Vehicle control method, vehicle, and computer-readable storage medium

Abstract

Embodiments of the present application provide a vehicle control method, a vehicle and a computer readable storage medium. The method comprises: generating a context state vector based on context information of the vehicle, wherein the context information is used to represent a driving state and a passenger state of the vehicle during driving; determining a historical state vector sequence based on the context state vector; in response to the context state vector and the historical state vector sequence satisfying a preset trigger condition, performing a strategy evolution operation based on the context state vector and the historical state vector sequence to obtain action description information, wherein the action description information is used to describe an evolved vehicle function strategy; generating a vehicle control instruction based on the action description information; and controlling the vehicle based on the vehicle control instruction. The present application solves the technical problem of low continuity, intelligence and driving safety of human-computer interaction caused by the control mode triggered based on instantaneous input or static rules in the traditional method.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and more specifically, to a vehicle control method, a vehicle, and a computer-readable storage medium. Background Technology

[0002] As driving scenarios become more complex, users expect vehicle systems to automatically optimize functional strategies based on continuous changes in driving situations without requiring repeated intervention, thereby improving the consistency and intelligence of interactions.

[0003] However, existing in-vehicle systems generally adopt instantaneous decision-making models based on single input or static rules, which only execute fixed functions for the current moment. They cannot dynamically evolve and adaptively adjust the executed strategies during driving, resulting in low continuity, intelligence, and driving safety of human-machine interaction in long-cycle scenarios such as fatigue driving.

[0004] There is currently no good solution to the above problems. Summary of the Invention

[0005] This application provides a vehicle control method, a vehicle, and a computer-readable storage medium to at least solve the technical problem that traditional methods using control methods based on instantaneous input or static rule triggering result in low continuity of human-machine interaction, low intelligence, and low driving safety.

[0006] According to one aspect of the embodiments of this application, a vehicle control method is provided, comprising: generating a context state vector based on vehicle context information, wherein the context information is used to represent the driving state and occupant state of the vehicle during driving; determining a historical state vector sequence based on the context state vector; in response to the context state vector and the historical state vector sequence satisfying a preset trigger condition, performing a strategy evolution operation based on the context state vector and the historical state vector sequence to obtain action description information, wherein the action description information is used to describe the evolved vehicle function strategy; generating vehicle control commands based on the action description information; and controlling the vehicle based on the vehicle control commands.

[0007] Furthermore, the method also includes: obtaining initial context information of the vehicle, wherein the initial context information includes at least one of the following: driving behavior data, driving object state data, vehicle condition data, road condition data, human-machine interaction data, and trip data; and aligning the initial context information with timestamps to obtain the context information.

[0008] Furthermore, generating a context state vector based on vehicle context information includes: performing feature mapping on the context information to obtain context state features, wherein the context state features include at least one of the following: fatigue level features, fatigue change trend features, driving stress features, traffic environment impact features, environmental comfort features, and remaining travel time features; normalizing the context state features to obtain a context state vector, wherein the context state vector carries a current timestamp and a target function strategy identifier, and the target function strategy identifier is used to index the vehicle function strategy adopted during the most recent execution of the strategy evolution operation.

[0009] Furthermore, each context state feature carries a confidence level.

[0010] Furthermore, determining the historical state vector sequence based on the context state vector includes: determining the historical state vector sequence based on the context state vector and a preset sliding time window, wherein the historical state vector sequence includes multiple context state vectors arranged in chronological order.

[0011] Further, the preset triggering conditions include at least one of the following: the target state feature in the context state vector is greater than the corresponding state feature threshold; the rate of change of the target state feature is greater than the preset deterioration threshold, wherein the rate of change of the target state feature is used to represent the direction and magnitude of change of the target state feature within the preset sliding time window; the difference between the current value of the target state feature and the preset baseline value is greater than the preset deviation threshold; the state difference between the context state vector and the target state snapshot is greater than the corresponding state difference threshold, wherein the target state snapshot is the context state vector recorded when the policy evolution operation was most recently executed; the first time difference between the current timestamp and the target evaluation timestamp is greater than the preset evaluation time interval, wherein the target evaluation timestamp is the timestamp of the most recently executed evolution trigger evaluation operation, and the evolution trigger evaluation operation is used to determine whether the context state vector and the historical state vector sequence meet the preset triggering conditions.

[0012] Furthermore, the action description information obtained by performing a policy evolution operation based on the context state vector and the historical state vector sequence includes: determining the target state feature change rate based on the context state vector and the historical state vector sequence; determining the adaptation value of the policy indexed by the indicator function policy identifier based on the target state feature change rate, wherein the indicator function policy identifier is used to index the current vehicle function policy, or the vehicle function policy adopted in the most recent policy evolution operation, and the adaptation value is used to indicate the degree of adaptation between the policy indexed by the indicator function policy identifier and the current driving scenario; determining the evolution type based on the adaptation value; and performing a policy evolution operation based on the evolution type to obtain the action description information.

[0013] Furthermore, the evolution type is one of the following: maintaining the vehicle function policy with indicator function policy identifier index, adjusting the parameters of the vehicle function policy with indicator function policy identifier index, or adding new execution actions to the vehicle function policy with indicator function policy identifier index.

[0014] Furthermore, the action description information includes at least one of the following: action type, action parameters, triggering conditions for executing the action, and the order of execution of the actions.

[0015] Furthermore, the method also includes: in response to the context state vector and the historical state vector sequence satisfying a preset triggering condition, calculating a second time difference between the current timestamp and the target trigger timestamp, wherein the target trigger timestamp is the timestamp of the most recent execution of the policy evolution operation; in response to the second time difference being less than a preset cooling time interval and the evolution type of the policy evolution operation being the target evolution type, delaying the execution of the policy evolution operation, wherein the target evolution type is the evolution type of the most recent executed policy evolution operation.

[0016] Furthermore, the method also includes: delaying the execution of a strategy evolution operation in response to the detection of a driving intervention operation.

[0017] According to another aspect of the embodiments of this application, a vehicle control device is also provided, comprising: a generation module, configured to generate a context state vector based on vehicle context information, wherein the context information is used to represent the driving state and occupant state of the vehicle during driving; a determination module, configured to determine a historical state vector sequence based on the context state vector; an execution module, configured to perform a strategy evolution operation based on the context state vector and the historical state vector sequence in response to the context state vector and the historical state vector sequence satisfying a preset trigger condition, thereby obtaining action description information, wherein the action description information is used to describe the evolved vehicle function strategy; and a control module, configured to generate vehicle control commands based on the action description information, and control the vehicle based on the vehicle control commands.

[0018] According to another aspect of the embodiments of this application, a vehicle is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0019] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0020] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0021] According to another aspect of the embodiments of this application, a computer program product is also provided, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the methods in various embodiments of this application.

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

[0023] In this embodiment, a vehicle control method is employed. A context state vector is generated based on the vehicle's context information, where the context information represents the vehicle's driving state and occupant state during operation. A historical state vector sequence is determined based on the context state vector. In response to the context state vector and historical state vector sequence satisfying a preset trigger condition, a strategy evolution operation is performed based on the context state vector and historical state vector sequence to obtain action description information, which describes the evolved vehicle function strategy. Vehicle control commands are generated based on the action description information, and the vehicle is controlled based on these commands. This achieves the goal of dynamically adjusting the vehicle function strategy based on continuous context states, thereby realizing the technical effect of adaptive evolution of the vehicle function strategy as the driving state and occupant state change during operation. This solves the technical problem of low continuity, intelligence, and driving safety in human-machine interaction caused by traditional methods that use control methods based on instantaneous input or static rule triggering. Attached Figure Description

[0024] 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:

[0025] Figure 1 This is a hardware structure block diagram of an optional computing terminal for a vehicle control method according to an embodiment of this application;

[0026] Figure 2 This is a flowchart of a vehicle control method according to an embodiment of this application;

[0027] Figure 3 This is a structural block diagram of a vehicle-mounted function dynamic adjustment system according to an embodiment of this application;

[0028] Figure 4This is a structural block diagram of a vehicle control device according to an embodiment of this application;

[0029] Figure 5 This is a structural block diagram of a vehicle according to an embodiment of this application. Detailed Implementation

[0030] 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.

[0031] 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.

[0032] Figure 1 This is a hardware structure block diagram of an optional computing terminal for a vehicle control method according to an embodiment of this application, such as... Figure 1 As shown, a computing terminal (e.g., a computer terminal, a mobile smart terminal, a vehicle terminal, or a cloud computing virtual terminal) may include: one or more processors 102, a memory 104, and a transmission device 106 configured to implement communication functions. Each processor 102 may include, but is not limited to, a processing component such as a microprocessor (MCU) or a field programmable gate array (FPGA).

[0033] The aforementioned computing terminal may further include: a display device 110, an input / output device 108, and a Universal Serial Bus (USB) port (this USB port can be used as one of the ports of the computer bus). Figure 1 (not shown in the image), network interface ( Figure 1 (not shown in the image), power supply ( Figure 1 (not shown in the image) and camera ( Figure 1 (Not shown in the image). Those skilled in the art will understand that... Figure 1 The structure of the computing terminal shown is for illustrative purposes only and does not impose strict limitations on the structure of the computing terminal described above. For example, the computing terminal may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0034] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits may be embodied, in whole or in part, as software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuitry may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the vehicle terminal (or mobile device).

[0035] The memory 104 can be configured to store software programs and modules for application software, such as the program instructions / data storage device corresponding to the vehicle control method in this embodiment. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the aforementioned vehicle control 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 102, and these remote memories can be connected to a computing terminal 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.

[0036] The transmission device 106 is configured to receive or transmit data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computing terminal. In one example, the transmission device 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 device 106 may be a Radio Frequency (RF) module, configured to communicate with the Internet wirelessly.

[0037] Under the above operating environment, the embodiments of this application provide the following: Figure 2 The vehicle control method shown is as follows: Figure 2 This is a flowchart of a vehicle control method according to an embodiment of this application, such as... Figure 2 As shown, the method includes the following implementation steps S201 to S204.

[0038] Step S201: Generate a context state vector based on the vehicle's context information, where the context information is used to represent the vehicle's driving state and occupant state during the driving process.

[0039] In this embodiment, the vehicle control system generates a context state vector based on the vehicle's context information.

[0040] Contextual information is used to represent the vehicle's driving status and occupant status during operation. Contextual information (i.e., multi-source contextual data) refers to a collection of various raw data reflecting the real-time environment and occupant status during vehicle operation. Contextual information may include, for example, driving behavior data such as steering wheel angle, vehicle speed, and throttle opening; occupant status data such as driver eye closure frequency and head posture; and vehicle condition data such as interior and exterior temperature and tire pressure.

[0041] A context state vector is a unified structured data representation obtained by processing the aforementioned context information. A context state vector can be represented, for example, as an array or object containing dimensions such as timestamps, fatigue level values, stress level values, and traffic impact indices, where the values ​​of each dimension are processed to be of a uniform magnitude for computation.

[0042] As can be seen, the vehicle control system first acquires the raw context information collected by the vehicle sensors and the system, then performs feature mapping and normalization on this information, and finally generates a unified context state vector containing key state indicators at the current moment. This context state vector can simultaneously represent the driving state of the vehicle and the state of the occupants.

[0043] For example, the vehicle control system collects raw data (contextual information) such as the driver's blinking frequency decreases and the steering wheel correction frequency increases. After processing, a context state vector is generated, in which the fatigue level feature value is 0.6 and the fatigue change trend feature value is positive. This context state vector represents that the driver is currently in a state of moderate fatigue and the fatigue level is increasing.

[0044] Therefore, this application achieves standardized data representation by transforming multi-source heterogeneous context information into a unified context state vector, which facilitates subsequent modules to directly use the context state vector for trend analysis and strategy evolution judgment.

[0045] Step S202: Determine the historical state vector sequence based on the context state vector.

[0046] In this embodiment of the application, the vehicle control system determines the historical state vector sequence based on the context state vector.

[0047] In this context, a historical state vector sequence refers to a collection of multiple context state vectors arranged in chronological order. For example, a historical state vector sequence could be a list containing context state vectors generated every 10 seconds over the past 5 minutes, such as [vector t-4, vector t-3, vector t-2, vector t-1, vector t] sorted by timestamp. These context state vectors record indicators such as fatigue levels and traffic impact at the corresponding moments.

[0048] It can be seen that after generating the context state vector at the current moment, the vehicle control system adds the context state vector to a predefined time window and sorts the vectors in the window according to the timestamp, thereby forming an ordered sequence (historical state vector sequence) containing the state information of the current moment and multiple previous moments.

[0049] For example, the vehicle control system currently generates a context state vector with a timestamp of 12:00:00 and stores the timestamp in a historical state vector sequence. At this time, the historical state vector sequence contains context state vectors generated at times such as 11:59:50, 11:59:55 and 12:00:00, which constitute a historical state vector sequence reflecting the state changes in the last 10 seconds.

[0050] Therefore, this application constructs a historical state vector sequence to continuously record the trend of state changes during driving, providing necessary historical data support for subsequent time-based strategy evolution evaluation. Compared with traditional solutions, this application, by retaining a continuous historical state vector sequence, enables the vehicle control system to calculate the rate of change of state indicators (such as the rate of increase in fatigue), rather than simply judging whether the current state exceeds the limit, thus more accurately identifying potential risks.

[0051] Step S203: In response to the context state vector and the historical state vector sequence satisfying the preset triggering condition, a strategy evolution operation is performed based on the context state vector and the historical state vector sequence to obtain action description information, wherein the action description information is used to describe the evolved vehicle function strategy.

[0052] In this embodiment of the application, when the context state vector and the historical state vector sequence meet the preset triggering conditions, the vehicle control system performs a strategy evolution operation based on the context state vector and the historical state vector sequence to obtain action description information.

[0053] Among them, the preset trigger conditions refer to the set of rules used to determine whether the policy adjustment rules need to be activated, such as conditions such as the state value exceeding the threshold, the trend deteriorating, the state being too different from the state when the last policy was generated, or periodic evaluation.

[0054] Strategy evolution refers to the process of re-evaluating, adjusting parameters, or adding new actions to existing functional strategies based on the latest status and historical trends. For example, after detecting increasing fatigue, a strategy evolution operation might add a navigation command to the nearest service area in addition to the existing seat vibration. Alternatively, when a sudden drop in outside light intensity and an increase in interior humidity are detected, a command might be added to switch to dehumidification and heating mode and simultaneously dim the ambient lighting, in addition to the existing air conditioning operation mode.

[0055] Action description information is used to describe the evolved vehicle function strategies. Action description information refers to a structured data format representing the specific content of the evolved function strategies. Action description information may include, for example, function type, parameters, triggering conditions, and sequence, and is not limited here. The action description information in this application is not bound to a specific execution language; for example, it can use JavaScript Object Notation (JSON), Extensible Markup Language (XML), Protocol Buffers, or other common data exchange formats.

[0056] Optionally, the action description information includes at least one of the following: action type, action parameters, triggering conditions for executing the action, and the order of execution of the actions.

[0057] In this embodiment of the application, the action description information includes at least one of the following: action type, action parameters, triggering conditions for executing the action, and the order of execution of the actions.

[0058] Among them, the action type refers to the field in the action description information used to identify the specific function category, such as navigation prompts, seat adjustment, audio playback, air conditioning control, etc., which indicates which type of function the vehicle control system needs to perform.

[0059] Action parameters refer to numerical or status fields in the action description information that are used to define the specific details of the function execution, such as intensity, duration, volume, temperature setting, etc., and are used to refine the execution effect of the action type.

[0060] The triggering condition for an action refers to the prerequisite rules or logical judgments in the action description information that define when a specific functional action should be executed, such as a specific time, a specific vehicle status, a specific user operation, or a specific environmental indicator.

[0061] The sequential relationship of actions refers to the logical arrangement or dependency of multiple action types on the timeline. For example, adjusting the volume before broadcasting the content, or executing seat heating and air conditioning temperature rise in parallel, is used to determine the execution flow of multi-action coordination.

[0062] It can be seen that the vehicle control system defines the internal components of the motion description information, clarifying that the motion description information is a data structure containing multiple dimensions. The vehicle control system can selectively fill in one or more fields according to the evolution results to fully describe the functional instructions to be executed.

[0063] For example, the action description information generated by the vehicle control system includes: the action type is to increase seat ventilation, the action parameter is gear 3, the trigger condition for executing the action is that the current temperature is greater than 30 degrees, and the order of execution of the actions is to be executed after the audio prompt.

[0064] Therefore, this application achieves decoupling between functional strategies and specific execution interfaces by defining action description information in a structured manner, thereby improving the flexibility and adaptability of vehicle function control. It can be seen that the vehicle control system first checks whether the current context state vector and the historical state vector sequence meet the preset triggering conditions. If they do, it uses the current context state vector and the historical state vector sequence to perform evolutionary calculations on the existing strategy, ultimately generating action description information containing specific functional adjustments.

[0065] For example, the vehicle control system detects that the driver's fatigue level continues to rise and exceeds the threshold (meeting the trigger condition), so it performs a strategy evolution operation and decides to add a navigation prompt to recommend a rest area while maintaining the original seat vibration wake-up strategy. The final generated action description information is structured data containing two actions: seat vibration and navigation prompt.

[0066] Therefore, this application achieves adaptive adjustment of in-vehicle function strategies as driving scenarios change through a dynamic evaluation mechanism based on continuous time states, optimizing the interactive experience without repeated user intervention. Compared to traditional solutions, this application decouples the strategy generation and execution language, achieving independence between the function strategy adjustment logic and the underlying execution framework, thus improving the flexibility of the in-vehicle function evolution mechanism.

[0067] Step S204: Generate vehicle control commands based on action description information, and control the vehicle based on the vehicle control commands.

[0068] In this embodiment, the vehicle control system generates vehicle control commands based on motion description information, and controls the vehicle based on the vehicle control commands.

[0069] Among them, vehicle control commands refer to the underlying command codes or signals that map the abstract content such as function type and parameters in the action description information into specific in-vehicle systems (such as audio, air conditioning, navigation, seats, etc.) that can be recognized and executed.

[0070] As can be seen, the vehicle control system first receives the action description information generated in the previous step, converts the information such as function type and parameters into low-level control instructions that can be understood by each subsystem of the vehicle, and then drives the corresponding vehicle components or software modules to perform operations based on these low-level control instructions.

[0071] For example, the motion description information includes seat vibration at 50% intensity. The vehicle control system converts this into an instruction for the seat control module. After the vehicle control system executes the vehicle control instruction, the vibration motor inside the seat starts working at 50% power.

[0072] Thus, this application realizes the transformation from abstract strategy to concrete execution, ensuring that the evolved functional strategy can be accurately implemented and form a complete control loop.

[0073] In summary, firstly, this application generates a context state vector based on the vehicle's context information, where the context information represents the vehicle's driving state and occupant state during the driving process. Thus, this application achieves standardized data representation by transforming multi-source heterogeneous context information into a unified context state vector, which facilitates subsequent modules to directly utilize the context state vector for trend analysis and strategy evolution judgment.

[0074] Secondly, this application determines the historical state vector sequence based on the context state vector. Thus, by constructing the historical state vector sequence, this application realizes the continuous recording of the state change trend during driving, providing necessary historical data support for subsequent time-based strategy evolution evaluation.

[0075] Then, when the context state vector and the historical state vector sequence meet the preset triggering conditions, this application performs a strategy evolution operation based on the context state vector and the historical state vector sequence to obtain action description information. The action description information is used to describe the evolved vehicle function strategy. Thus, this application realizes the adaptive dynamic adjustment of the vehicle function strategy with the driving situation through a dynamic evaluation mechanism based on continuous time states, which can optimize the interactive experience without repeated user intervention.

[0076] Finally, this application generates vehicle control commands based on action description information, and controls the vehicle based on the vehicle control commands. Thus, this application realizes the transformation from abstract strategy to specific execution, ensuring that the evolved functional strategy can be accurately implemented and formed a complete control closed loop.

[0077] In this embodiment, a vehicle control method is employed. A context state vector is generated based on the vehicle's context information, where the context information represents the vehicle's driving state and occupant state during operation. A historical state vector sequence is determined based on the context state vector. In response to the context state vector and historical state vector sequence satisfying a preset trigger condition, a strategy evolution operation is performed based on the context state vector and historical state vector sequence to obtain action description information, which describes the evolved vehicle function strategy. Vehicle control commands are generated based on the action description information, and the vehicle is controlled based on these commands. This achieves the goal of dynamically adjusting the vehicle function strategy based on continuous context states, thereby realizing the technical effect of adaptive evolution of the vehicle function strategy as the driving state and occupant state change during operation. This solves the technical problem of low continuity, intelligence, and driving safety in human-machine interaction caused by traditional methods that use control methods based on instantaneous input or static rule triggering.

[0078] Optionally, the method further includes:

[0079] Step S205: Obtain the initial context information of the vehicle, wherein the initial context information includes at least one of the following: driving behavior data, driving object status data, vehicle condition data, road condition data, human-machine interaction data, and trip data;

[0080] Step S206: Timestamp alignment is performed on the initial context information to obtain the context information.

[0081] In this embodiment, the vehicle control system first obtains the initial context information of the vehicle.

[0082] The initial context information includes at least one of the following: driving behavior data, driving object status data, vehicle condition data, road condition data, human-computer interaction data, and trip data.

[0083] Driving behavior data reflects the driver's operational characteristics, such as steering wheel angle, pedal opening, and lane departure. Driving object state data reflects the driver's physiological or psychological state, such as blinking frequency, head posture, and fatigue levels. Vehicle condition data reflects the vehicle's own operating status, such as tire pressure, battery charge, and internal / external temperature. Road condition data reflects the external environment in which the vehicle is located, such as traffic density, weather conditions, and road grade. Human-machine interaction data reflects the user's interaction with the vehicle's infotainment system, such as voice content, emotional characteristics, and recent operation time. Trip data reflects the vehicle's travel plan, such as destination, remaining distance, or time.

[0084] It can be seen that the vehicle control system collects the above-mentioned raw data in real time or periodically by calling the hardware or software interfaces of the vehicle's internal sensors, bus interfaces, cameras, microphones and navigation systems.

[0085] For example, the vehicle control system obtains the current vehicle speed as 80 km / h from the Controller Area Network (CAN) bus, the driver's head yaw angle as 10 degrees from the Driver Monitoring System (DMS) camera, and the remaining travel time as 2 hours from the Navigation Application Programming Interface (API). The sum of these various raw data constitutes the initial context information.

[0086] Therefore, this application provides a rich and fundamental source of information for constructing accurate context state vectors by comprehensively collecting multi-dimensional raw data.

[0087] Secondly, the vehicle control system timestamps the initial context information to obtain the context information.

[0088] Timestamp alignment refers to the process of mapping multiple data points with different timestamps onto a unified time base through interpolation, discarding, or association.

[0089] It can be seen that the vehicle control system receives the initial context information with its own timestamp from the previous step, and performs time matching or synchronization processing on the data from each data source with a unified time reference (such as the current sampling time) as the center, eliminating time differences and forming a synchronized set of context information.

[0090] For example, the vehicle control system needs to generate a state vector at 12:00:00.0. At this time, the vehicle control system searches for data near 12:00:00.0, and uses interpolation or nearest neighbor matching to uniformly mark the tire pressure data at 12:00:00.05 and the fatigue index data at 12:00:00.02 as data at 12:00:00.0, thus obtaining the synchronization context information at that time.

[0091] Therefore, this application eliminates the time error between the initial context information of multi-source data through time synchronization processing, ensuring the accuracy and logical consistency of subsequent state vector generation.

[0092] Optionally, generating a context state vector based on vehicle context information includes:

[0093] Step S2011: Perform feature mapping on the context information to obtain context state features, wherein the context state features include at least one of the following: fatigue level features, fatigue change trend features, driving stress features, traffic environment impact features, environmental comfort features, and remaining travel time features.

[0094] Step S2012: Normalize the context state features to obtain a context state vector. The context state vector carries the current timestamp and the target function policy identifier. The target function policy identifier is used to index the vehicle function policy adopted when the policy evolution operation was most recently executed.

[0095] In this embodiment of the application, the vehicle control system first performs feature mapping on the context information to obtain context state features.

[0096] Feature mapping refers to the process of converting raw data into numerical or symbolic representations that reflect specific driving context dimensions using specific algorithms, models, or rules.

[0097] Contextual state features include at least one of the following: fatigue level features, fatigue change trend features, driving stress features, traffic environment impact features, environmental comfort features, and remaining travel time features. Contextual state features refer to indicator data obtained after feature mapping, used to characterize specific dimensions of the driving situation.

[0098] Fatigue level characteristics (denoted as fatigue_level) refer to indicators that quantify the fatigue level of drivers, such as the fatigue index calculated based on blink frequency.

[0099] Fatigue trend characteristics (denoted as fatigue_trend) refer to indicators that reflect the direction of change of fatigue level over time, such as the difference or slope of the fatigue index.

[0100] Driving stress characteristics (denoted as stress_level) refer to indicators that reflect the driver's tension or stress state, such as values ​​calculated based on the frequency of steering wheel fine-tuning or heart rate variability.

[0101] Traffic impact characteristics (referred to as traffic_impact) refer to indicators that reflect the degree of impact of current traffic conditions on driving, such as congestion or danger coefficients calculated based on vehicle distance and acceleration / deceleration frequency.

[0102] Environmental comfort characteristics (denoted as thermal_discomfort) refer to indicators that reflect the comfort of the in-vehicle environment, such as the comfort index calculated based on temperature, humidity, and noise levels.

[0103] The remaining trip time feature (denoted as remaining_trip_time) refers to an indicator that reflects the remaining travel time to the destination.

[0104] It can be seen that the vehicle control system receives the aligned context information, extracts and calculates the above-mentioned feature values ​​through a preset mapping logic or model, and forms a structured set of context state features.

[0105] For example, the vehicle control system reads the aligned driver monitoring system data, calculates the current fatigue index as 0.6, and the fatigue growth rate over the past five minutes as +0.1, obtaining fatigue degree characteristics and fatigue change trend characteristics, respectively. The vehicle control system reads the steering wheel angle variance and calculates the driving stress characteristic as 0.4. The vehicle control system reads the distance to surrounding vehicles and calculates the traffic environment impact characteristic as 0.7. The vehicle control system reads the in-vehicle temperature deviation and calculates the environmental comfort characteristic as 0.3. The vehicle control system reads navigation information and calculates the remaining travel time characteristic as 120 minutes.

[0106] Therefore, this application transforms the original data into higher-level features with more semantic meaning through feature mapping, providing a data foundation for generating a unified state vector.

[0107] Secondly, the vehicle control system normalizes the context state features to obtain the context state vector.

[0108] Normalization refers to the process of mapping values ​​of different ranges and dimensions to a unified interval (such as [0, 1]) to eliminate the influence of dimensions and facilitate subsequent calculations and comparisons.

[0109] The context state vector (denoted as ContextVector) carries the current timestamp and the target function policy identifier. The target function policy identifier is used to index the vehicle function policy adopted during the most recent policy evolution operation. The context state vector refers to a normalized, ordered set of data containing multiple state dimensions.

[0110] The current timestamp is a time stamp that identifies (denoted as timestamp) the moment when the context state vector was generated.

[0111] The target functional strategy identifier (denoted as last_strategy_id) is used to uniquely identify the number or identifier (ID) of the most recently executed functional strategy.

[0112] It can be seen that the vehicle control system receives the context state features, performs normalization calculations on each feature value to make the context state features fall within a unified range, and appends the current timestamp and the target function strategy identifier to the context state vector to form a complete context state vector.

[0113] For example, the vehicle control system maintains or maps the fatigue level feature (0.6), fatigue change trend feature (0.1), driving stress feature (0.4), traffic environment impact feature (0.7), and environmental comfort feature (0.3) to the zero-to-one interval, maps the remaining travel time feature (120 minutes) to 0.8, and then constructs a vector structure containing the current timestamp, each normalized feature value, and the target function strategy identifier.

[0114] Therefore, this application forms a standard and complete context state vector through normalization and additional metadata, which is convenient for subsequent modules to directly call and process.

[0115] Optionally, any context state feature carries a confidence level.

[0116] In this embodiment, each context state feature carries a confidence level. A higher confidence level indicates that the context state feature more accurately reflects the current state. A lower confidence level may indicate high noise in the sensor data, high uncertainty in model predictions, or incomplete data acquisition.

[0117] For example, for a calculated fatigue level feature of 0.6, the vehicle control system simultaneously generates a confidence level of 0.85 for the corresponding fatigue level feature, indicating that the fatigue assessment result has 85% confidence. For a driving stress feature of 0.4, the vehicle control system generates a confidence level of 0.60 for the corresponding driving stress feature, indicating that the stress assessment result has 60% confidence.

[0118] Therefore, by introducing confidence levels, this application provides a basis for data quality assessment for subsequent strategy evolution, which helps the vehicle control system to adopt a more conservative strategy or rely on other features with higher confidence levels when the confidence level is low, thereby improving the robustness of decision-making.

[0119] Optionally, determining the historical state vector sequence based on the context state vector includes:

[0120] Step S2021: Based on the context state vector and the preset sliding time window, determine the historical state vector sequence, wherein the historical state vector sequence includes multiple context state vectors arranged in chronological order.

[0121] In this embodiment, the vehicle control system determines the historical state vector sequence based on the context state vector and a preset sliding time window.

[0122] The preset sliding time window refers to a fixed length or number of time periods on the time axis. This preset sliding time window slides forward as the latest state vector is generated, and is used to capture the most recent segment of historical data. The preset sliding time window can be set to the past 5 minutes or the most recent 10 sampling points.

[0123] The historical state vector sequence includes multiple context state vectors arranged in chronological order.

[0124] It can be seen that the vehicle control system uses the currently generated context state vector as a basis, combined with the preset sliding time window range, to select or generate a series of context state vectors arranged in chronological order from historical data, thus forming a historical state vector sequence.

[0125] For example, the vehicle control system sets the sliding time window to the most recent 5 sampling periods. When a new context state vector is generated at the current moment, the vehicle control system combines the new context state vector with the context state vectors of the previous 4 moments to form a historical state vector sequence containing 5 vectors, arranged from earliest to latest time.

[0126] Therefore, this application provides time-dimensional information on the evolution of driving scenarios by constructing a historical state vector sequence, providing a data foundation for subsequent analysis of state change trends.

[0127] Optionally, the preset trigger conditions include at least one of the following:

[0128] The target state feature in the context state vector is greater than the corresponding state feature threshold.

[0129] The rate of change of the target state feature is greater than the preset deterioration threshold, where the rate of change of the target state feature is used to represent the direction and magnitude of change of the target state feature within the preset sliding time window;

[0130] The difference between the current value of the target state feature and the preset baseline value is greater than the preset deviation threshold;

[0131] The state difference between the context state vector and the target state snapshot is greater than the corresponding state difference threshold, where the target state snapshot is the context state vector recorded during the most recent execution of the policy evolution operation;

[0132] The first time difference between the current timestamp and the target evaluation timestamp is greater than the preset evaluation time interval. The target evaluation timestamp is the timestamp of the most recent execution of the evolution-triggered evaluation operation. The evolution-triggered evaluation operation is used to determine whether the context state vector and the sequence of historical state vectors meet the preset triggering conditions.

[0133] In this embodiment, the target state feature refers to the state indicator in the context state vector used to monitor a specific dimension (such as fatigue level, stress level, etc.). The state feature threshold refers to the warning value set for a specific state feature. The state feature threshold can be obtained from the historical state vector sequence. When the feature value exceeds this warning value, it indicates that the state of that dimension is abnormal.

[0134] It can be seen that when the vehicle control system detects that the target state feature in the context state vector is greater than the corresponding state feature threshold, and the change conforms to the evolution pattern reflected by the historical sequence, it determines that the context state vector and the historical state vector sequence meet the preset triggering conditions.

[0135] The target state feature change rate is calculated based on the context state vector and the historical state vector sequence. It represents the direction and magnitude of change of the target state feature within a preset sliding time window. For example, the target state feature change rate refers to the rate of change of the target state feature within the preset sliding time window, including the direction of change (e.g., increase or decrease) and the magnitude (e.g., the amount of change). The preset deterioration threshold is a critical value used to determine whether the change of the target state feature is too drastic or developing in a detrimental direction.

[0136] It can be seen that by introducing historical state vector sequences to calculate the rate of change, the vehicle control system can dynamically capture the real-time evolution trend of the target state characteristics, rather than relying solely on static thresholds, thereby more accurately identifying deteriorating scenarios with temporal continuity, such as fatigue accumulation and rapid power consumption.

[0137] Therefore, this application calculates the rate of change through historical state vector sequences, which can sensitively reflect the dynamic process of state changes, avoid misjudgments caused by data fluctuations at a single moment, and improve the recognition accuracy of progressively deteriorating state scenarios. The preset baseline value refers to the reference value that the target state characteristics should have under normal or ideal driving conditions. The preset deviation threshold refers to the maximum allowable difference between the target state characteristics and the baseline value.

[0138] It can be seen that the vehicle control system obtains the current target state feature value from the context state vector and subtracts it from the static or dynamic baseline value calculated from the historical state vector sequence to obtain the difference, thereby quantifying the degree to which the current state deviates from the normal reference.

[0139] Therefore, by comparing the current value with a preset baseline value generated based on the historical state vector sequence, this application can accurately identify the abnormal deviation of the current target state from the normal level, providing a reliable basis for policy evolution.

[0140] The target state snapshot is the context state vector recorded during the most recent execution of the policy evolution operation. Historical state vector sequences can support the generation of the target state snapshot or serve as a time reference for verifying the validity of the current context state vector. The state difference value refers to the numerical or dimensional difference between the current context state vector and the target state snapshot.

[0141] It can be seen that the vehicle control system obtains the context state vector at the current moment, subtracts the feature values ​​of each dimension from the feature values ​​of the corresponding dimension in the target state snapshot, or calculates the Euclidean distance / cosine similarity to obtain the state difference value, thereby measuring the extent to which the driving or riding situation has changed since the last policy adjustment.

[0142] Therefore, by comparing the current context state vector with the target state snapshot, this application can accurately capture the context drift between two policy executions, ensuring that the functional policy always adapts to the latest dynamic driving environment.

[0143] The target evaluation timestamp is the timestamp of the most recent execution of the evolution-triggered evaluation operation. The evolution-triggered evaluation operation is used to determine whether the context state vector and historical state vector sequence meet preset trigger conditions. The preset evaluation time interval refers to the minimum time interval allowed between two evolution-triggered evaluations by the vehicle control system, used to prevent evaluations from being too frequent.

[0144] As can be seen, the vehicle control system reads the current timestamp and retrieves the target evaluation timestamp from the historical state vector sequence or system log, calculates the difference between the two, and thus determines whether a sufficient time period has passed since the last evaluation operation, so as to decide whether a new round of state evaluation is needed.

[0145] Therefore, this application, through a time-interval-based periodic triggering mechanism, can ensure the real-time response of the system while rationally allocating computing resources, and ensure that the vehicle function policy can be periodically verified and updated based on the latest context state vector.

[0146] It can be seen that the vehicle control system defines the current value of the target state characteristic as one of the conditions for triggering strategy evolution when it exceeds its corresponding preset warning value.

[0147] For example, the vehicle control system sets the state feature threshold for fatigue level to 0.7. When the vehicle control system generates a new context state vector, it finds that the fatigue level feature value is 0.75. Since 0.75 is greater than 0.7, the triggering condition that the target state feature is greater than the corresponding state feature threshold is met.

[0148] Therefore, by setting a static warning threshold, this application can quickly identify the absolute level of state deterioration and ensure that strategy adjustments are initiated in a timely manner when the state is severely abnormal.

[0149] Optionally, a policy evolution operation is performed based on the context state vector and the sequence of historical state vectors to obtain action description information, including:

[0150] Step S2031: Determine the target state feature change rate based on the context state vector and the historical state vector sequence;

[0151] Step S2032: Determine the adaptation value of the policy index of the indicator function policy identifier based on the target state feature change rate. The indicator function policy identifier is used to index the current vehicle function policy or the vehicle function policy used when the policy evolution operation was last performed. The adaptation value is used to indicate the degree of adaptation between the policy index of the indicator function policy identifier and the current driving scenario.

[0152] Step S2033: Determine the evolution type based on the fitness value;

[0153] Step S2034: Perform a strategy evolution operation based on the evolution type to obtain action description information.

[0154] In this embodiment, the vehicle control system first determines the target state characteristic change rate based on the context state vector and the historical state vector sequence.

[0155] Among them, the target state feature change rate refers to the rate of change of the target state feature within a preset sliding time window. It is calculated by comparing the context state vector and the historical state vector sequence and is used to represent the direction (such as rising or falling) and magnitude (such as the amount of change) of the target state feature over time.

[0156] It can be seen that the vehicle control system uses the currently generated context state vector and the previously accumulated historical state vector sequence to obtain the rate of change of the target state characteristics within a specified time window through specific calculation methods (such as differential or regression analysis).

[0157] For example, the vehicle control system collects the current fatigue level as 0.7 and reads the historical fatigue level sequence of the past 3 minutes as [0.5, 0.55, 0.6, 0.65, 0.7], and calculates that the target state characteristic change rate of fatigue level increases by 0.1 per minute.

[0158] Therefore, this application provides a dynamic data basis for subsequent evaluation of the effectiveness of the current strategy by quantifying the rate of change of state characteristics.

[0159] Secondly, the vehicle control system determines the appropriate value for the strategy of the indicator function strategy identifier index based on the rate of change of the target state characteristics.

[0160] Among them, the indicator function policy identifier is the function policy identifier that is currently being executed or recently executed, used to index the current vehicle function policy, or the vehicle function policy used when the policy evolution operation was most recently executed.

[0161] The adaptation value indicates the degree to which the strategy indicated by the function strategy identifier index is adapted to the current driving scenario. In other words, the adaptation value is used to analyze the direction and magnitude of changes in the target state characteristics and to determine whether the strategy indicated by the function strategy identifier index is effective.

[0162] It can be seen that the vehicle control system evaluates and calculates an adaptation value based on the calculated rate of change of the target state characteristics and the indication function strategy. This adaptation value reflects the suitability of the existing strategy under the current changing situation.

[0163] For example, the vehicle control system detects that the current fatigue change rate exceeds a threshold and indicates that the function strategy is seat vibration. The vehicle control system calculates the adaptation value of the indication function strategy to the current fatigue state based on the magnitude of the change rate, which is 0.4, indicating that the current indication function strategy is not effective.

[0164] Therefore, this application objectively judges whether and how to adjust the strategy by quantifying the degree of adaptability between the functional strategy and the scenario.

[0165] At the same time, the vehicle control system will determine the evolution type based on the adaptation value.

[0166] Evolution type refers to the policy adjustment category determined by the adaptation value, which usually includes specific types such as maintaining the current policy, making parameter-level adjustments to the original policy, or adding new functional actions.

[0167] It can be seen that the vehicle control system determines the specific type of this strategy evolution operation by comparing the adaptation value calculated in the previous steps with the preset judgment rules or mapping relationship.

[0168] For example, the vehicle control system calculates an adaptation value of 0.3, which is lower than the preset minimum adaptation threshold of 0.5, and therefore determines the evolution type as adding a new navigation suggestion function.

[0169] Therefore, this application transforms the abstract degree of adaptation into a specific direction of strategy adjustment, guiding subsequent evolutionary operations.

[0170] Finally, the vehicle control system will perform strategy evolution operations based on the evolution type to obtain action description information.

[0171] It can be seen that the vehicle control system executes specific strategy evolution operations based on the determined evolution type and outputs the final action description information, which will be used to guide the subsequent function execution.

[0172] For example, based on the determined additional function action type, the vehicle control system generates action description information such as popping up a navigation prompt for the nearest service area while maintaining audio wake-up.

[0173] Therefore, this application generates instruction descriptions that are dynamically adapted to the current driving scenario, enabling continuous optimization of in-vehicle function strategies.

[0174] Optionally, the evolution type is one of the following: maintaining the vehicle function policy with the indicator function policy identifier index, adjusting the parameters of the vehicle function policy with the indicator function policy identifier index, or adding new execution actions to the vehicle function policy with the indicator function policy identifier index.

[0175] In this application embodiment, the evolution type includes at least one of the following: maintaining the vehicle function policy identifier index, adjusting the parameters of the vehicle function policy identifier index, and adding a new execution action to the vehicle function policy identifier index.

[0176] The "maintain" strategy refers to the vehicle control system not modifying any aspects of the currently executing or most recently executed vehicle function strategy when the evolution type is determined to be "maintain." This includes not changing function actions, not changing execution parameters, and not adding or reducing function actions. For example, if soothing music is currently playing and the vehicle control system determines that the soothing music is effective in relieving fatigue, then the evolution type is "maintain," meaning the vehicle control system continues to play the music in the current playlist without changing any settings.

[0177] Parameter adjustment refers to modifying the specific execution parameters of the core function, such as intensity, duration, frequency, and volume, while keeping the core function of the original in-vehicle function strategy unchanged, in order to optimize the experience or effect in the current scenario. For example, if the current seat vibration frequency is once per second, and the vehicle control system evaluates that the frequency is insufficient, the evolution type is parameter adjustment, that is, the vehicle control system adjusts the vibration frequency to twice per second, but does not change the vibration action itself.

[0178] Additional actions refer to adding one or more functions not included in the original strategy, while retaining the functions already present in the original in-vehicle function strategy, to form a combined function execution strategy. For example, if music is currently playing, and the vehicle control system determines that music alone is insufficient to wake a fatigued driver, then the evolution type is an additional action, meaning that the vehicle control system continues to play music while adding the action of air conditioning vents blowing air.

[0179] It can be seen that the vehicle control system has specific limitations on the evolution results, which clarifies that when the vehicle control system performs strategy evolution, it selects one of the above three preset operation modes as the final evolution type, thereby determining the specific generation method of subsequent action description information.

[0180] Therefore, by defining a specific evolution type, this application makes the strategy evolution process deterministic and predictable, which facilitates the implementation and control of system logic.

[0181] Optionally, the method further includes:

[0182] Step S207: In response to the context state vector and the historical state vector sequence satisfying the preset triggering condition, calculate the second time difference between the current timestamp and the target trigger timestamp, where the target trigger timestamp is the timestamp of the most recent execution of the policy evolution operation;

[0183] In step S208, in response to the second time difference being less than the preset cooling time interval and the evolution type of the strategy evolution operation being the target evolution type, the execution of the strategy evolution operation is delayed, wherein the target evolution type is the evolution type of the most recently executed strategy evolution operation.

[0184] In this embodiment of the application, when the context state vector and the historical state vector sequence meet the preset triggering conditions, the vehicle control system calculates the second time difference between the current timestamp and the target trigger timestamp.

[0185] The target trigger timestamp is the timestamp of the most recent execution of the strategy evolution operation.

[0186] The second time difference refers to the time interval between the current timestamp and the target trigger timestamp, used to measure the length of time since the last policy evolution operation.

[0187] It can be seen that when the vehicle control system detects that the preset triggering conditions are met, it obtains the current timestamp and calculates the second time difference between this time and the target trigger timestamp of the most recent strategy evolution operation, thereby quantifying the time interval between the two strategy evolution operations.

[0188] For example, the vehicle control system detects the fatigue state triggering evolution condition at 10:05:00. At this time, the vehicle control system records the current time as 10:05:00 and reads the last evolution execution time as 10:04:30, and calculates the second time difference as 30 seconds.

[0189] Therefore, this application provides basic data support for determining whether image stabilization is needed by calculating time intervals.

[0190] When the second time difference is less than the preset cooling time interval, and the evolution type of the strategy evolution operation is the target evolution type, the vehicle control system delays the execution of the strategy evolution operation.

[0191] The preset cooling time interval refers to a minimum time threshold preset by the vehicle control system. It is used to limit the frequent triggering of the same type of strategy evolution operation, so that the same strategy evolution operation is not repeatedly triggered within the cooling time, and to prevent the vehicle control system from being overloaded or frequently interfering with the user due to minor fluctuations in state.

[0192] The target evolution type is the evolution type of the most recently executed policy evolution operation.

[0193] It can be seen that the vehicle control system determines whether the currently calculated second time difference is less than the preset cooling time interval, and whether the currently triggered evolution type is the same as the most recently executed evolution type. If both conditions are met, the vehicle control system decides not to immediately execute the current strategy evolution operation, but to perform delayed processing to avoid repeated or excessive strategy adjustments.

[0194] For example, the preset cooling interval is 60 seconds. The vehicle control system detects the trigger condition at 10:05:00 and calculates the second time difference to be 30 seconds (less than 60 seconds). Simultaneously, the currently triggered evolution type is seat vibration, and the previously executed evolution type was also seat vibration. Since the second time difference is less than the preset cooling interval and the evolution type is the same, the vehicle control system determines to delay the execution of the current seat vibration strategy evolution.

[0195] Therefore, this application effectively prevents the frequent repetitive triggering of strategy evolution operations by introducing a dual judgment mechanism based on time and evolution type, thereby reducing system load and improving the stability of user experience.

[0196] Optionally, the method further includes:

[0197] Step S209: In response to the detection of a driving intervention operation, delay the execution of the strategy evolution operation.

[0198] In this embodiment of the application, when a driving intervention operation is detected, the vehicle control system delays the execution of the strategy evolution operation.

[0199] Driving intervention refers to actions taken by the driver during driving to control vehicle functions or change system states, such as manually adjusting the air conditioning temperature, manually switching audio sources, manually setting a navigation destination, or manually disabling a certain reminder function. This driving intervention reflects the driver's subjective intention.

[0200] It can be seen that when the vehicle control system detects that the driver has made an active driving intervention, it immediately suspends or terminates the strategy evolution operation that is currently waiting to be executed, and prioritizes responding to the driver's immediate manual control to avoid conflicts or interference between the automatic evolution strategy and the driver's active intention.

[0201] For example, the vehicle control system calculates that an evolution operation to trigger the recommended service area needs to be performed. Just before this evolution operation is about to be executed, the vehicle control system detects that the driver has manually activated Do Not Disturb mode or manually entered a command to skip this recommendation. The vehicle control system recognizes the driving intervention and therefore delays the execution of the recommended service area evolution operation, thus no longer disturbing the driver.

[0202] Therefore, by prioritizing the response to the driver's active intervention, this application avoids interference from the vehicle control system's automatic evolution function on user operation, ensuring that the driver always has the initiative in human-computer interaction, thus improving the safety and user-friendly experience.

[0203] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0204] According to one embodiment of this application, this application provides a vehicle function dynamic adjustment system based on time-continuous state driving. Figure 3 This is a structural block diagram of a vehicle-mounted function dynamic adjustment system according to an embodiment of this application, such as... Figure 3 As shown, the vehicle function dynamic adjustment system 300 includes at least a multi-source context acquisition module 301, a context state vector generation module 302, a state change evaluation and evolution triggering module 303, an evolution strategy generation module 304, an action description output module 305, and an optional execution interface module 306. It should be noted that the core of this application lies in the evolution generation mechanism of the function strategy, and it does not limit the execution interface through which the evolution strategy is ultimately implemented. The multi-source context acquisition module 301 is used to continuously collect multi-source information related to the driving and passenger states during vehicle operation, including driving behavior data such as steering wheel fine-tuning frequency, accelerator and brake pedal changes, lane departure, etc.; driver state data such as blinking frequency, head posture, fatigue indicators, etc.; vehicle condition data such as tire pressure, power battery charge, and interior and exterior temperatures; road condition data such as real-time traffic density, weather conditions, and road grade; human-computer interaction data such as voice interaction content, voice emotion characteristics, and the time of the most recent user active operation; and trip data such as destination, remaining trip time, or distance. The multi-source context acquisition module 301 outputs the raw context data stream in a timestamp-aligned manner, providing basic data support for subsequent state modeling.

[0205] The context state vector generation module 302 is used to convert multi-source heterogeneous data into a unified state vector expression to form a context state vector. This context state vector includes dimensions such as timestamp, fatigue level, fatigue change trend, stress level, traffic impact, thermal comfort, remaining travel time, and the previous strategy identifier. The state vector is continuous time series data, and each dimension of the state can be accompanied by a confidence level. It also supports sliding time windows to calculate the change trend.

[0206] The state change assessment and evolution triggering module 303 is used to determine whether the current functional strategy needs to be adjusted through evolution. Its triggering conditions include a state value exceeding a preset threshold, a continuously deteriorating state change trend or deviation from the baseline, a difference between the current state and the state at the time of the last strategy generation exceeding a threshold, or reaching a preset periodic assessment time point. Simultaneously, the state change assessment and evolution triggering module 303 is equipped with anti-jitter and suppression mechanisms to ensure that the same evolution type is not repeatedly triggered within the cooling-off period, and to temporarily suspend automatic evolution when the user actively intervenes, in order to avoid overloading the vehicle function dynamic adjustment system or interfering with user operation.

[0207] The evolutionary strategy generation module 304 is the core module of this application. It is used to generate new or adjusted functional strategies based on the already executed strategies and the latest state changes. The inputs to the evolutionary strategy generation module 304 include the current context state vector, the historical state vector sequence, and the identifier of the currently executing or most recently executed functional strategy. Logically, the evolutionary strategy generation module 304 analyzes the direction and magnitude of state changes, determines whether the original strategy is effective, and generates one of three evolutionary results: maintaining the current strategy, making parameter-level adjustments to the original strategy, or adding new functional actions based on the original strategy. Each evolution is causally related to the state change. The evolutionary result is not a repetition of the original strategy, but a dynamic adjustment based on model reasoning, rules, or a combination thereof, thereby ensuring that the functional control behavior remains consistent with the driving context.

[0208] The action description output module 305 is used to convert evolution strategies into action description information that is independent of specific execution methods. This action description information includes functional action types, action parameters, and triggering conditions or sequential relationships for executing actions, but is not limited to a specific execution language or framework. This means that the action description information generated by the action description output module 305 can be flexibly executed through functional semantic orchestration languages, rule engines, predefined action mapping tables, or other executable interfaces.

[0209] If the vehicle function dynamic adjustment system includes an execution interface module 306, the execution interface module 306 is not a necessary component of this application. The execution interface module 306 is used to receive action description information and convert the action description information into vehicle executable instructions. Different execution methods can be adopted in different embodiments, and the execution result can be fed back to the status acquisition module to form a closed-loop control, so that the vehicle function dynamic adjustment system can further optimize the subsequent context state evaluation and strategy evolution based on the execution effect.

[0210] Through the collaborative work of the above modules, this application realizes the continuous perception, dynamic reasoning and adaptive evolution of vehicle function strategies in the time dimension, effectively solving the problems of rigid function strategies and lack of continuous optimization capabilities in the prior art.

[0211] Compared with existing technologies, this application has at least the following beneficial effects. First, by constructing a continuous context state modeling mechanism, this application enables continuous perception of the driving process, overcoming the limitations of existing vehicle control systems that make decisions based solely on instantaneous input, thus achieving a more accurate and comprehensive understanding of the driving situation. Second, through a re-inference mechanism based on state changes, this application realizes the dynamic evolution of functional strategies, allowing in-vehicle functions to adapt to continuous changes in context states during driving, rather than being limited to one-time execution, significantly improving the intelligence level of the vehicle control system. Furthermore, because the vehicle control system can proactively optimize functional strategies based on situational changes, it reduces repetitive operations required by users in complex driving scenarios, reducing the user's operational burden and effectively improving driving safety and passenger comfort. Finally, the action description information proposed in this application has high abstraction and universality; the execution method of the action description information is not limited by a specific execution language or framework, and can flexibly adapt to different execution interfaces according to actual application scenarios, greatly improving the flexibility of functional strategy implementation and the compatibility of the vehicle control system.

[0212] This application provides specific implementation methods to further clarify the technical solution of this application. Example 1 mainly demonstrates the continuous functional evolution process based on a fatigue driving scenario. Example 1 is as follows: During long-distance driving, the driver's state dynamically changes over time. This example aims to illustrate how the vehicle control system responds to such continuous changes in contextual state. First, the vehicle control system continuously collects physiological characteristic data such as the driver's eyelid closure frequency, gaze direction changes, and head posture changes through a driver state monitoring camera. Simultaneously, it acquires driving behavior data such as steering wheel correction frequency, vehicle lateral offset, and speed changes in real time through the vehicle bus, and combines this with navigation system data to determine the remaining travel time, achieving continuous collection of multi-source contextual data. Subsequently, the vehicle control system generates a contextual state vector for the current moment based on the collected heterogeneous data. This contextual state vector keyly includes a fatigue level index (fatigue_level) and a fatigue trend (fatigue_trend). In the initial stage, if the detected contextual state vector shows that the fatigue level is at a moderate level and the fatigue trend is slowly increasing, the vehicle control system will continue monitoring.

[0213] When the state change assessment module detects that the fatigue level has reached a preset first threshold, the evolution strategy generation module generates a first-stage functional strategy. This first-stage functional strategy aims to alleviate fatigue through short-term wake-up, specifically including functions such as increasing the in-vehicle audio rhythm and triggering slight seat vibrations, and outputs the above strategy in the form of action description information. After the first-stage strategy has been executed for a period of time, the vehicle control system continues to collect context data and update the state vector. If the fatigue level index is not significantly reduced and the fatigue trend continues to rise, it indicates that the original strategy is not effective enough, and the vehicle control system will determine that strategy evolution needs to be triggered. The evolution strategy generation module adds new functional action descriptions while retaining some effective functional actions in the original strategy, such as recommending the nearest service area or rest area and triggering navigation prompts, while increasing the intensity of visual reminders, thereby guiding the driver to rest. The evolved functional strategy is executed through the execution interface module, and the execution results and the driver's response will be fed back to the context acquisition module, forming a continuous closed loop, providing a basis for subsequent strategy evolution.

[0214] Example 2 focuses on illustrating the flexibility and independence of the functional strategy execution method in this application. This example 2 aims to demonstrate that the functional evolution mechanism in continuous context described in this application does not depend on a specific functional semantic orchestration language, but can be implemented through other general execution methods. Example 2 is as follows: the evolution strategy generation module generates action description information containing multiple functional actions based on changes in the context state. This action description information represents the function type, action parameters, and execution order in the form of structured data. Subsequently, this action description information is directly input to the predefined action mapping module. This predefined action mapping module directly maps the action descriptions to corresponding vehicle control interface calls, such as audio control interfaces, air conditioning control interfaces, or navigation system interfaces, through fixed mapping relationships. The vehicle control module executes the corresponding control commands sequentially or in parallel according to the mapping results.

[0215] In this process, the vehicle control system does not use a functional semantic orchestration language for complex parsing or scheduling of actions. Instead, it completes execution through direct mapping relationships, thus verifying that the continuous context evolution mechanism of this application can exist independently of a specific execution language or framework. After execution, the vehicle control system feeds back the execution status and effect to the context state update module for subsequent state evaluation and strategy evolution judgment. As can be seen from the above embodiments, this application can not only realize the dynamic evolution of functional strategies through continuous state awareness, but also provide diverse implementation paths by decoupling strategy generation and specific execution methods, thereby adapting to the architectural requirements of different vehicle control systems.

[0216] According to an embodiment of this application, a vehicle control device is provided. It should be noted that the device can be used to execute the above-described vehicle control method.

[0217] Figure 4 This is a structural block diagram of a vehicle control device according to an embodiment of this application. Figure 4 The vehicle control device 400 includes: a generation module 401, used to generate a context state vector based on the vehicle's context information, wherein the context information represents the vehicle's driving state and occupant state during driving; a determination module 402, used to determine a historical state vector sequence based on the context state vector; an execution module 403, used to perform a strategy evolution operation based on the context state vector and the historical state vector sequence in response to the context state vector and the historical state vector sequence satisfying a preset trigger condition, to obtain action description information, wherein the action description information describes the evolved vehicle function strategy; and a control module 404, used to generate vehicle control commands based on the action description information, and to control the vehicle based on the vehicle control commands.

[0218] Figure 5 This is a structural block diagram of a vehicle according to an embodiment of this application. For example... Figure 5 As shown, embodiments of this application also provide a vehicle 500, including: a memory 501 storing an executable program; and a processor 502 for running the program, wherein the program executes the methods in various embodiments of this application during runtime.

[0219] Embodiments of this application also provide a computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0220] Embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0221] Embodiments of this application also provide a computer program product, including a non-volatile computer-readable storage medium for storing a computer program that, when executed by a processor, implements the methods in various embodiments of this application.

[0222] Embodiments of this application also provide a computer program that, when executed by a processor, implements the methods described in the various embodiments of this application.

[0223] 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.

[0224] 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.

[0225] 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.

[0226] 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.

[0227] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0228] 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 vehicle control method, characterized in that, The method includes: A context state vector is generated based on the vehicle's context information, wherein the context information is used to represent the vehicle's driving state and occupant state during the driving process; Determine the historical state vector sequence based on the context state vector; In response to the context state vector and the historical state vector sequence satisfying a preset triggering condition, a strategy evolution operation is performed based on the context state vector and the historical state vector sequence to obtain action description information, wherein the action description information is used to describe the evolved vehicle function strategy; The system generates vehicle control commands based on the action description information and controls the vehicle based on the vehicle control commands.

2. The method according to claim 1, characterized in that, The method further includes: Obtain the initial context information of the vehicle, wherein the initial context information includes at least one of the following: driving behavior data, driving object state data, vehicle condition data, road condition data, human-machine interaction data, and trip data; The initial context information is timestamped to obtain the context information.

3. The method according to claim 1, characterized in that, The generation of the context state vector based on vehicle context information includes: The context information is feature-mapped to obtain context state features, wherein the context state features include at least one of the following: fatigue level features, fatigue change trend features, driving stress features, traffic environment impact features, environmental comfort features, and remaining travel time features. The context state features are normalized to obtain the context state vector, wherein the context state vector carries the current timestamp and the target function policy identifier, and the target function policy identifier is used to index the vehicle function policy adopted when the policy evolution operation was most recently executed.

4. The method according to claim 3, characterized in that, Each of the aforementioned context state features carries a confidence level.

5. The method according to claim 1, characterized in that, Determining the historical state vector sequence based on the context state vector includes: Based on the context state vector and a preset sliding time window, the historical state vector sequence is determined, wherein the historical state vector sequence includes multiple context state vectors arranged in chronological order.

6. The method according to claim 1, characterized in that, The preset triggering conditions include at least one of the following: The target state feature in the context state vector is greater than the corresponding state feature threshold; The rate of change of the target state feature is greater than a preset deterioration threshold, wherein the rate of change of the target state feature is used to represent the direction and magnitude of change of the target state feature within a preset sliding time window; The difference between the current value of the target state feature and the preset baseline value is greater than the preset deviation threshold; The state difference between the context state vector and the target state snapshot is greater than the corresponding state difference threshold, wherein the target state snapshot is the context state vector recorded during the most recent execution of the policy evolution operation; The first time difference between the current timestamp and the target evaluation timestamp is greater than the preset evaluation time interval, wherein the target evaluation timestamp is the timestamp of the most recent execution of the evolution-triggered evaluation operation, and the evolution-triggered evaluation operation is used to determine whether the context state vector and the historical state vector sequence satisfy the preset triggering condition.

7. The method according to claim 1, characterized in that, The step of performing a policy evolution operation based on the context state vector and the historical state vector sequence to obtain action description information includes: The target state feature change rate is determined based on the context state vector and the historical state vector sequence; The adaptation value of the policy index of the indicator function policy identifier is determined based on the target state feature change rate, wherein the indicator function policy identifier is used to index the current vehicle function policy, or the vehicle function policy adopted when the policy evolution operation was performed most recently, and the adaptation value is used to indicate the degree of adaptation between the policy indexed by the indicator function policy identifier and the current driving scenario. The evolution type is determined based on the fitness value; The policy evolution operation is performed based on the evolution type to obtain the action description information.

8. The method according to claim 7, characterized in that, The evolution type is one of the following: maintaining the vehicle function policy of the indicator function policy identifier index, adjusting the parameters of the vehicle function policy of the indicator function policy identifier index, or adding a new execution action to the vehicle function policy of the indicator function policy identifier index.

9. The method according to claim 1, characterized in that, The action description information includes at least one of the following: action type, action parameters, triggering conditions for executing the action, and the order of execution of the actions.

10. The method according to any one of claims 1-9, characterized in that, The method further includes: In response to the context state vector and the historical state vector sequence satisfying the preset triggering condition, a second time difference between the current timestamp and the target trigger timestamp is calculated, wherein the target trigger timestamp is the timestamp of the most recent execution of the policy evolution operation; In response to the second time difference being less than a preset cooling time interval, and the evolution type of the strategy evolution operation being the target evolution type, the execution of the strategy evolution operation is delayed, wherein the target evolution type is the evolution type of the most recently executed strategy evolution operation.

11. The method according to any one of claims 1-9, characterized in that, The method further includes: In response to the detection of a driving intervention, the execution of the policy evolution operation is delayed.

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

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program is configured to execute the vehicle control method according to any one of claims 1 to 11 when run on a computer or processor.

Images