion: if only physiological limits are exceeded but behavior is normal, it is judged as "mild discomfort"; if physiological limits are exceeded plus a single behavioral abnormality (such as eyelid closure), it is judged as "severe discomfort"; if physiological limits are exceeded plus multiple behavioral abnormalities (grip strength zero + eyelid closure + no voice response), and lasts for more than 5 seconds, it is judged as "loss of operational ability", triggering the highest priority takeover process.

[0061] Existing technologies rely solely on single behavioral characteristics (such as releasing the steering wheel or lane departure) to determine the timing of takeover, failing to identify latent incapacity caused by sudden physiological illnesses in the driver. This results in a critical technological bottleneck: severely delayed or completely ineffective emergency response. In contrast, this application's embodiment uses physiological parameters as an active trigger source, constructing a dual verification system from the inside out. This allows the system to identify and intervene before the driver exhibits obvious operational abnormalities, securing a critical time window for subsequent takeover, stopping, and rescue, significantly reducing the risk of traffic accidents caused by driver incapacity.

[0062] Step S206: If the assessment results indicate that the target vehicle meets the takeover trigger conditions, control the target vehicle to enter the emergency takeover mode.

[0063] In step S206 above, the takeover trigger condition is a set of logical criteria for driving the vehicle to switch from manual driving mode to emergency intelligent driving mode. Emergency takeover mode refers to an intelligent driving operation state with L4 level control authority that is automatically activated by the system when the driver loses effective control ability. In this mode, the vehicle's steering, braking, power output, lighting system and communication module are all taken over by the on-board decision system, which executes path planning, smooth deceleration, precise stopping and multi-dimensional warnings according to preset safety strategies until rescue arrives or the driver regains control. Its function is to enable the vehicle to autonomously avoid risks in extreme situations where the human driver fails, and avoid collisions or secondary accidents caused by loss of control.

[0064] In some embodiments of this application, the takeover triggering conditions include: a first takeover triggering condition passively triggered based on an operation command of the target object, wherein the operation command includes a hardware trigger command in the target vehicle or a voice command of the target object; and a second takeover triggering condition actively triggered based on a preset driving state of the target object, wherein the preset driving state includes a state in which the evaluation result meets a preset level.

[0065] It should be noted that the first takeover trigger condition refers to the takeover mechanism triggered by a human intervention command issued by the driver, clearly expressing the intention to take over. This can originate from physical hardware buttons installed in the target vehicle (such as a dedicated one-button takeover button on the left side of the steering wheel) or preset voice commands issued by the driver through the voice system (such as "emergency takeover"). The core function of this condition is to grant the driver the highest priority of active control, especially when the driver feels unwell but is still conscious. This allows bypassing the system assessment process and directly initiating an emergency response, avoiding missing the golden opportunity for intervention due to system delays.

[0066] The second takeover trigger condition refers to the takeover mechanism triggered by the system's automatic judgment of the driver's preset driving state. Its core basis is whether the assessment result output in step S204 reaches a preset level (such as "severe discomfort" or "loss of operational ability"). This condition does not rely on the driver's subjective behavior, but rather, through the fusion of multi-dimensional physiological parameters and behavioral characteristics, the system autonomously identifies whether the driver has lost effective control ability. Its function is to compensate for the lack of human subjective judgment and achieve a zero-delay response to sudden latent illnesses (such as myocardial infarction or stroke) or sudden loss of consciousness.

[0067] In some embodiments of this application, the system presets "severe discomfort" and "loss of operational ability" as trigger level thresholds. When the fusion judgment module outputs the result of this level, the system automatically initiates the takeover process provided that its confidence level is higher than 95%. To prevent misjudgment, the system may also require that the state must last for ≥5 seconds without any signs of recovery, and perform secondary verification in combination with environmental factors (such as the vehicle being in a high-speed driving state, without emergency braking or obstacle avoidance actions).

[0068] To address the technical shortcomings of existing intelligent driver assistance systems, such as slow response and limited intervention, the following steps can be performed: Upon receiving an operation command from the target vehicle, control the target vehicle to enter emergency takeover mode; or, if the assessment result indicates that the target vehicle's driving status is at Level 1, send a prompt message to the target vehicle, where Level 1 indicates that the target vehicle has full control over the target vehicle, and the prompt message is used to remind the target vehicle to rest; or, if the assessment result indicates that the target vehicle's driving status is at Level 2, send a takeover warning message to the target vehicle, and control the target vehicle to enter emergency takeover mode when the target vehicle's inactivity time meets a preset duration, where Level 2 indicates that the target vehicle has lost some control over the target vehicle; or, if the assessment result indicates that the target vehicle's driving status is at Level 3, control the target vehicle to enter emergency takeover mode, where Level 3 indicates that the target vehicle has completely lost control over the target vehicle.

[0069] Specifically, upon receiving an operational command from the target, the system controls the target vehicle to enter emergency takeover mode, and the actions to be executed based on the assessment results include:

[0070] (1) Level 1 refers to the driver’s physiological state being abnormal but his behavior being within the normal range, and he having the ability to fully control the vehicle. Although there is no direct control action at this level, the system will still actively remind the driver to rest through prompts. Its role is to achieve preventive intervention, resolve risks at the bud stage, and avoid fatigue accumulation leading to a deterioration of the condition.

[0071] (2) Level 2 refers to a situation where the driver exhibits obvious physiological or behavioral abnormalities but has not completely lost control. The system determines this as a partial loss of control, such as a slight increase in heart rate, persistent head tilt, and decreased grip strength but still with intermittent corrective actions. The core function of this level is to establish a warning buffer zone, giving the driver a final opportunity to correct their behavior by issuing a takeover warning in advance, thus avoiding the system taking over abruptly and causing fright or misjudgment. For example, if the driver's hands tremble due to hypoglycemia but they can still grip the steering wheel briefly, the system will determine this as Level 2 and issue a voice warning: "Your abnormal condition has been detected. Assisted takeover will be initiated soon. Please resume operation as soon as possible." If there is no response within 10 seconds (i.e., the preset duration), takeover will be automatically triggered.

[0072] Furthermore, the preset response time can be dynamically determined based on the vehicle's environmental conditions, enabling a faster and more precise response in emergency situations. Specifically, environmental parameters such as road type, vehicle speed, traffic density, weather conditions, visibility, lane width, and emergency lane availability are collected. After standardization, these parameters are input into a preset multi-dimensional environmental risk assessment function, which outputs a four-level environmental risk level (extremely high, high, medium, and low) based on a nonlinear weighted model. When a takeover warning is triggered in the second-level driving state, the system dynamically adjusts the preset response time according to the current environmental risk level: for example, in extremely high-risk environments, the time is compressed to within 3 seconds and the multimodal warning intensity is enhanced to force driver intervention; in high-risk environments, the time is shortened to 5 seconds to balance response efficiency and operational error tolerance; and in medium- and low-risk environments, the time is extended to 8–12 seconds to prioritize the driver's autonomous control.

[0073] (3) Level 3 refers to the driver having completely lost the ability to operate the vehicle. The system determines that the driver has completely lost control. Its characteristics include severely abnormal physiological parameters (such as heart rate > 120 beats / minute), and behavioral characteristics that remain at zero (eyelid closure for more than 5 seconds, grip strength consistently below 5N, and no voice response), with no signs of recovery. This level is the only final trigger condition for the system to initiate mandatory takeover. Its function is to achieve zero-delay safety intervention in an unconscious state. For example, if the driver suddenly suffers a cerebral hemorrhage and falls into a coma, the system will quickly complete the takeover after confirming Level 3, automatically decelerating, changing lanes, and stopping, thus buying golden time for subsequent emergency treatment.

[0074] In some embodiments of this application, the target vehicle can be controlled to enter the emergency takeover mode by: obtaining the driving information of the target vehicle, wherein the driving information includes the target vehicle status information, the environmental information of the road where the target vehicle is located, and the information of traffic participants; determining the parking area based on the driving information; determining the parking path of the target vehicle to the parking area, and performing an emergency parking operation according to the parking path.

[0075] It should be noted that driving information refers to the comprehensive set of environmental and vehicle status perception data that the system relies on during emergency takeover, including: the target vehicle's own status information (such as vehicle speed, steering angle, braking pressure, power system operating status, and electronic parking brake status), the environmental information of the road where the target vehicle is located (such as road type, number of lanes, presence of emergency lanes, shoulder width, and distribution of surrounding facilities), and information on traffic participants (such as the position, speed, and driving trajectory of vehicles in front, behind, and to the side, as well as the distribution of pedestrians, non-motorized vehicles, and static obstacles).

[0076] Specifically, vehicle status information can be read in real time by the vehicle's CAN bus, road environment information is provided by a high-precision map module and a Beidou / GPS dual-mode positioning system, combined with lane-level positioning (accuracy ±0.5m) to identify the current road structure, emergency lane location and surrounding facilities, and traffic participant information is perceived collaboratively by lidar, millimeter-wave radar and visual cameras, and the dynamic position, speed and movement trend of surrounding vehicles and obstacles are continuously output through target tracking algorithms.

[0077] Furthermore, parking areas can be determined based on driving information in the following ways: at least one candidate parking area is determined based on driving information that is less than a preset distance from the target vehicle; obstacles in the candidate parking areas are eliminated to obtain at least one target candidate area; a safety factor is determined for each target candidate area, wherein the safety factor is used to quantify the collision risk level caused by the density of traffic participants around the target vehicle; and the target candidate area with the highest safety factor is determined as the parking area.

[0078] Specifically, the system first reads the vehicle's real-time absolute position (using BeiDou / GPS dual-mode positioning with an accuracy of ±0.5 meters) and then calls the built-in road structure topology database. This database pre-stores the geographic coordinates and geometric boundaries of emergency lanes, shoulders, service areas, bus stops, and other parking areas on major roads across the country. The system then retrieves all area points within 500 meters (i.e., the preset distance) that meet the preset type (such as "emergency parking lane" or "non-driving area") and generates an initial candidate set. This process is a fast query based on spatial indexing, relying on the high-precision updates of map data and the high stability of vehicle positioning to ensure that no candidate areas are missed or falsely detected.

[0079] After obtaining candidate parking areas, the system constructs real-time 3D point clouds and semantic segmentation maps of the vehicle's surroundings using LiDAR and visual cameras, identifying all static and dynamic obstacles located within or bounding the candidate areas. The system performs virtual occupancy detection on each candidate area: specifically, if unauthorized objects (such as other parked vehicles, pedestrians, cones, or construction equipment) are detected within the area, the area is deemed unusable; if only low vegetation, drainage ditches, or road markings that do not affect parking exist, the area is retained. This process can, for example, employ a dynamic masking algorithm to map obstacle entities onto a 2D projection plane of the parking area, calculate their occupancy percentage, and if it exceeds a threshold (e.g., 20%), the area is removed, ultimately outputting a list of unblocked target candidate areas.

[0080] After obtaining the target candidate areas, the surrounding traffic flow characteristics of each target candidate area are extracted, including: the number of vehicles within 100 meters behind, average speed, relative approach rate, vehicle density in the side lanes, and whether it is in the influence zone of a curve or slope. These parameters are normalized into multiple risk factors, and a safety factor is calculated through a weighted linear comprehensive model. For example, the closer, faster, and more numerous the vehicles behind, the higher the weight of the risk factor and the lower the safety factor. If the candidate area is located at the end of a curve or on an unlit road section at night, the system automatically adds an environmental correction coefficient. Finally, each target area outputs a continuous safety value of 0–1, which directly reflects "the probability level of being rear-ended or side-impacted by a vehicle if parked here".

[0081] Finally, the safety coefficients of all candidate areas are ranked, and the one with the highest score is selected as the final parking target.

[0082] In some embodiments of this application, during the process of determining the parking area based on driving information, the parameters for screening and evaluating the parking area can be dynamically adjusted by the driver's state assessment level (i.e., assessment result). This enables an intelligent grading strategy under different degrees of abnormal driving conditions, ensuring that the system can complete a safe stop at the fastest speed and with the highest priority when the driver is in severe discomfort or even completely loses the ability to operate.

[0083] For example, when a driver is in a state of "mild discomfort," the system determines that they still have autonomous control capabilities. In this case, the system prioritizes the driving experience and operational autonomy, and does not rush to forcibly intervene. Therefore, in the process of determining parking areas, the system adopts a lenient screening strategy: the preset candidate area search distance is expanded to 800 meters, allowing the selection of parking spots that are farther away but have a better environment (such as service areas and parking lot entrances); the obstacle removal criteria are appropriately relaxed, tolerating minor static interference (such as non-motorized vehicles parked on the roadside and low road signs); when calculating the safety factor, the weight of rear traffic density and lateral interference is reduced, allowing areas with a safety factor below 0.75 to enter the candidate pool.

[0084] Correspondingly, when the driver is in a state of "severe discomfort," the system has confirmed that their physiological indicators are significantly abnormal (such as heart rate > 120 beats / min, blood pressure exceeding the standard, eyelid closure lasting more than 3 seconds, and steering wheel grip strength almost zero), and combined with behavioral characteristics, judges that they are about to completely lose control. At this time, the system immediately activates the emergency stop enhancement mode, systematically tightening and accelerating all parking area screening parameters, including but not limited to:

[0085] The candidate area search distance is compressed to within 300 meters, focusing only on the nearest, quickly accessible emergency lane or shoulder, rejecting any distant areas that require long-distance driving or complex lane changes;

[0086] The safety threshold has been raised, requiring the target area to have a safety factor of ≥0.85. Areas with a safety factor below this value are excluded even if they are closer or have a shorter path, in order to avoid choosing high-risk areas for the sake of "closeness".

[0087] Introducing a "time-distance" penalty factor: The system calculates the arrival time for each candidate area (based on the current vehicle speed and path length). If an area has a slightly higher safety factor but the estimated arrival time exceeds 8 seconds, 0.15 safety weight is automatically deducted to ensure safety priority, but timeliness is still required.

[0088] It should be noted that the specific values ​​mentioned above are for illustrative purposes only and can be modified according to requirements.

[0089] Furthermore, using the optimal parking area as the endpoint, and combining the current vehicle speed, lane markings, and the trajectory of vehicles approaching from behind, a continuous trajectory is generated, comprising five sub-stages: "deceleration—turn signal activation—lane change—further deceleration—precise stopping." Each step ensures lateral displacement ≤0.3 m / s and longitudinal deceleration controlled within the 1.2–2.0 m / s² range. Before each lane change, a millimeter-wave radar confirms a safe distance of ≥50 meters behind the vehicle (specific values ​​are for illustrative purposes only). Subsequently, the planned trajectory is translated into control commands, precisely executed through steer-by-wire, brake-by-wire, and power adjustment modules. Simultaneously, hazard warning lights, parking lights, and V2X broadcast emergency stopping information are triggered, ensuring high predictability during vehicle execution and guaranteeing safe interaction with other road users.

[0090] In some embodiments of this application, the following steps may also be performed: generating warning information, wherein the warning information includes at least one of the following: a first warning information for reminding that an object inside the target vehicle has taken over the vehicle, a second warning information for reminding that an object around the target vehicle has taken over the vehicle, and a third warning information for reminding that a platform connected to the target vehicle has taken over the vehicle; and issuing a warning based on the warning information.

[0091] Specifically, the first warning message refers to the prompt message issued by the system to the occupants (including the driver and passengers) inside the target vehicle after the emergency takeover mode is activated. This message informs them that the control of the vehicle has been switched from manual operation to automatic system takeover. The message is presented through in-vehicle voice broadcast, flashing dashboard warning lights, steering wheel vibration, or slight seat vibration. Its core function is to reduce the panic caused by the sudden takeover, clearly inform the occupants of the current status, guide them to remain calm, fasten their seat belts, and avoid misoperation. It is a psychological buffer mechanism for achieving a stable transition in human-machine collaboration.

[0092] The second warning information refers to the active warning signal issued by the system to other traffic participants around the target vehicle (such as vehicles behind, vehicles in adjacent lanes, non-motorized vehicles, and pedestrians) during the emergency stop of the vehicle. This is to warn them that the vehicle in front is entering an abnormal operating state. This information is achieved through the vehicle's external lighting system (hazard lights and side marker lights), short horn beeps, and external communication devices (such as V2X broadcast). Its function is to alert the surrounding traffic participants in advance, leave reaction time for avoidance behavior, and effectively prevent secondary accidents such as rear-end collisions or side collisions. It is a key link in achieving vehicle-road cooperative safety.

[0093] The third early warning information refers to the structured notification information pushed by the system after completing an emergency stop to external platforms that have established communication connections with the vehicle (such as cloud collaboration platforms, traffic management command centers, emergency medical systems, vehicle brand after-sales service systems, and driver's preset emergency contacts). This information contains key data such as vehicle location, driver status, vehicle model information, and power system status. Its role is to establish a closed-loop rescue link between "vehicle-cloud-person-platform", realize automated emergency response, and significantly improve the speed of rescue response and the accuracy of resource allocation.

[0094] Through steps S202 to S206 above, a method of integrating multi-dimensional driving state information is adopted. By collecting the physiological and behavioral states of the target object and comprehensively evaluating its control ability over the vehicle, the vehicle emergency takeover mode is automatically triggered when the evaluation results meet the preset takeover conditions. This achieves the purpose of timely identification of abnormal driver states and rapid intervention in vehicle control, thereby realizing the technical effect of instant takeover of the intelligent driving system without relying on the driver's active operation. This solves the technical problem that the emergency takeover of the intelligent driving assistance system used in related technologies is passively triggered by the driver's complex manual operation, which makes it difficult to quickly activate the intelligent driving emergency takeover function and delays the best handling time.

[0095] This application also provides a vehicle takeover system, including a perception layer, a decision-making layer, an execution layer, and a linkage layer, wherein:

[0096] (1) Perception layer.

[0097] The perception layer is the system's data input source, responsible for collecting driver status information, vehicle status information, road environment information, and traffic participant information. It employs a multi-sensor redundancy design, specifically including:

[0098] Driver state perception module: equipped with an infrared camera (to collect facial expressions, eyelid closure, and head posture), a steering wheel grip force sensor (to collect grip force magnitude and changes), a seat physiological sensor (to collect physiological parameters such as heart rate, blood pressure, and respiratory rate), and a voice recognition module (to recognize abnormal voice commands or groans from the driver).

[0099] Vehicle status perception module: collects vehicle speed, steering angle, braking status, power system status, tire pressure, and intelligent driving system hardware status (camera, radar, high-precision map positioning accuracy).

[0100] Environmental perception module: equipped with lidar, millimeter-wave radar, visual camera (collecting information on surrounding vehicles, pedestrians, and obstacles), high-precision map module (providing information on road type, emergency lane location, and surrounding facilities), Beidou / GPS dual-mode positioning (positioning accuracy ±0.5m), and V2X communication module (receiving traffic flow and traffic light status information pushed by roadside equipment).

[0101] (2) Decision-making level.

[0102] The decision-making layer is the "brain" of the system, responsible for fusing and analyzing data collected by the perception layer, making logical judgments, and generating strategies. It is built on an embedded chip computing platform, and its core modules include:

[0103] Driver state fusion and judgment module: It adopts a deep learning algorithm (CNN+LSTM) to fuse the driver's physiological parameters and behavioral characteristics, and outputs a four-level state assessment result of "normal / mild discomfort / severe discomfort / loss of operation ability";

[0104] Takeover Trigger Judgment Module: Receives "driver active trigger signal" (one-button / voice command) or "driver severe discomfort / loss of operation ability" automatic judgment signal to trigger intelligent driving emergency takeover process;

[0105] Scenario-based parking planning module: Based on high-precision map information and real-time traffic data, it filters parking areas that meet safety requirements (priority emergency lanes, secondary priority open areas on the right side of roads, and no-occupancy fast lanes / intersections) and plans the optimal parking route (avoiding obstacles and maintaining smooth lane transitions).

[0106] Early warning and rescue strategy module: Based on the driver's status and parking location, it generates tiered early warning information (in-vehicle warning, surrounding vehicle warning, platform alarm) and rescue request information, and determines the rescue priority (prioritizes sending to emergency centers for drivers with severe discomfort).

[0107] (3) Execution layer.

[0108] The execution layer is responsible for executing the instructions issued by the decision-making layer, realizing intelligent driving takeover, emergency braking, and proactive warning, specifically including:

[0109] Intelligent driving takeover module: Takes over the vehicle's steering, braking, and power control, switches to L4 level emergency intelligent driving mode, and drives according to the planned path;

[0110] Emergency parking execution module: Controls the vehicle to decelerate smoothly (deceleration rate ≤2m / s², to avoid passengers leaning forward), switches to the target lane, stops at the preset parking area, engages the electronic parking brake, shuts off power output, and turns on the hazard warning lights;

[0111] Active warning module: It provides multi-dimensional warnings through in-vehicle voice (reminding passengers to keep quiet), lights (hazard lights + side marker lights on), horn (short beeping to alert surrounding vehicles) and V2X communication (pushing "emergency stop, please give way" information to vehicles within 100m).

[0112] (4) Linkage layer.

[0113] The linkage layer is responsible for enabling information exchange between the vehicle system and external platforms, constructing a closed-loop rescue link, specifically including:

[0114] Vehicle terminal communication module: Establishes a connection with the cloud platform through 4G / 5G network to push information such as driver status, vehicle location, vehicle model, license plate, and power system status;

[0115] Cloud-based collaborative platform: Receives information from vehicle terminals, performs data analysis and classification, and pushes it to traffic management platforms (for traffic control), emergency centers (for medical rescue), vehicle brand after-sales platforms (for vehicle towing), and driver's preset contacts (for family notification).

[0116] Feedback receiving module: Receives rescue progress information pushed by the cloud platform (such as the estimated arrival time of the ambulance and the status of traffic police handling), and displays it to passengers through the in-vehicle screen.

[0117] Figure 3 This is an overall flowchart of a vehicle takeover method according to an embodiment of this application, such as... Figure 3 As shown, in conjunction with the vehicle takeover system, the method includes:

[0118] S302, the vehicle is driving normally.

[0119] The vehicle is in manual driving mode, with the power system, steering system, and braking system directly controlled by the driver. The intelligent driving function is in standby mode, the perception layer continuously collects vehicle operation data and external environmental information, and the system is in low-power monitoring mode without intervening in control decisions.

[0120] S304, the perception layer continuously collects data.

[0121] The perception layer uses a multi-sensor fusion mechanism to collect driver physiological parameters (heart rate, blood pressure, respiratory rate), facial features (eyelid closure, head posture), steering wheel grip force, and voice commands in real time. It also simultaneously acquires vehicle status (vehicle speed, steering angle, braking pressure, powertrain health), environmental information (road type, lane lines, emergency lane location, surrounding vehicle density, weather conditions, visibility), and the dynamics of traffic participants. Furthermore, it synchronizes the data completion time of each channel with spatial coordinates to form a unified multi-dimensional environment and status perception matrix.

[0122] S306, Decision-making level: Determine whether the takeover trigger conditions are met?

[0123] The decision-making level performs fusion analysis on the perception data based on preset logic. If a driver actively triggers a signal (such as a one-button operation or the voice command "emergency takeover"), it is immediately determined that the triggering condition is met. If no active trigger is detected, the automatic judgment logic is activated. When physiological parameters exceed the safety threshold, behavioral characteristics show continuous abnormalities (such as no steering or grip change for 5 consecutive seconds) and there is no operational response, the deep learning model outputs the state of "severe discomfort" or "loss of operational ability", which is determined to be the automatic triggering condition met, and the takeover process is triggered.

[0124] S308, Execution layer: Initiate driving takeover.

[0125] The decision-making level sends a takeover command to the execution level. The intelligent driving control system immediately takes over the vehicle's steering, braking and power control authority, switches to L4 level emergency driving mode (i.e. emergency takeover mode), releases the driver's intervention authority over the control system, and the system controls the vehicle to decelerate at a preset safe rate. At the same time, it activates the electronic parking brake and pre-activates the hazard warning lights to prepare for subsequent parking actions.

[0126] S310, Execution Layer: Scenario-based Emergency Parking.

[0127] Based on the parking area and route planning instructions output by the decision-making layer, the execution layer plans a parking trajectory that meets safety constraints, combining high-precision maps and real-time environmental perception data. It prioritizes emergency lanes or open roadside areas, controls the vehicle to smoothly switch lanes one by one, confirms the safe distance behind the vehicle with radar before each lane change, controls the deceleration rate to within 2 meters per second squared, and accurately stops in the target area with the distance between the vehicle body and the curb not exceeding 30 centimeters. After the vehicle comes to a complete stop, it automatically engages the electronic handbrake and cuts off the power output.

[0128] S312, Execution Layer: Multi-dimensional proactive early warning.

[0129] During and after parking, the system simultaneously activates a tiered warning mechanism: inside the vehicle, a voice announcement reminds passengers to fasten their seat belts; outside the vehicle, hazard lights, side marker lights, and three short blasts from the horn enhance visual and auditory warnings; at the same time, the system broadcasts "emergency parking" information to all traffic participants within a 100-meter radius that support the protocol via the V2X communication module, ensuring that surrounding vehicles can detect the emergency parking in advance and take evasive action.

[0130] S314, Linkage Layer: Initiate closed-loop rescue linkage.

[0131] After the vehicle comes to a complete stop, the onboard terminal automatically integrates the driver's physiological state, precise geographical location, vehicle model, license plate, power system status, and environmental data. After encryption, the data is sent to the cloud-based collaborative platform via the 5G network. The platform prioritizes the information and pushes the data to emergency centers, traffic management departments, vehicle brand after-sales platforms, and preset emergency contacts to achieve targeted dispatch of rescue resources. At the same time, it receives rescue progress feedback from the platform (such as the estimated arrival time of the ambulance) and sends it back to the onboard display screen for passengers to view.

[0132] S316, Driver / Rescue Personnel: Determine whether to cancel emergency mode?

[0133] When the driver regains consciousness and attempts to actively operate the steering wheel, accelerator, or brake pedal, the system uses grip force sensors and operational force analysis to confirm that the driver has the ability to regain control. Alternatively, when rescue personnel arrive at the scene and initiate a cancellation request via physical buttons or remote commands, the system enters a mode to exit the judgment process.

[0134] S318, transfer of vehicle control, system returns to standby.

[0135] After confirming that the cancellation request is legal and valid, the system gradually releases the intelligent driving control authority, smoothly returns steering, braking and power control to the driver, turns off all warning signals, the sensors enter low-frequency inspection mode, and the system returns to the initial standby state, waiting for the next takeover trigger condition to be met again.

[0136] Figure 4 This is a flowchart of a vehicle takeover method according to an embodiment of this application, showing the state assessment process. Figure 4 As shown, it includes:

[0137] S402, physiological sensor for data acquisition.

[0138] The seat integrates multi-parameter physiological sensors that continuously collect key physiological indicators such as the driver's heart rate, blood pressure, and respiratory rate at a sampling frequency of no less than 10 Hz. The data is filtered and baseline corrected before being transmitted to the decision-making layer as the initial input for state assessment, ensuring data continuity and stability.

[0139] S404, determine if physiological parameters are abnormal?

[0140] The system compares the collected physiological parameters with preset individualized health threshold ranges, which are based on a dynamic benchmark model established according to dimensions such as age, gender, and physical signs. If all parameters are within the safe range, it is determined that there are no abnormalities; if any parameter exceeds the set upper or lower limit for more than two seconds consecutively, it is determined that the physiological parameter is abnormal and enters the review process.

[0141] S406 outputs the normal state and ends the judgment.

[0142] When physiological parameters do not trigger abnormal judgment conditions, the system directly determines that the driver's state is normal, terminates the current evaluation process, and does not activate the vision or behavior perception module in order to reduce the computing load and system power consumption and maintain the system's light-load operation.

[0143] S408 triggers the infrared camera, grip sensor, and voice recognition.

[0144] Under the premise of abnormal physiological parameters, the system actively activates the high-precision perception sub-module, including an infrared camera for facial feature capture, a steering wheel grip force sensor for dynamic grip force change monitoring, and a voice recognition module for semantic recognition of abnormal voice patterns (such as groans and calls for help), to achieve synchronous acquisition of multimodal behavioral features.

[0145] S410 collects facial / head / grip strength / voice features.

[0146] An infrared camera captures facial images of the driver at a frequency of 30 frames per second, extracting static characteristics such as eyelid closure duration, eyebrow tension, and degree of drooping of the corners of the mouth; a head posture sensor records the head tilt angle and sway frequency; a grip force sensor records the trend and duration of grip force amplitude changes; and a speech recognition module performs voiceprint analysis on environmental speech to identify abnormal intonation, speech rate, or keywords.

[0147] S412 uses CNN to extract static features and LSTM to extract temporal features.

[0148] Convolutional neural networks extract spatial features from infrared images to identify static patterns of facial micro-expressions; long short-term memory networks model temporal data such as grip force sequences, head posture changes, and voice energy fluctuations to capture the dynamic patterns of behavioral states evolving over time, achieving independent deep representation of static performance and dynamic trends.

[0149] S414, Feature fusion, generating state score.

[0150] The static feature vector output by the CNN and the temporal feature vector output by the LSTM are concatenated in multiple dimensions and input into the fusion neural network layer. A comprehensive state score is generated through weighted aggregation and nonlinear mapping. This score reflects the degree of overall functional decline of the driver and provides a quantitative basis for level determination.

[0151] S416, Determine the corresponding level of the status score?

[0152] The system divides the comprehensive score according to the preset four-level status scoring threshold range: a score below the first threshold is normal, a score between the first and second thresholds is mild discomfort, a score between the second and third thresholds is severe discomfort, and a score above the third threshold is loss of operational ability. Each threshold is determined by training and optimization using millions of real driving behavior samples to ensure that the discrimination accuracy is not less than 98%.

[0153] S418 outputs a slight discomfort status and sends a reminder message.

[0154] When the score falls into the mild discomfort range, the system determines that the driver has early fatigue or mild discomfort, triggering in-vehicle voice prompts and instrument panel light effects to remind the driver to rest or adjust their state, but does not initiate the takeover process, keeping driving control in the driver's hands, and only providing early warning intervention.

[0155] S420 outputs a severe discomfort status, triggering a takeover warning message.

[0156] When the score reaches the severe discomfort level, the system determines that the driver is no longer able to safely maintain vehicle control, immediately generates a takeover warning command, and initiates pre-action actions such as in-vehicle voice alarm, hazard light pre-activation, and V2X broadcast preparation to provide decision signals for subsequent execution layer takeover, but does not relinquish control at this time.

[0157] S422 outputs a state of loss of operational capability, directly triggering takeover.

[0158] When the score exceeds the highest threshold, the system determines that the driver has completely lost the ability to operate the vehicle, immediately executes the highest priority response, sends a forced takeover command to the execution layer, interrupts all manual control input channels, and seamlessly switches to L4 level emergency driving mode to ensure that the vehicle enters a safe parking process and minimizes the risk of an accident.

[0159] Figure 5 This is an emergency parking flowchart of a vehicle takeover method according to an embodiment of this application, such as... Figure 5 As shown, it includes:

[0160] S502, obtain road information.

[0161] The system calls the built-in high-precision map database to obtain the topological structure information of the road segment where the vehicle is currently located, including key road elements such as the number of lanes, lane type, shoulder width, location of emergency parking lanes, bus stops, bridge and tunnel entrances, and intersections. Combined with the Beidou and GPS dual-mode positioning results, the system accurately locates the absolute position of the vehicle in the road coordinate system, providing a geographical benchmark for subsequent parking area selection.

[0162] S504, filter candidate parking areas.

[0163] Based on the acquired road information, the system filters all potential areas that meet the parking conditions within a range of 300 to 500 meters in front of and to the side of the vehicle, including emergency lanes, non-driving areas on the right side of the road, buffer zones at service area entrances, and open ramps on the roadside, excluding prohibited or high-risk parking areas such as highway main lanes, intersections, inside curves, and tunnel exits, forming a preliminary set of candidate areas.

[0164] S506, collects surrounding information.

[0165] A three-dimensional environmental perception model of the vehicle's surroundings is constructed using lidar, millimeter-wave radar, and visual cameras. This model acquires real-time information on the distribution of static and dynamic obstacles within and around the candidate area, including physical targets such as construction barriers, bollards, pedestrians, non-motorized vehicles, and animals, thus completing a preliminary spatial assessment of the area's parking availability.

[0166] S508 excludes areas containing obstacles / pedestrians / non-motorized vehicles.

[0167] The system performs 3D point cloud semantic segmentation and dynamic target tracking on each candidate region to determine whether there are physical obstacles that hinder the safe parking of vehicles within the region. If any obstacle is detected occupying more than a preset threshold (e.g., 20%) of the parking area or is located within the wheel track, the region is marked as invalid and removed from the candidate set to ensure that the remaining areas have complete, continuous, and unobstructed parking space.

[0168] S510 acquires real-time traffic flow and calculates the safety factor.

[0169] The system receives traffic density, speed, and relative motion trends of adjacent lanes from roadside units via a V2X communication module. Combined with radar and visual detection results of surrounding vehicles, it quantitatively assesses the traffic pressure around each candidate target area. The safety factor calculation comprehensively considers the distance to vehicles approaching from behind, relative speed, lateral traffic density, nighttime visibility correction factors, and road slope. A weighted linear model generates continuous values ​​between 0 and 1, representing the potential risk level of a secondary collision in the area under the current traffic conditions.

[0170] S512, determine if the safety factor is greater than or equal to the threshold?

[0171] The system sets a minimum safety threshold of 0.85, which is determined through optimization based on historical accident data and simulation verification. If at least one candidate region meets this threshold, the optimal region selection process begins; if none of the regions meet the threshold, the current environment is deemed not to meet safe parking conditions, and an expansion strategy is initiated.

[0172] S514, Expand the filtering range, return to step S504.

[0173] When the safety factor of all candidate areas within the initial range (within 500 meters) is lower than the threshold, the system automatically expands the search range to 1000 meters, calls up the high-precision map and environmental perception module again, and repeats the process of candidate area screening, obstacle removal and safety factor calculation to ensure that parking spots that meet safety standards can still be found in extremely complex traffic scenarios, and avoids the lack of available areas due to local environmental limitations.

[0174] S516, select the area with the highest safety factor as the target parking area.

[0175] Among all candidate areas that meet the safety threshold, the system selects the one with the highest safety coefficient as the final target parking point. This selection mechanism prioritizes safety performance over proximity or shortest path. Even if the target area is located further away, it is identified as the optimal parking target as long as its risk level is the lowest, ensuring that safety is the top priority in emergency response.

[0176] S518, planned lane switching route.

[0177] Based on the coordinates and road topology of the selected target parking area, the system generates a complete path plan from the current lane to the target area. The path includes one or more lane change maneuvers. Before each lane change, radar and visual sensors confirm that there are no vehicles approaching at high speed from behind in adjacent lanes, and that the lateral safety distance is greater than two meters. The path planning follows the principles of "minimum number of lane changes, maximum smoothness, and shortest response time" to ensure that the vehicle smoothly transitions to the target position during deceleration, avoiding the risk of loss of control caused by sharp turns or frequent lane changes.

[0178] S520 outputs parking planning instructions and ends the process.

[0179] The system encapsulates the final selected target parking area coordinates, lane switching sequence, deceleration rate curve, steering angle command and other control parameters into a structured command package, sends it to the intelligent driving takeover module of the execution layer, initiates the vehicle's automatic parking process, completes the decision-making closed loop of the parking strategy, and then waits for the execution layer to provide feedback on the parking completion signal before entering the next stage of the early warning and rescue linkage process.

[0180] Figure 6This is a flowchart illustrating the early warning and rescue process of a vehicle takeover method according to an embodiment of this application, such as... Figure 6 As shown, it includes:

[0181] S602, the vehicle has completed an emergency stop.

[0182] After completing actions such as lane switching, smooth deceleration, precise parking, electronic handbrake engagement, power system disconnection, and hazard warning light activation, the execution layer sends a parking completion signal to the decision-making layer. The system confirms that the vehicle is in a stable and stationary state and enters the conditions for initiating emergency response.

[0183] S604, the vehicle terminal organizes information (driver status / location / vehicle model).

[0184] The vehicle terminal system automatically integrates structured data collected from the perception and decision-making layers, including driver physiological state assessment results (such as heart rate, blood pressure, eyelid closure duration and corresponding four-level status levels), vehicle's precise geographical location (longitude, latitude, altitude), vehicle model number, vehicle identification number, license plate information, power system operating status (such as power battery SOC, fault codes), environmental conditions (such as visibility, weather), and other key elements, to generate JSON format information packages that conform to industry standards, ensuring that the data is complete, formatted uniformly, and can be parsed by external platforms.

[0185] S606 sends information to the cloud-based collaborative platform.

[0186] The vehicle-mounted terminal establishes a stable link through the 5G mobile communication network and sends the processed information packets to the collaborative platform deployed in the cloud in an encrypted manner. The communication protocol adopts a secure authentication mechanism, supports TLS encryption and identity authentication, and ensures the integrity, confidentiality and non-repudiation of data during transmission. If 5G communication is interrupted, the system automatically switches to the 4G network. Under extreme conditions, satellite short messages are used as a supplementary channel to ensure the reliability of information delivery.

[0187] S608 uses a cloud-based collaborative platform for data processing.

[0188] After receiving the information, the cloud platform performs format verification, field mapping, and semantic normalization on the structured data by the data parsing module, eliminating redundancy or outliers and completing data standardization. Subsequently, the intelligent routing engine classifies and prioritizes the information based on its content, providing a basis for decision-making in subsequent distribution strategies.

[0189] S610, Driver Status Level?

[0190] The system performs a secondary assessment of the driver's condition level based on a preset grading logic, distinguishing between four categories: "severe discomfort," "loss of operational ability," "mild discomfort," or "normal." This assessment is based on the raw physiological data and algorithm scores reported by the vehicle terminal, ensuring consistency with the vehicle's local assessment results and preventing judgment deviations caused by network transmission.

[0191] S612, if the patient is severely unwell / loses the ability to operate the device, they should be prioritized for transport to the emergency center.

[0192] When the status level is determined to be severe discomfort or loss of operational ability, the system initiates the highest priority response, immediately pushing key information such as vehicle location, driver's vital signs, and vehicle type to the nearest regional emergency center. The pushed content includes an emergency level label and an estimated rescue window suggestion, triggering the emergency vehicle dispatch process to ensure that medical resources respond as soon as possible.

[0193] For S614 and other levels, the information will be pushed to the traffic management platform.

[0194] If the status level is mild discomfort or the system determines that the situation is not critical, the system will push the information to the traffic management department of the relevant area to assist in road network monitoring and traffic management, indicating that there is an abnormal parking event on the road section, so that traffic police can arrive at the scene in time to deal with it or issue detour notices to reduce the impact on the overall traffic flow.

[0195] S616, push to after-sales platform, preset contact person.

[0196] After completing the emergency rescue and traffic management notifications, the system simultaneously sends the vehicle's location and status information to the brand's after-sales service system to initiate vehicle towing or remote diagnostic requests. At the same time, a simplified version of the information is pushed to the driver's pre-set emergency contacts (such as family members or guardians) to notify them of the vehicle's status and location, thus realizing a supplementary mechanism for humanistic care and family response.

[0197] S618, external platform receives information and initiates processing / rescue.

[0198] After receiving the push information, each receiving platform (emergency center, traffic management system, after-sales system, contact terminal) automatically triggers the response process according to its business logic. For example, the emergency center initiates ambulance dispatch, the traffic management platform releases an early warning through the electronic information board, the after-sales platform generates a service work order, and the contact person pushes a notification to their mobile phone, thereby realizing the automated response and task distribution of rescue resources.

[0199] S620, external platform reports rescue progress to the cloud.

[0200] After initiating the response process, each external platform transmits the processing status in real time, including key information such as whether the ambulance has been dispatched, the estimated arrival time, the arrival of traffic police, and the tow truck is en route. This information is uploaded to the cloud platform through standardized interfaces, forming full lifecycle tracking data for the rescue mission.

[0201] S622, pushes rescue progress to the vehicle terminal via cloud.

[0202] The cloud platform integrates the feedback information received from various external platforms to form a unified summary of the rescue progress, which is then pushed back to the vehicle's onboard terminal through an encrypted channel to ensure real-time information synchronization and avoid data conflicts or delays from multiple sources.

[0203] The S624 features an in-vehicle screen displaying the progress of the rescue operation.

[0204] The in-vehicle human-machine interaction system dynamically presents the rescue progress on the central control display screen with a visual interface, including text prompts and progress bars such as "Ambulance has been dispatched and is expected to arrive in 8 minutes", "Traffic police have arrived at the scene", and "Tow truck is on its way". It also supports voice broadcast, so that the people in the vehicle can clearly understand the external response status, alleviate anxiety, and improve their sense of security and control during the emergency.

[0205] It should be noted that, Figures 3 to 6 Preferred embodiments of the shown examples can be found in [reference needed]. Figure 2 The corresponding solutions in the illustrated embodiments will not be described in detail here.

[0206] Figure 7 This is a structural diagram of a vehicle takeover device according to an embodiment of this application, as shown below. Figure 7 As shown, the device includes:

[0207] The acquisition module 702 is used to acquire the driving status of the target object in the target vehicle, wherein the driving status is used to indicate the physiological and behavioral status of the target object;

[0208] The evaluation module 704 is used to evaluate the target object based on its physiological and behavioral states and obtain evaluation results, wherein the evaluation results are used to quantitatively represent the target object's control capability over the target vehicle.

[0209] The control module 706 is used to control the target vehicle to enter the emergency takeover mode when the evaluation results indicate that the target vehicle meets the takeover trigger conditions.

[0210] It should be noted that, Figure 7 The vehicle takeover device shown is used to perform Figure 2 The method of vehicle takeover shown, therefore Figure 2 The relevant explanations and instructions in the vehicle takeover method also apply to Figure 7The vehicle takeover device shown is not described in detail here.

[0211] This application also provides an electronic device, which includes a memory and a processor, wherein the memory is used to store program instructions; the processor is connected to the memory and is used to execute the steps of implementing the vehicle takeover method in various embodiments of this application.

[0212] This application also provides a non-volatile storage medium including a stored computer program, wherein the device containing the non-volatile storage medium executes the steps of the vehicle takeover method in various embodiments of this application by running the computer program.

[0213] This application also provides a computer program product, including computer instructions that, when executed by a processor, implement the steps of the vehicle takeover method in various embodiments of this application.

[0214] This application also provides a computer program that, when executed by a processor, implements the steps of the vehicle takeover method in various embodiments of this application.

[0215] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0216] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0217] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0218] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0219] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0220] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0221] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for taking over vehicle control, characterized in that, include: The driving state of a target object in a target vehicle is obtained, wherein the driving state is used to indicate the physiological and behavioral state of the target object; The target object is evaluated based on the physiological state and the behavioral state to obtain an evaluation result, wherein the evaluation result is used to quantitatively represent the target object's control capability over the target vehicle; If the assessment results indicate that the target vehicle meets the takeover trigger conditions, the target vehicle is controlled to enter emergency takeover mode.

2. The method according to claim 1, characterized in that, The target object is evaluated based on the physiological state and the behavioral state to obtain evaluation results, including: The physiological parameters of the target object are collected, wherein the physiological parameters are used to reflect the physiological state of the target object; The physiological parameters are compared with preset thresholds to obtain the comparison results; When the comparison result indicates that the physiological parameter is greater than the preset threshold, the behavioral characteristics of the target object are collected, wherein the behavioral characteristics are used to characterize the behavioral state of the target object in the process of driving the target vehicle in multiple dimensions; The evaluation result is determined based on the comparison results and the behavioral characteristics.

3. The method according to claim 1, characterized in that, The takeover triggering conditions include: The first takeover trigger condition is passively triggered based on the operation instructions of the target object, wherein the operation instructions include hardware trigger instructions in the target vehicle or voice instructions of the target object; The second takeover trigger condition is actively triggered based on the preset driving state of the target object, wherein the preset driving state includes the state in which the evaluation result meets the preset level.

4. The method according to claim 3, characterized in that, The method further includes: Upon receiving an operation command from the target object, control the target vehicle to enter emergency takeover mode; or... If the assessment result indicates that the target's driving status is at Level 1, a prompt message is sent to the target, where Level 1 indicates that the target has full control over the target vehicle, and the prompt message is used to remind the target to take a rest; or... If the assessment result indicates that the target vehicle's driving status is at Level 2, a takeover warning message is sent to the target vehicle. Furthermore, if the target vehicle's inactivity duration meets a preset time, the vehicle is controlled to enter emergency takeover mode. Here, Level 2 indicates that the target vehicle has lost partial control over itself. Alternatively... If the assessment result indicates that the target vehicle's driving status is at level three, the target vehicle is controlled to enter emergency takeover mode, wherein level three indicates that the target vehicle has completely lost control of the target vehicle.

5. The method according to claim 1, characterized in that, Controlling the target vehicle to enter emergency takeover mode includes: The driving information of the target vehicle is obtained, wherein the driving information includes the target vehicle status information, the environmental information of the road where the target vehicle is located, and the traffic participant information; The parking area is determined based on the driving information; Determine the parking path of the target vehicle to the parking area, and perform an emergency parking operation according to the parking path.

6. The method according to claim 5, characterized in that, Determining the parking area based on the driving information includes: Based on the driving information, at least one parking candidate area is determined that is less than a preset distance from the target vehicle; By eliminating obstacles within the parking candidate area, at least one target candidate area is obtained; Determine a safety factor for each of the target candidate regions, wherein the safety factor is used to quantify the collision risk level caused by the density of traffic participants around the target vehicle; The target candidate area with the highest safety factor is determined as the parking area.

7. The method according to claim 5, characterized in that, The method further includes: Generate early warning information, wherein the early warning information includes at least one of the following: a first early warning information for reminding objects inside the target vehicle that have taken over the vehicle, a second early warning information for reminding objects around the target vehicle that have taken over the vehicle, and a third early warning information for reminding a platform connected to the target vehicle that has taken over the vehicle. Warnings are issued based on the aforementioned warning information.

8. A device for vehicle takeover, characterized in that, include: The acquisition module is used to acquire the driving status of the target object in the target vehicle, wherein the driving status is used to indicate the physiological and behavioral status of the target object; An evaluation module is used to evaluate the target object based on the physiological state and the behavioral state to obtain an evaluation result, wherein the evaluation result is used to quantitatively represent the target object's control capability over the target vehicle; The control module is used to control the target vehicle to enter emergency takeover mode when the evaluation result indicates that the target vehicle meets the takeover triggering conditions.

9. An electronic device, characterized in that, include: A memory and a processor, the memory being used to store program instructions; the processor being connected to the memory and used to execute the method for implementing the vehicle takeover method according to any one of claims 1 to 7.

10. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the vehicle takeover method according to any one of claims 1 to 7.

Images