Vehicle control method and device and automobile

Abstract

The invention relates to a vehicle control method and device and an automobile. The method comprises the following steps: acquiring multi-modal data of a seat of an automobile and a driver and passengers on the seat; based on the multi-modal data, first evaluation information of the seat and second evaluation information of the driver and passengers are obtained; obtaining a current seat comfort value of the seat according to the first evaluation information and the second evaluation information; according to the current seat comfort value and the target seat comfort value of the seat, current parameters of the seat component associated with the first evaluation information and the second evaluation information are adjusted, and to-be-adjusted parameters of the seat component are obtained; according to the to-be-adjusted parameters, a seat adjusting instruction is generated; the adjusting instruction is used for indicating to adjust the seat component based on the to-be-adjusted parameter. By adopting the method, the active control effect on the vehicle seat can be improved, so that the comfort of the seat is improved, and the driving safety is also improved.

Description

Technical Field

[0001] This application relates to the field of vehicle control technology, and in particular to a vehicle control method, device and automobile. Background Technology

[0002] Vehicle control technology, as a core interaction technology between drivers and passengers and the vehicle, directly affects the comfort and safety of drivers and passengers. For example, the control technology for vehicle seats is one of the vehicle control technologies that drivers and passengers frequently use when driving.

[0003] Current vehicle seat control technology mainly employs manual adjustment or simple electric adjustment schemes, with some high-end models equipped with memory functions that can store a limited number of fixed seat position parameters. However, these adjustment methods primarily rely on passive adjustment mechanisms triggered manually by the driver and passengers, lacking awareness of the driver's and passengers' status and active seat adjustment, resulting in poor vehicle seat control performance. Summary of the Invention

[0004] Based on this, this application addresses the aforementioned technical problems by providing a vehicle control method, device, automobile, computer-readable storage medium, and computer program product that can at least improve the vehicle seat control effect.

[0005] In a first aspect, this application provides a vehicle control method, including:

[0006] Acquire multimodal data of the car seats and the occupants in the seats;

[0007] Based on the multimodal data, a first evaluation information of the seat and a second evaluation information of the driver and passenger are obtained; the first evaluation information includes at least pressure distribution evaluation information and comfort evaluation information; the second evaluation information includes at least muscle fatigue evaluation information and driving posture evaluation information.

[0008] Based on the first evaluation information and the second evaluation information, the current seat comfort value of the seat is obtained;

[0009] Based on the current seat comfort value and the target seat comfort value, the current parameters of the seat components associated with the first evaluation information and the second evaluation information are adjusted to obtain the parameters of the seat components to be adjusted.

[0010] An adjustment command for the seat is generated based on the parameter to be adjusted; the adjustment command is used to instruct the seat components to be adjusted based on the parameter to be adjusted.

[0011] The aforementioned vehicle control method acquires multimodal data of the car seat and occupants, and based on this data, obtains first evaluation information covering seat pressure distribution and comfort, as well as second evaluation information including occupant muscle fatigue and driving posture. This achieves integrated perception and evaluation from four dimensions: pressure, comfort, physiology, and posture. Furthermore, it calculates the current seat comfort value by combining the first and second evaluation information, and compares it with a preset target seat comfort value. Based on the comfort value difference, it intelligently decides the parameters to be adjusted for the associated seat components, and generates adjustment commands accordingly. Ultimately, it drives the seat components to perform adjustment actions. By transforming subjective comfort feelings into quantifiable objective comfort value indicators, it achieves dynamic adaptive adjustment of parameters such as seat support and posture, effectively improving seat comfort while optimizing occupant driving posture, alleviating occupant muscle fatigue, and improving driving safety.

[0012] In an optional embodiment of the first aspect, after acquiring multimodal data of the vehicle seats and the occupants seated in the seats, the method further includes:

[0013] The multimodal data, the seat, and the historical multimodal data of the driver and passengers are input into a pre-trained human state prediction model to obtain the predicted second evaluation information of the driver and passengers in the first time period in the future.

[0014] Based on the predicted second evaluation information, fatigue identification processing is performed on the driver and passenger to obtain the predicted fatigue state of the driver and passenger in the future first time period.

[0015] If the predicted fatigue state is detected to meet the preset fatigue driving conditions, then the first safety parameter of the seat is obtained based on the predicted second evaluation information.

[0016] Based on the first safety parameter, a first safety adjustment command for the seat is generated; the first safety adjustment command is used to instruct the adjustment of the vibration element of the seat to be completed before the arrival of the first future time period based on the first safety parameter.

[0017] In this embodiment, by inputting real-time multimodal data of the seat and the occupants, as well as historical multimodal data, into a pre-trained human state prediction model, the model can accurately obtain the second evaluation information of the occupants in the first time period in the future, thus achieving a forward-looking prediction of the occupants' state. Based on this second evaluation information, fatigue identification processing can be performed to determine the predicted fatigue state of the occupants in the first time period in the future, effectively improving the predictability and reliability of fatigue detection. When the predicted fatigue state is detected to meet the preset fatigue driving conditions, the first safety parameter of the seat can be automatically determined based on the second evaluation information. This allows for early warning of driver fatigue and proactive intervention and adjustment of seat parameters by adjusting the seat's vibration elements in advance, avoiding safety hazards caused by driver fatigue and improving the safety of the occupants while driving and the level of intelligent adjustment of the car seat.

[0018] In an optional embodiment of the first aspect, after acquiring multimodal data of the vehicle seats and the occupants seated in the seats, the method further includes:

[0019] The multimodal data, the seat, and the historical multimodal data of the driver and passengers are input into a pre-trained seat state prediction model to obtain the first evaluation information of the seat in the second future time period.

[0020] Based on the predicted first evaluation information, the driver and passenger are subjected to driving state identification processing to obtain the predicted driving state of the driver and passenger in the future second time period.

[0021] If the predicted driving state is detected to meet the preset aggressive driving conditions, then the second safety parameter of the seat is obtained based on the predicted first evaluation information;

[0022] Based on the second safety parameter, a second safety adjustment command for the seat is generated; the second safety adjustment command is used to instruct the adjustment of the vibration element of the seat to be completed before the arrival of the future second time period based on the second safety parameter.

[0023] In this embodiment, based on multimodal data and historical multimodal data, the first evaluation information of the seat in the second time period in the future is predicted, and the driving state of the driver and passengers in the second time period in the future is predicted based on the predicted first evaluation information. When the conditions for aggressive driving are met, the second safety parameters of the seat are calculated, and the corresponding second safety adjustment command is generated so that the seat components such as vibration elements are adjusted in advance before the arrival of the second time period in the future. The seat adjustment is elevated from a passive response to an active prediction level. By pre-optimizing the seat support and vibration response before the occurrence of aggressive driving behavior, the physical stability and driving safety of the driver and passengers in aggressive driving conditions such as emergency lane changes and sharp turns are effectively improved, providing a forward-looking technical solution for the active safety protection of car seats.

[0024] In an optional embodiment of the first aspect, obtaining the current seat comfort value of the seat based on the first evaluation information and the second evaluation information includes:

[0025] Based on the current driving condition data of the vehicle and the historical adjustment data of the seat, the first weight corresponding to the pressure distribution evaluation information, the second weight corresponding to the comfort evaluation information, the third weight corresponding to the muscle fatigue evaluation information, and the fourth weight corresponding to the driving posture evaluation information are determined.

[0026] Based on the first weight, the second weight, the third weight, and the fourth weight, the pressure distribution evaluation information, the comfort evaluation information, the muscle fatigue evaluation information, and the driving posture evaluation information are fused to obtain the current seat comfort value of the seat.

[0027] In this embodiment, based on the current driving conditions of the vehicle and the historical adjustment data of the seat, the weights of the pressure distribution evaluation information, comfort evaluation information, muscle fatigue evaluation information, and driving posture evaluation information are adaptively determined to better suit the usage habits and actual driving scenarios of different drivers and passengers. Based on the above weights, the current seat comfort value is calculated. By utilizing pressure distribution, comfort, muscle fatigue, and driving posture, subjective comfort feelings are transformed into quantifiable objective comfort value indicators, providing a scientific and reliable basis for subsequent adjustments to seat components.

[0028] In an optional embodiment of the first aspect, the current parameters of the seat component associated with the first evaluation information and the second evaluation information are adjusted based on the current seat comfort value and the target seat comfort value of the seat to obtain the parameters of the seat component to be adjusted, including:

[0029] Based on the first weight, the second weight, the third weight, the fourth weight, and the target seat comfort value, the target pressure distribution evaluation information corresponding to the pressure distribution evaluation information, the target comfort evaluation information corresponding to the comfort evaluation information, the target muscle fatigue evaluation information corresponding to the muscle fatigue evaluation information, and the target driving posture evaluation information corresponding to the driving posture evaluation information are determined.

[0030] Based on the target pressure distribution evaluation information, the current parameters of the first seat component associated with the pressure distribution evaluation information are optimized to obtain the parameters to be adjusted for the first seat component.

[0031] Based on the target comfort evaluation information, the current parameters of the second seat component associated with the comfort evaluation information are optimized to obtain the parameters to be adjusted for the second seat component.

[0032] Based on the target muscle fatigue evaluation information, the current parameters of the third seat component associated with the muscle fatigue evaluation information are optimized to obtain the parameters to be adjusted for the third seat component.

[0033] Based on the target driving posture evaluation information, the current parameters of the fourth seat component associated with the driving posture evaluation information are optimized to obtain the parameters to be adjusted for the fourth seat component.

[0034] In this embodiment, based on the first to fourth weights and the target seat comfort value, the target evaluation information corresponding to the four evaluation indicators of pressure distribution, comfort, muscle fatigue, and driving posture is determined respectively. Then, for the seat components associated with each evaluation indicator, parameter optimization is performed independently to generate the corresponding adjustable parameters for each seat component, realizing the precise disassembly of seat components that affect the seat comfort value. Through component-level parameter decoupling optimization, the problem of neglecting one aspect due to parameter coupling in the traditional overall parameter adjustment method is avoided. This allows different seat components to independently find the optimal adjustable parameters under their respective optimization objectives, providing a flexible, precise, and interpretable seat adjustment strategy for automobile seats. This fully meets the personalized comfort needs of different users in different scenarios, effectively improves the accuracy of seat adjustment, and enhances the driving comfort and driving safety of the seats.

[0035] In an optional embodiment of the first aspect, obtaining first evaluation information of the seat and second evaluation information of the driver / passenger based on the multimodal data includes:

[0036] The multimodal data is input into a pre-trained comfort evaluation model to obtain the comfort evaluation information of the seat; the comfort evaluation model is trained through supervised learning and reinforcement learning.

[0037] Based on the target pressure data in the multimodal data, the pressure distribution evaluation information of the seat is obtained;

[0038] The multimodal data is input into a pre-trained muscle fatigue evaluation model to obtain the muscle fatigue evaluation information of the driver and passengers.

[0039] The multimodal data is input into a pre-trained driving posture evaluation model to obtain the driving posture evaluation information of the driver and passengers.

[0040] In this embodiment, by inputting multimodal data into a pre-trained comfort evaluation model, seat comfort evaluation information can be intelligently output. Based on the target pressure data in the multimodal data, pressure distribution evaluation information can be accurately obtained. At the same time, by inputting multimodal data into a pre-trained muscle fatigue evaluation model and a pre-trained driving posture evaluation model, the muscle fatigue evaluation information and driving posture evaluation information of the driver and passengers can be accurately identified. This achieves a multi-dimensional comprehensive evaluation of seat comfort and pressure distribution, as well as the muscle fatigue and driving posture of the driver and passengers, providing a reliable processing basis for subsequent adjustment of seat parameters.

[0041] In an alternative embodiment of the first aspect, acquiring multimodal data of the vehicle's seats and occupants seated in the seats includes:

[0042] Acquire raw data on the seat and the occupants seated therein in terms of vision, touch, hearing, and physiology;

[0043] The various original data are preprocessed to obtain various preprocessed data of the seat and the driver / passenger;

[0044] The various preprocessed data are spatiotemporally synchronized to obtain various synchronized data of the seat and the driver / passenger.

[0045] The various synchronized data are input into a multimodal fusion model to obtain multimodal data of the seat and the driver / passenger.

[0046] In this embodiment, raw data from seats and occupants in multiple dimensions, including visual, tactile, auditory, and physiological aspects, are collected and then preprocessed, spatiotemporally synchronized, and fused using a multimodal model. Spatiotemporal synchronization processing enables the standardization and temporal alignment of multi-source data, ensuring consistency of data across different dimensions in time and space, and improving data reliability and the accuracy of subsequent analysis. The multimodal fusion model integrates the synchronized data to form unified and efficient multimodal data, providing a high-quality data foundation for subsequent evaluations of seats and occupants.

[0047] In an optional embodiment of the first aspect, the raw data includes at least video data, posture data, stress data, audio data, and physiological data;

[0048] The acquisition of raw data on the seat and the occupants seated therein, including visual, tactile, auditory, and physiological data, includes:

[0049] Acquire video data of the driver and passenger sitting in the seat, perform human posture detection processing on the video data, and obtain the posture data of the driver and passenger in the visual aspect;

[0050] The tactile pressure data is obtained based on the captured information about the change in resistance of the conductive material associated with the seat caused by the physical deformation of the seat.

[0051] The audio data related to hearing is obtained based on the captured voice signals and ambient sound signals of the car's interior environment.

[0052] The physiological data related to the physiological aspects are obtained based on the captured physiological signals of the drivers and passengers.

[0053] In this embodiment, by simultaneously collecting raw data from multiple dimensions, including video, posture, pressure, audio, and physiological data, a comprehensive and accurate perception of the state of drivers and passengers and the in-vehicle environment is achieved. This avoids the one-sidedness of single-data collection and can truly reflect the state of the driving and riding scenario. Human posture detection through video data can accurately obtain the posture data of drivers and passengers, providing a visual basis for posture analysis. Pressure data obtained through the resistance value change information of the seat's conductive material can reflect the seat deformation and the contact pressure of drivers and passengers in real time, closely matching actual driving and riding interactions. Audio data obtained through in-vehicle voice and ambient sound signals enriches the dimensions of scene perception. Physiological data obtained through the physiological signals of drivers and passengers can intuitively reflect the physical functional state of drivers and passengers. By using raw data from multiple sources, the driving and riding scenario can be comprehensively and accurately depicted, providing reliable data support for subsequent evaluation of seats and drivers and passengers.

[0054] Secondly, this application also provides a vehicle control device, comprising:

[0055] The data acquisition module is used to acquire multimodal data of the car seats and the occupants sitting in the seats;

[0056] The information evaluation module is used to obtain first evaluation information of the seat and second evaluation information of the driver and passenger based on the multimodal data; the first evaluation information includes at least pressure distribution evaluation information and comfort evaluation information; the second evaluation information includes at least muscle fatigue evaluation information and driving posture evaluation information.

[0057] The comfort evaluation module is used to obtain the current seat comfort value of the seat based on the first evaluation information and the second evaluation information;

[0058] The parameter optimization module is used to adjust the current parameters of the seat components associated with the first evaluation information and the second evaluation information based on the current seat comfort value and the target seat comfort value of the seat, so as to obtain the parameters of the seat components to be adjusted.

[0059] A seat adjustment module is used to generate an adjustment command for the seat based on the parameter to be adjusted; the adjustment command is used to instruct the seat components to be adjusted based on the parameter to be adjusted.

[0060] Thirdly, this application also provides an automobile, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the method described above.

[0061] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0062] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in any of the above aspects.

[0063] Regarding the beneficial effects of any of the technical solutions in the second to fifth aspects mentioned above, refer to the beneficial effects of the corresponding technical solutions in the first aspect; repeated examples will not be listed here. Attached Figure Description

[0064] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0065] Figure 1 This is a schematic diagram of an optional flow of a vehicle control method in one embodiment;

[0066] Figure 2 This is a schematic diagram of the logic architecture for adjusting a car seat in one embodiment;

[0067] Figure 3 This is an optional flowchart illustrating one embodiment of the step of adjusting the parameters of the hardware unit corresponding to the seat before the arrival of the first time period in the future;

[0068] Figure 4 This is a schematic diagram of another optional process for the vehicle control method in one embodiment;

[0069] Figure 5 This is a schematic diagram of another optional process for the vehicle control method in one embodiment;

[0070] Figure 6 This is a schematic diagram of an optional structure of the vehicle control device in one embodiment;

[0071] Figure 7 This is a schematic diagram of an optional internal structure of a car in one embodiment. Detailed Implementation

[0072] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application.

[0073] The terms "first," "second," etc., used in this application may be used to describe various elements, but these elements are not limited by these terms. These terms are used only to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0074] In one embodiment, such as Figure 1 As shown, a vehicle control method is provided. This embodiment illustrates the application of this method to a vehicle's processor. The vehicle includes, but is not limited to, various types of gasoline-powered vehicles, electric vehicles, hybrid vehicles, range-extended vehicles, and hydrogen fuel cell vehicles. In this embodiment, the method includes the following steps:

[0075] Step S101: Obtain multimodal data of the car seats and the occupants sitting in the seats.

[0076] Multimodal data refers to various types of data from different sources and in different formats, but all of which describe the seating experience of drivers and passengers.

[0077] Figure 2 A schematic diagram illustrating the logical architecture for adjusting car seats. (For example...) Figure 2As shown, the logical architecture for adjusting the car seat consists of a perception layer, a decision layer, and an execution layer. Data interaction between each layer is achieved through in-vehicle Ethernet and CAN (Controller Area Network) bus, forming a complete perception-decision-execution closed loop. The total data bandwidth requirement is 64Mbps, and the average control delay is ≤100ms, which meets the real-time requirements.

[0078] The perception layer is used to provide perception functions such as visual perception, physiological perception, pressure perception, and posture perception.

[0079] Visual perception can be achieved using various types of cameras, millimeter-wave radar, and other devices to capture the human posture of drivers and passengers. Cameras include DMS (Driver Monitoring System) cameras and OMS (Occupant Monitoring System) cameras. DMS cameras have a 1920×1080 resolution, are equipped with infrared illumination, support all-weather day / night operation, and have a basic acquisition frequency of 30fps. OMS cameras also have a 1920×1080 resolution and are primarily used to detect passenger body shape, posture, and limb position. Millimeter-wave radar can achieve 3D posture detection without acquiring image data, offering better protection for user privacy than cameras. However, the posture recognition accuracy of the 3D point cloud captured by millimeter-wave radar is approximately 15%-20% lower than that of images captured by cameras.

[0080] Physiological sensing can be accomplished using various types of sensors. For example, sensors used for physiological sensing include heart rate sensors, electromyography (EMG) sensors, and body temperature sensors.

[0081] Pressure sensing can also be accomplished using various types of sensors. For example, sensors used for pressure sensing include sensor arrays that detect seat pressure and sensors that detect contact area. When constructing a sensor array for detecting seat pressure, capacitive thin-film sensors can be used instead of piezoresistive sensors. Capacitive thin-film sensors have better flexibility and durability (lifespan > 1 million cycles), but they have a larger temperature drift (±1% / ℃), requiring the addition of a temperature compensation algorithm.

[0082] Attitude perception can also be accomplished using various types of sensors. For example, sensors used for attitude perception include accelerometers, angle sensors, and position sensors.

[0083] The decision layer is used to provide data processing functions such as data fusion (e.g., preprocessing, feature extraction, and spatiotemporal synchronization) and intelligent decision-making (e.g., stress distribution evaluation, muscle fatigue evaluation, comfort evaluation, and driving posture evaluation).

[0084] The execution layer is used to provide functions involving the operation of automotive hardware, such as seat adjustment, environmental adjustment, and safety intervention.

[0085] For example, raw data on the pressure of the car seat is collected using various micro sensors; raw visual data of the occupants in the seat is collected using DMS and OMS cameras; raw physiological data of the occupants is collected using various micro sensors; and then the raw data is transmitted to the processor.

[0086] Step S102: Based on multimodal data, obtain first evaluation information of the seat and second evaluation information of the driver and passengers; the first evaluation information includes at least pressure distribution evaluation information and comfort evaluation information; the second evaluation information includes at least muscle fatigue evaluation information and driving posture evaluation information.

[0087] The first evaluation information refers to the evaluation indicators used to evaluate the condition of the seat. The second evaluation information refers to the evaluation indicators used to evaluate the driving condition of the driver and passengers.

[0088] For example, the processor can input multimodal data into the model (including a comfort evaluation model, a muscle fatigue evaluation model, and a driving posture evaluation model), and obtain first evaluation information for the seat and second evaluation information for the driver and passengers through the model output. In practical applications, the model can be built based on a large model or deep learning algorithm, and then trained using a combination of supervised learning and reinforcement learning. In the supervised learning phase, a dataset labeled by professional reviewers is used as the training set for the model, and in the reinforcement learning phase, user feedback and improvements in physiological data are used as reward signals for the model.

[0089] Step S103: Obtain the current seat comfort value based on the first evaluation information and the second evaluation information.

[0090] The current seat comfort value refers to a numerical value that describes the comfort level of the car seat at the current time period.

[0091] For example, by combining the first evaluation information and the second evaluation information, the seat status and human sensation are comprehensively characterized, and then the current seat comfort value is calculated.

[0092] Step S104: Based on the current seat comfort value and the target seat comfort value, adjust the current parameters of the seat components associated with the first evaluation information and the second evaluation information to obtain the parameters to be adjusted for the seat components.

[0093] The target seat comfort value refers to a parameter set for the comfort level of the seat. For example, the target seat comfort value can be used to set the firmness or softness of the seat. In practical applications, the target seat comfort value can be set between 0.5 and 2; the smaller the target seat comfort value, the softer the seat, and the larger the target seat comfort value, the firmer the seat.

[0094] Among them, the parameter to be adjusted refers to the parameter value that needs to be adjusted for the corresponding seat index.

[0095] For example, a target seat comfort value preset for seat comfort is obtained; based on the difference between the current seat comfort value and the target seat comfort value, the current parameters of the seat components associated with the first evaluation information and the second evaluation information are optimized to obtain the parameters to be adjusted for the seat components, so that the seat components can achieve the comfort requirements of the target seat comfort value after being adjusted according to the parameters to be adjusted.

[0096] Step S105: Generate adjustment instructions for the seat based on the parameters to be adjusted; the adjustment instructions are used to instruct the seat components to be adjusted based on the parameters to be adjusted.

[0097] For example, the processor generates an adjustment instruction for the seat component based on the parameter to be adjusted, and sends the adjustment instruction to the controller of the seat component; the controller controls the relevant seat components (such as airbags, temperature control elements, vibration elements, etc.) to adjust the current parameter according to the parameter to be adjusted in the adjustment instruction.

[0098] In practical applications, by continuously monitoring the spinal pressure distribution and muscle status of drivers and passengers and proactively adjusting the seat posture, chronic injuries caused by prolonged poor posture can be avoided. The processor has built-in recognition of various common poor posture patterns (such as scoliosis, pelvic tilt, and forward head tilt). After detection, the processor proactively makes micro-adjustments to correct the seat and provides healthy driving suggestions to drivers and passengers, realizing human-computer interaction and proactive services for car seats.

[0099] In the aforementioned vehicle control method, multimodal data of the car seat and occupants are acquired. Based on this multimodal data, a first evaluation information covering seat pressure distribution and comfort, and a second evaluation information including occupant muscle fatigue and driving posture are obtained, achieving integrated perception and evaluation from four dimensions: pressure, comfort, physiology, and posture. Furthermore, the current seat comfort value is calculated by combining the first and second evaluation information and compared with a preset target seat comfort value. Based on the comfort value difference, intelligent decisions are made regarding the adjustment parameters of the associated seat components, and adjustment commands are generated accordingly. Ultimately, the seat components are driven to perform adjustment actions. By transforming subjective comfort feelings into quantifiable objective comfort value indicators, dynamic adaptive adjustment of parameters such as seat support and posture is achieved, effectively improving seat comfort. Simultaneously, it optimizes the occupant's driving posture, alleviates occupant muscle fatigue, and improves driving safety.

[0100] In one exemplary embodiment, such as Figure 3 As shown, step S101 above, after acquiring the multimodal data of the car seat and the occupants in the seat, further includes steps S301 to S304. Wherein:

[0101] Step S301: Input the multimodal data, the seat and the historical multimodal data of the driver and passengers into the pre-trained human state prediction model to obtain the second evaluation information of the driver and passengers in the first time period in the future.

[0102] Among them, the human state prediction model refers to a model that predicts the second evaluation information of drivers and passengers in the first time period in the future (i.e., predicting the second evaluation information) based on input data (such as multimodal data and historical multimodal data). The model is constructed using deep learning algorithms (such as the publicly available LSTM (Long Short-Term Memory) network), large models, or fuzzy logic to build a base model to be trained. Sample multimodal data is used as the training set, and the labeled results of the second evaluation information of the sample multimodal data in the first time period in the future are used as labels. Supervised learning and reinforcement learning are used to train the base model, resulting in a pre-trained human state prediction model.

[0103] Traditional seat adjustment mechanisms employ a passive response approach, relying solely on manual adjustments by the driver or passengers, or only adjusting after noticeable discomfort arises. This leads to a continuous accumulation of discomfort and delayed adjustments. This application proactively predicts and intervenes in advance, adjusting the seat's vibration elements and related components. For example, historical multimodal data of the seat and the driver / passenger are input into a pre-trained human state prediction model to predict changes in the driver / passenger's state in a future timeframe (e.g., 5-10 seconds later), such as whether muscle fatigue is gradually accumulating or whether driving posture is changing, outputting a predicted second evaluation.

[0104] The processor can also input historical multimodal data of the seat and the occupants into a pre-trained seat state prediction model, and output the first evaluation information of the seat in the first time period in the future.

[0105] Step S302: Based on the predicted second evaluation information, fatigue identification processing is performed on the driver and passengers to obtain the predicted fatigue state of the driver and passengers in the first time period in the future.

[0106] For example, the processor utilizes a Driver Monitoring System (DMS) to perform fatigue recognition processing on the predicted second evaluation information (including predicted muscle fatigue evaluation information and predicted driving posture evaluation information). For instance, it can use keypoint detection, posture recognition, and state determination algorithms to extract features such as the driver's eye state, head posture, blinking frequency, duration of eye closure, and mouth state in real time for a future first time period. If fatigue characteristics such as prolonged eye closure, frequent nodding, head drooping, yawning, and prolonged gaze deviation are identified based on these features, it can be determined that the driver may be in a state of fatigued driving in the future first time period, and the predicted fatigue state of the driver in the future first time period is output. The predicted fatigue state can be represented by a fatigue level.

[0107] Step S303: If the predicted fatigue state is detected to meet the preset fatigue driving conditions, the first safety parameter of the seat is obtained based on the predicted second evaluation information.

[0108] The first safety parameter refers to the seat adjustment parameters used to improve the driving safety of drivers and passengers.

[0109] For example, if the predicted fatigue state is detected to meet the preset fatigue driving conditions, such as the fatigue level corresponding to the predicted fatigue state being greater than or equal to the standard fatigue level corresponding to the fatigue driving conditions, then the first safety parameter of the seat is obtained based on the predicted second evaluation information and the predicted first evaluation information. It can be understood that the method of obtaining the first safety parameter can be the same as the method of obtaining the parameter to be adjusted in step S104 above, and will not be described again here.

[0110] In addition, in emergency braking situations, seat parameters can be adjusted in advance to optimize seat adjustment.

[0111] Step S304: Generate a first safety adjustment command for the seat based on the first safety parameter; the first safety adjustment command is used to instruct the seat's vibration element adjustment to be completed before the arrival of a first time period in the future, based on the first safety parameter.

[0112] For example, a first safety adjustment command for the seat is generated based on a first safety parameter, and this command is sent to the seat's controller. The controller, according to the first safety parameter carried in the first safety adjustment command, pre-emptively adjusts at least the seat's vibration element before the arrival of a future first time period. Alternatively, it can pre-emptively adjust other related seat components (such as airbags, temperature control elements, or voice control elements) before the arrival of the future first time period. For instance, by controlling the seat's vibration element to vibrate using the first safety parameter, the seat's voice control element can also be controlled to issue an alarm prompt.

[0113] In practical applications, the adaptive adjustment function of the seat reduces the frequency of manual seat operation by drivers and passengers (for example, the number of times drivers and passengers manually operate the seat can be reduced from an average of 8 times per hour to 2 times per hour), and the adjustment time is also shortened (for example, from 5-8 seconds to 1-2 seconds), significantly reducing the time spent driving while distracted. At the same time, by detecting and warning of fatigue driving conditions, it can also reduce the risk of accidents related to driver fatigue.

[0114] In this embodiment, by inputting real-time multimodal data of the seat and the occupants, as well as historical multimodal data, into a pre-trained human state prediction model, the model can accurately obtain the second evaluation information of the occupants in the first time period in the future, thus achieving a forward-looking prediction of the occupants' state. Based on this second evaluation information, fatigue identification processing can be performed to determine the predicted fatigue state of the occupants in the first time period in the future, effectively improving the predictability and reliability of fatigue detection. When the predicted fatigue state is detected to meet the preset fatigue driving conditions, the first safety parameter of the seat can be automatically determined based on the second evaluation information. This allows for early warning of driver fatigue by adjusting the seat's vibration elements in advance, avoiding safety hazards caused by driver fatigue and improving the safety of the occupants while driving and the level of intelligent adjustment of the car seat.

[0115] In an exemplary embodiment, after acquiring the multimodal data of the car seat and the occupants in the seat, step S101 further includes: inputting the multimodal data, the historical multimodal data of the seat and the occupants into a pre-trained seat state prediction model to obtain a first evaluation information of the seat in a future second time period; performing driving state recognition processing on the occupants based on the first evaluation information to obtain a predicted driving state of the occupants in the future second time period; if the predicted driving state is detected to meet preset aggressive driving conditions, obtaining a second safety parameter of the seat based on the first evaluation information; generating a second safety adjustment command for the seat based on the second safety parameter; the second safety adjustment command is used to instruct the seat vibration element adjustment to be completed before the arrival of the future second time period based on the second safety parameter.

[0116] The seat condition prediction model refers to a model that predicts the first evaluation information of a seat in a second future time period based on input data (such as multimodal data or historical multimodal data). The base model to be trained for the seat condition prediction model is constructed based on deep learning algorithms (such as neural networks), large models, or fuzzy logic. Sample multimodal data is used as the training set, and the labeled results of the first evaluation information of the sample multimodal data in the second future time period are used as labels. Supervised learning and reinforcement learning are used to train the base model, resulting in a pre-trained seat condition prediction model.

[0117] The second future time period can be exactly the same as the first future time period, or it can be completely different, or it can share some time periods.

[0118] For example, the processor inputs multimodal data, historical multimodal data of the seat and occupants into a pre-trained seat state prediction model to analyze the pressure distribution change trend of the seat in a future second time period and output a first evaluation information prediction of the seat in the future second time period. Similarly, historical multimodal data of the seat and occupants, along with other multimodal data, are input into a pre-trained human state prediction model to output a first evaluation information prediction of the occupants in the future second time period.

[0119] The processor performs driving state recognition processing on the driver and passengers based on the predicted pressure distribution evaluation information in the predicted first evaluation information of the second time period and the predicted driving posture evaluation information in the predicted second time period of the second time period, in order to determine the predicted driving state of the driver and passengers in the second time period of the future. For example, in a state of intense driving (such as high-speed cornering), the human body will be biased to the outside of the center by the centripetal force, which will cause the posture of the driver and passengers to be biased to the outside of the center, and the pressure distribution of the seat will also be greater at the outside of the center.

[0120] If the predicted driving state is detected to match preset aggressive driving conditions, such as the predicted driving posture evaluation information matching any one of the preset aggressive driving states, then the second safety parameter of the seat is obtained based on the first and second predicted evaluation information. It is understood that the method for obtaining the second safety parameter can be the same as the method for obtaining the parameter to be adjusted in step S104 above, and will not be elaborated further here.

[0121] Based on the second safety parameter, a second safety adjustment command for the seat is generated and sent to the seat controller. The controller, according to the second safety adjustment parameter carried in the second safety adjustment command, completes the adjustment of at least the seat's vibration element before the arrival of the second time period. Of course, it can also complete the adjustment of other related seat components (such as airbags, temperature control elements, or voice control elements) before the arrival of the second time period. For example, the second safety parameter can be used to control the seat's vibration element to vibrate, the seat's voice control element can be controlled to issue an alarm prompt, and the seat's airbag can be controlled.

[0122] For example, during aggressive driving, such as when cornering at high speed, the human body is subjected to centripetal force and deflects outward from the center. To counteract this force, the driver and passengers need to rely on their own abdominal and back strength to maintain a normal driving posture. The processor in this application can identify the high-speed cornering (i.e., aggressive driving) in the second time period in advance and inflate the side airbags of the seat in advance before the arrival of the second time period, actively providing sufficient side support for the driver and passengers when cornering, and controlling the vibration element of the car seat to vibrate and issue a warning.

[0123] In this embodiment, based on multimodal data and historical multimodal data, the first evaluation information of the seat in the second time period in the future is predicted, and the driving state of the driver and passengers in the second time period in the future is predicted based on the predicted first evaluation information. When the conditions for aggressive driving are met, the second safety parameters of the seat are calculated, and the corresponding second safety adjustment command is generated so that the seat components such as vibration elements are adjusted in advance before the arrival of the second time period in the future. The seat adjustment is elevated from a passive response to an active prediction level. By pre-optimizing the seat support and vibration response before the occurrence of aggressive driving behavior, the physical stability and driving safety of the driver and passengers in aggressive driving conditions such as emergency lane changes and sharp turns are effectively improved, providing a forward-looking technical solution for the active safety protection of car seats.

[0124] In an exemplary embodiment, step S103 above, obtaining the current seat comfort value of the seat based on the first evaluation information and the second evaluation information, specifically includes the following: determining the first weight corresponding to the pressure distribution evaluation information, the second weight corresponding to the comfort evaluation information, the third weight corresponding to the muscle fatigue evaluation information, and the fourth weight corresponding to the driving posture evaluation information based on the current driving condition data of the vehicle and the historical adjustment data of the seat; and fusing the pressure distribution evaluation information, comfort evaluation information, muscle fatigue evaluation information, and driving posture evaluation information according to the first weight, the second weight, the third weight, and the fourth weight to obtain the current seat comfort value of the seat.

[0125] Among them, the first weight, the second weight, the third weight, and the fourth weight are dynamic weight coefficients that can be adaptively adjusted according to driving scenarios (such as current driving conditions data), user preferences (such as historical manual adjustment data), and user status.

[0126] Historical adjustment data refers to data on manual adjustments made to the seats within a historical time period.

[0127] For example, based on the vehicle's current driving condition data and the seat's historical adjustment data, the weights corresponding to the first evaluation information (e.g., the first and second weights) and the second evaluation information (e.g., the third and fourth weights) are adaptively and dynamically adjusted. Then, based on the first, second, third, and fourth weights, the pressure distribution evaluation information, comfort evaluation information, muscle fatigue evaluation information, and driving posture evaluation information are weighted and summed to obtain the current seat comfort value. The formula for calculating the current seat comfort value is as follows:

[0128] minC(P)=|w1*P_pressure|+|w2*P_fatigue|+|w3*P_comfort|+|w4*P_safety|

[0129] In the formula, minC(P) represents the current seat comfort value; w1 represents the first weight; P_pressure represents the pressure distribution evaluation information; w2 represents the second weight; P_fatigue represents the muscle fatigue evaluation information; w3 represents the third weight; P_comfort represents the comfort evaluation information; w4 represents the fourth weight; and P_safety represents the driving posture evaluation information.

[0130] In this embodiment, based on the current driving conditions of the vehicle and the historical adjustment data of the seat, the weights of the pressure distribution evaluation information, comfort evaluation information, muscle fatigue evaluation information, and driving posture evaluation information are adaptively determined to better suit the usage habits and actual driving scenarios of different drivers and passengers. Based on the above weights, the current seat comfort value is calculated. By utilizing pressure distribution, comfort, muscle fatigue, and driving posture, subjective comfort feelings are transformed into quantifiable objective comfort value indicators, providing a scientific and reliable basis for subsequent adjustments to seat components.

[0131] In an exemplary embodiment, step S104 above, based on the current seat comfort value and the target seat comfort value, adjusts the current parameters of the seat components associated with the first evaluation information and the second evaluation information to obtain the parameters to be adjusted for the seat components, including the following: determining the target pressure distribution evaluation information corresponding to the pressure distribution evaluation information, the target comfort evaluation information corresponding to the comfort evaluation information, the target muscle fatigue evaluation information corresponding to the muscle fatigue evaluation information, and the target driving posture evaluation information corresponding to the driving posture evaluation information based on the first weight, second weight, third weight, fourth weight, and target seat comfort value; and based on the target pressure distribution evaluation information... The parameters of the first seat component associated with the pressure distribution evaluation information are optimized to obtain the parameters to be adjusted for the first seat component. Based on the target comfort evaluation information, the parameters of the second seat component associated with the comfort evaluation information are optimized to obtain the parameters to be adjusted for the second seat component. Based on the target muscle fatigue evaluation information, the parameters of the third seat component associated with the muscle fatigue evaluation information are optimized to obtain the parameters to be adjusted for the third seat component. Based on the target driving posture evaluation information, the parameters of the fourth seat component associated with the driving posture evaluation information are optimized to obtain the parameters to be adjusted for the fourth seat component.

[0132] Among them, the first seat component, the second seat component, the third seat component, and the seat component are completely different, partially the same, or completely the same.

[0133] Among them, such as Figure 2As shown, the seat component includes multiple airbags corresponding to different areas of the passenger's body (e.g., 2 for the waist, 2 for the shoulders, and 4 for the side support, for a total of 8), multiple massage elements corresponding to different areas of the passenger's body, the seat component also includes a seat body, lumbar support, leg support, headrest, and a temperature control element for controlling the seat temperature.

[0134] In practical applications, airbags, seat bodies, lumbar support, leg support, and headrests support both mechanical and electric adjustments. Shape memory alloy actuators can also be used to replace electric adjustments for a more compact design; however, these actuators have a slower response time (from 200ms to 500ms) and higher power consumption. Airbags in different areas can be controlled independently, with air pressure adjustment ranging from 50-300kPa and a response time of <200ms. The seat body, lumbar support, leg support, and headrest support support multiple (e.g., 10-20) parameter adjustments. For example, seat position adjustment parameters include forward / backward ±120mm, height adjustment parameters include ±60mm, angle adjustment parameters include backrest 0-120° and seat cushion ±15°, lumbar support adjustment parameters include height ±30mm and protrusion ±20mm, leg support adjustment parameters include length 0-80mm and angle 0-30°, and headrest adjustment parameters include height ±50mm and angle ±20°. The adjustment parameters of the seat body, lumbar support, leg support, and headrest support are accurate to 1mm or 0.5°.

[0135] The massage elements feature a modular design, including a waist kneading unit (2 rotating massage heads), a back tapping unit (4 tapping units), and a shoulder massage unit (2 movable massage heads), supporting 5 massage modes and 3 intensity levels. The operating noise of the massage elements is <45dB(A) to ensure no disturbance during driving.

[0136] The temperature control element includes a ventilation unit and a heating unit. The ventilation unit can be a micro fan integrated into the seat surface (airflow 5-15 CFM), and the heating unit can be a carbon fiber heating wire (power 20-60W). The temperature adjustment range of the temperature control element is 18-45℃, with a control accuracy of ±1℃. The temperature control element can automatically adjust the ventilation / heating intensity of the ventilation unit and the heating unit according to the ambient temperature and humidity inside the car, as well as physiological data such as the body temperature of the driver and passengers, to achieve thermal comfort control.

[0137] For example, based on each weight (i.e., the first weight, the second weight, the third weight, and the fourth weight) and the target seat comfort value, the target evaluation information (i.e., target pressure distribution evaluation information, target comfort evaluation information, target muscle fatigue evaluation information, and target driving posture evaluation information) corresponding to each evaluation index (i.e., pressure distribution evaluation information, comfort evaluation information, muscle fatigue evaluation information, and driving posture evaluation information) is calculated. For instance, assuming the target driving posture evaluation information is 1, and the first weight, the second weight, the third weight, and the fourth weight are all 0.25, then the target pressure distribution evaluation information, the target comfort evaluation information, the target muscle fatigue evaluation information, and the target driving posture evaluation information are all 1.

[0138] Each evaluation index can also be calculated based on the parameters of its associated seat components. For example, the parameters of the seat components associated with each evaluation index can be weighted and summed to calculate each evaluation index. It is then determined whether each evaluation index equals the corresponding target evaluation information. If not, parameter optimization is performed on the current parameters of the seat components associated with each evaluation index (i.e., the first, second, third, and fourth seat components) to ensure that the updated evaluation index calculated based on the adjusted parameters of the seat components equals its corresponding target evaluation information. Finally, the current seat comfort value obtained after fusing the target evaluation information is equal to the target seat comfort value. In other words, by adjusting the parameters, the current seat comfort value minC(P) is brought back to the target seat comfort value of 1, achieving a balance between seat comfort and safety.

[0139] For example, suppose the processor collects multimodal data from the seat of car number 1 and processes it to obtain P_pressure=0.8 (slightly uneven pressure), P_fatigue=0.6 (mild lower back fatigue), P_comfort=0.7 (moderate comfort), and P_safety=0.9 (basically normal posture). Assuming w1=w2=w3=w4=0.25, then minC(P)=|0.25*0.8|+|0.25*0.6|+|0.25*0.7|+|0.25*0.9|=0.2+0.15+0.175+0.225=0.75. The current seat comfort value of 0.75 is less than the target seat comfort value of 1, indicating that the seat of car number 1 has insufficient support and low comfort. Looking at individual evaluation indicators, the pressure distribution evaluation information, comfort evaluation information, muscle fatigue evaluation information, and driving posture evaluation information are all less than their corresponding target evaluation information of 1. Assuming the first seat component associated with the pressure distribution evaluation information is the lumbar airbag, side airbag, and shoulder airbag, by adjusting the lumbar support airbag of car seat No. 1 to inflate by 20% and the shoulder airbag to inflate by 15%, the updated pressure distribution evaluation information can be made equal to the corresponding target pressure distribution evaluation information 1. Therefore, the parameters to be adjusted are the lumbar support airbag inflation of 20% and the shoulder airbag inflation of 15%.

[0140] To illustrate further, suppose the processor collects multimodal data from the seat of car number 2 and processes it to obtain P_pressure=1.3 (local pressure is too high), P_fatigue=1.1 (slight muscle tension), P_comfort=1.2 (comfort is acceptable), and P_safety=0.5 (posture is too forward, safety is low). Assuming w1=w2=w3=w4=0.25, then minC(P)=|0.25*1.3|+|0.25*1.1|+|0.25*1.2|+|0.25*0.5|= 0.325 + 0.275 + 0.3 + 0.125 = 1.025. The current seat comfort value of 1.025 is greater than the target seat comfort value of 1. Although the current seat comfort value is close to the target seat comfort value, the driving posture evaluation information of 0.5 is too low, significantly lower than its corresponding target driving posture evaluation information of 1. Therefore, the driving posture of the occupants can be improved by adjusting the fourth seat components associated with the driving posture evaluation information (such as side airbags and seat back angle), thereby enhancing driving safety. Assuming that adjusting the side airbag deflation by 10% and the seat back angle adjustment by 5° can bring the driving posture evaluation information back to the target driving posture evaluation information of 1, then the parameters to be adjusted are the side airbag deflation by 10% and the seat back angle adjustment by 5°.

[0141] In this embodiment, based on the first to fourth weights and the target seat comfort value, the target evaluation information corresponding to the four evaluation indicators of pressure distribution, comfort, muscle fatigue, and driving posture is determined respectively. Then, for the seat components associated with each evaluation indicator, parameter optimization is performed independently to generate the corresponding adjustable parameters for each seat component, realizing the precise disassembly of seat components that affect the seat comfort value. Through component-level parameter decoupling optimization, the problem of neglecting one aspect due to parameter coupling in the traditional overall parameter adjustment method is avoided. This allows different seat components to independently find the optimal adjustable parameters under their respective optimization objectives, providing a flexible, precise, and interpretable seat adjustment strategy for automobile seats. This fully meets the personalized comfort needs of different users in different scenarios, effectively improves the accuracy of seat adjustment, and enhances the driving comfort and driving safety of the seats.

[0142] In an exemplary embodiment, step S102 above, which obtains first evaluation information of the seat and second evaluation information of the driver and passenger based on multimodal data, specifically includes the following: inputting multimodal data into a pre-trained comfort evaluation model to obtain comfort evaluation information of the seat; the comfort evaluation model is trained through supervised learning and reinforcement learning; obtaining pressure distribution evaluation information of the seat based on target pressure data in the multimodal data; inputting multimodal data into a pre-trained muscle fatigue evaluation model to obtain muscle fatigue evaluation information of the driver and passenger; and inputting multimodal data into a pre-trained driving posture evaluation model to obtain driving posture evaluation information of the driver and passenger.

[0143] The target pressure data includes pressure distribution entropy and pressure concentration coefficient. Pressure distribution entropy indicates the uniformity of pressure distribution on the seat; a higher entropy indicates a more uniform pressure distribution and a more comfortable seat, while a lower entropy indicates a more concentrated pressure distribution and a less comfortable seat. Pressure concentration coefficient indicates the degree of localized pressure accumulation; a higher coefficient indicates more concentrated localized pressure and a less comfortable seat, while a lower coefficient indicates more dispersed pressure and a more comfortable seat.

[0144] The comfort evaluation model refers to a model that evaluates the comfort of a seat based on input data (such as multimodal data). The base model to be trained for the comfort evaluation model is constructed based on deep learning algorithms (such as neural networks), large models, or fuzzy logic. Sample multimodal data is used as the training set, and the labeled results of comfort evaluation information for the sample multimodal data are used as labels. Supervised learning and reinforcement learning are then used to train the base model, resulting in a pre-trained comfort evaluation model.

[0145] Among them, the muscle fatigue assessment model refers to a model that evaluates the muscle fatigue level of drivers and passengers based on input data (such as multimodal data). The training model is constructed using deep learning algorithms (such as neural networks), large models, or fuzzy logic. Sample multimodal data is used as the training set, and the labeled results of muscle fatigue assessment information for the sample multimodal data are used as labels. Supervised learning and reinforcement learning are then used to train the base model, resulting in a pre-trained muscle fatigue assessment model.

[0146] Among them, the driving posture evaluation model refers to a model that evaluates the driving posture of drivers and passengers based on input data (such as multimodal data). The base model to be trained for the driving posture evaluation model is constructed based on deep learning algorithms (such as neural networks), large models, or fuzzy logic. Sample multimodal data is used as the training set, and the labeled results of driving posture evaluation information for the sample multimodal data are used as labels. Supervised learning and reinforcement learning are used to train the base model, resulting in a pre-trained driving posture evaluation model.

[0147] Based on deep learning algorithms, large models, or fuzzy logic, base models to be trained for comfort evaluation, muscle fatigue evaluation, and driving posture evaluation are constructed respectively. Then, supervised learning and reinforcement learning are used to train the base models respectively, resulting in pre-trained comfort evaluation models, pre-trained muscle fatigue evaluation models, and pre-trained driving posture evaluation models.

[0148] For example, multimodal data is input into a pre-trained comfort evaluation model to output seat comfort evaluation information (e.g., 0-100 points). Multimodal data is also input into a pre-trained muscle fatigue prediction model to detect abnormal behaviors such as hitting or kneading the shoulders or legs of the driver / passenger. This model, combined with the driver / passenger's audio data and facial images, comprehensively judges and outputs the driver / passenger's muscle fatigue evaluation information. Multimodal data is then input into a pre-trained driving posture prediction model to output the driver / passenger's driving posture data. If this driving posture data falls within a pre-defined standard driving posture data range, the driver / passenger's driving posture evaluation information is confirmed as a safe driving posture; otherwise, it is confirmed as an unsafe driving posture. Finally, a weighted summation of the pressure distribution entropy value and the pressure concentration coefficient is performed to obtain the seat's pressure distribution evaluation information.

[0149] In this embodiment, by inputting multimodal data into a pre-trained comfort evaluation model, seat comfort evaluation information can be intelligently output. Based on the target pressure data in the multimodal data, pressure distribution evaluation information can be accurately obtained. At the same time, by inputting multimodal data into a pre-trained muscle fatigue evaluation model and a pre-trained driving posture evaluation model, the muscle fatigue evaluation information and driving posture evaluation information of the driver and passengers can be accurately identified. This achieves a multi-dimensional comprehensive evaluation of seat comfort and pressure distribution, as well as the muscle fatigue and driving posture of the driver and passengers, providing a reliable processing basis for subsequent adjustment of seat parameters.

[0150] In an exemplary embodiment, step S101 above, acquiring multimodal data of the car seat and the occupants sitting on the seat, specifically includes the following: acquiring raw data of the seat and the occupants sitting on the seat in terms of vision, touch, hearing, and physiology; preprocessing the various raw data to obtain various preprocessed data of the seat and the occupants; performing spatiotemporal synchronization processing on the various preprocessed data to obtain various synchronized data of the seat and the occupants; and inputting the various synchronized data into a multimodal fusion model to obtain multimodal data of the seat and the occupants.

[0151] Among them, the multimodal fusion model refers to an intelligent model that performs feature fusion processing on input data (such as multiple synchronized data). The base model to be trained for the multimodal fusion model is constructed based on machine learning or deep learning (such as attention mechanism), using the synchronized data of the samples as the training set, using the annotation results of the multimodal data of the synchronized data of the samples as labels, and using supervised learning and reinforcement learning to train the base model to obtain the pre-trained multimodal fusion model.

[0152] The synchronized data includes, but is not limited to, spatiotemporally synchronized pressure data, spatiotemporally synchronized image data, spatiotemporally synchronized posture data, spatiotemporally synchronized audio data, and spatiotemporally synchronized physiological data.

[0153] For example, the processor acquires seat pressure data and multimodal raw data such as video data, posture data, audio data, and physiological data of the occupants in the seat. The processor employs Region of Interest (ROI) coding technology to perform high-resolution processing on the human body region in the visual data. For instance, it first detects and locates the human body region in the visual data, designating it as the ROI and other regions (such as the background region) as non-ROIs. Then, it performs pixel-level segmentation of the ROI (such as the human body region) and non-ROI (such as the background region) in the video data to obtain human body region images and non-human body region images. Next, it uses high-resolution coding methods to perform image encoding processing on the human body region images to obtain a high-resolution bitstream of the human body region images. For example, it increases the sampling rate of the human body region images, reduces the compression ratio of the human body region images, and preserves details such as the edges, textures, and postures of the human body in the human body region images to ensure that the human body region is clearly distinguishable after decoding. Low resolution is used in this case. The encoding method performs image encoding processing on the non-human region image to obtain a low-resolution bitstream of the non-human region image. For example, it reduces the sampling rate of the non-human region image and increases the compression ratio to save encoding bits for the non-human region image, and allows for a certain degree of blurring or distortion in the decoded non-human region image. Finally, the high-resolution bitstream of the human region image is integrated with the low-resolution bitstream of the non-human region image, and the location identifier of the human region is added before transmission / storage. The decoding end performs decoding processing on the human region and the non-human region separately according to the location identifier of the human region, and then stitches the decoded human image and the decoded non-human region image to restore the complete image, obtaining the preprocessed image data. This ensures that the human region is displayed at high resolution and the non-human region is displayed normally in the preprocessed image data. The processor can also use the preprocessed image data to analyze and obtain the posture data of the driver and passengers.

[0154] The processor can employ the 3σ criterion to detect outliers in the physiological data, remove these outliers, and obtain anomaly-processed physiological data. Then, using a combination of forward padding and interpolation, it can impute missing values ​​in the anomaly-processed physiological data, resulting in preprocessed physiological data. Furthermore, it can perform separate preprocessing on posture data, pressure data, and audio data to obtain preprocessed posture data, preprocessed pressure data, and preprocessed audio data. For example, preprocessing may include outlier handling for posture and pressure data, and noise reduction for audio data.

[0155] Spatiotemporal synchronization of various preprocessed data can be achieved by first performing time synchronization processing (time alignment accuracy ±10ms) on preprocessed stress data, preprocessed image data, preprocessed posture data, preprocessed audio data, and preprocessed physiological data based on the timestamps of each preprocessed data, resulting in time-aligned stress data, time-aligned image data, time-aligned posture data, time-aligned audio data, and time-aligned physiological data; then, standardizing these time-aligned data yields spatiotemporally synchronized stress data, spatiotemporally synchronized image data, spatiotemporally synchronized posture data, spatiotemporally synchronized audio data, and spatiotemporally synchronized physiological data.

[0156] The spatiotemporally synchronized pressure data, spatiotemporally synchronized image data, spatiotemporally synchronized posture data, spatiotemporally synchronized audio data, and spatiotemporally synchronized physiological data are input into a multimodal fusion model. The multimodal fusion model performs feature fusion processing on different types of synchronized data and outputs multimodal data.

[0157] In this embodiment, raw data from seats and occupants in multiple dimensions, including visual, tactile, auditory, and physiological aspects, are collected and then preprocessed, spatiotemporally synchronized, and fused using a multimodal model. Spatiotemporal synchronization processing enables the standardization and temporal alignment of multi-source data, ensuring consistency of data across different dimensions in time and space, and improving data reliability and the accuracy of subsequent analysis. The multimodal fusion model integrates the synchronized data to form unified and efficient multimodal data, providing a high-quality data foundation for subsequent evaluations of seats and occupants.

[0158] In practical applications, raw data includes at least video data, posture data, stress data, audio data, and physiological data.

[0159] In one embodiment, raw data on the seat and the occupants sitting on it, including visual, tactile, auditory, and physiological aspects, are acquired. Specifically, this includes: acquiring video data of the occupants sitting on the seat; performing human posture detection processing on the video data to obtain visual posture data; obtaining tactile pressure data based on captured information about changes in resistance values ​​of conductive materials associated with the seat caused by physical deformation of the seat; obtaining auditory audio data based on captured voice signals and ambient sound signals from the vehicle's interior environment; and obtaining physiological data based on captured physiological signals of the occupants.

[0160] For example, video data (or image data) of the driver / passenger in the seat is acquired through DMS and OMS cameras, and transmitted to the processor. The processor uses a human posture detection algorithm based on YOLOv8 (You Only Look Once version 8) or a human posture detection algorithm based on MediaPipe Pose (an open-source multimedia machine learning model) to extract multiple (e.g., 17 to 21) key skeletal feature points from the video data (or image data). Based on the skeletal feature points, the processor analyzes the driver / passenger's visual posture data, such as head posture angles (e.g., head pitch ±30°, head yaw ±45°), shoulder tilt, and spinal curvature, to detect the driver / passenger's body shape, sitting posture, and limb position.

[0161] The pressure is sensed by using flexible thin-film sensors to detect changes in the physical properties of the seat material. For example, when the seat undergoes physical deformation, the resistance of the conductive material inside the seat (such as carbon paste or nano-carbon material) changes. The pressure data of the seat in terms of touch can then be deduced from the measured resistance change information and transmitted to the processor.

[0162] The car's audio acquisition equipment collects voice signals and ambient sound signals from the car's interior environment, and transmits these signals as auditory audio data to the processor.

[0163] The system collects physiological signals (such as heart rate, muscle activity, and body temperature) from occupants by using miniature sensors embedded in the seat back (heart rate sensor, 1Hz sampling frequency, ±2bpm accuracy), electromyography (EMG) sensors in the thigh area of ​​the seat cushion (200Hz sampling frequency, used to detect muscle activity), and contact-type body temperature sensors on the seat surface (measurement range 32-42℃, ±0.3℃ accuracy). These physiological signals are then used as physiological data for the occupants. The physiological data is transmitted to the processor via Bluetooth Low Energy, and can also be encrypted using an encryption protocol before Bluetooth transmission to ensure data security.

[0164] It can also monitor the dynamic changes in the seat status in real time, such as vehicle acceleration / deceleration, lateral acceleration during cornering, and seat adjustment position information, by using a triaxial accelerometer (measurement range ±10g, resolution 0.01g), a backrest angle sensor (measurement range 0-180°, accuracy ±0.5°), and a seat position sensor (measurement range 0-250mm, accuracy ±0.1mm) installed at the bottom of the seat; and transmit the seat status change information to the processor.

[0165] In this embodiment, by simultaneously collecting raw data from multiple dimensions, including video, posture, pressure, audio, and physiological data, a comprehensive and accurate perception of the state of drivers and passengers and the in-vehicle environment is achieved. This avoids the one-sidedness of single-data collection and can truly reflect the state of the driving and riding scenario. Human posture detection through video data can accurately obtain the posture data of drivers and passengers, providing a visual basis for posture analysis. Pressure data obtained through the resistance value change information of the seat's conductive material can reflect the seat deformation and the contact pressure of drivers and passengers in real time, closely matching actual driving and riding interactions. Audio data obtained through in-vehicle voice and ambient sound signals enriches the dimensions of scene perception. Physiological data obtained through the physiological signals of drivers and passengers can intuitively reflect the physical functional state of drivers and passengers. By using raw data from multiple sources, the driving and riding scenario can be comprehensively and accurately depicted, providing reliable data support for subsequent evaluation of seats and drivers and passengers.

[0166] In one embodiment, such as Figure 4 As shown, an optional vehicle control method is provided, which is illustrated using an example of the method applied to a car's processor, and includes the following steps:

[0167] Step S401: Obtain raw data on the seat and the occupants in the seat in terms of vision, touch, hearing and physiology.

[0168] Step S402: Preprocess various raw data to obtain various preprocessed data of seats and occupants.

[0169] Step S403: Perform spatiotemporal synchronization processing on various preprocessed data to obtain various synchronized data of seats and occupants.

[0170] Step S404: Input the various synchronized data into the multimodal fusion model to obtain multimodal data of the seat and the driver / passenger.

[0171] Step S405: Based on multimodal data, obtain first evaluation information of the seat and second evaluation information of the driver and passengers; the first evaluation information includes at least pressure distribution evaluation information and comfort evaluation information; the second evaluation information includes at least muscle fatigue evaluation information and driving posture evaluation information.

[0172] Step S406: Based on the current driving condition data of the vehicle and the historical adjustment data of the seat, determine the first weight corresponding to the pressure distribution evaluation information, the second weight corresponding to the comfort evaluation information, the third weight corresponding to the muscle fatigue evaluation information, and the fourth weight corresponding to the driving posture evaluation information.

[0173] Step S407: Based on the first weight, second weight, third weight and fourth weight, the pressure distribution evaluation information, comfort evaluation information, muscle fatigue evaluation information and driving posture evaluation information are fused to obtain the current seat comfort value.

[0174] Step S408: Based on the current seat comfort value and the target seat comfort value, adjust the current parameters of the seat components associated with the first evaluation information and the second evaluation information to obtain the parameters to be adjusted for the seat components.

[0175] Step S409: Generate an adjustment command for the seat based on the parameters to be adjusted; the adjustment command is used to instruct the seat components to be adjusted based on the parameters to be adjusted.

[0176] The aforementioned vehicle control method achieves the following beneficial effects: By acquiring multimodal data of the car seat and occupants, it obtains first evaluation information covering seat pressure distribution and comfort, and second evaluation information including occupant muscle fatigue and driving posture, realizing integrated perception and evaluation from four dimensions: pressure, comfort, physiology, and posture. Furthermore, it calculates the current seat comfort value by combining the first and second evaluation information, and compares it with a preset target seat comfort value. Based on the comfort value difference, it intelligently decides the parameters to be adjusted for the associated seat components, and generates adjustment commands for the seat accordingly. Ultimately, it drives the seat components to perform adjustment actions. By transforming subjective comfort feelings into quantifiable objective comfort value indicators, it achieves dynamic adaptive adjustment of parameters such as seat support and posture, effectively improving seat comfort, optimizing occupant driving posture, alleviating occupant muscle fatigue, and improving driving safety.

[0177] To more clearly illustrate the vehicle control method provided in the embodiments of this disclosure, such as Figure 5 As shown, the above vehicle control method will be specifically described below with a specific embodiment, including the following:

[0178] Step S501, Multimodal data acquisition: Collect raw data on the status of drivers and passengers and related seats through the visual perception unit, physiological perception unit, pressure perception unit and posture perception unit of the perception layer;

[0179] Step S502, Data Preprocessing and Fusion: The collected raw data is preprocessed, features are extracted and spatiotemporally synchronized. Data fusion of the synchronized data is also achieved through a multimodal fusion model based on attention mechanism to obtain multimodal data.

[0180] Step S503, State Assessment and Context Recognition: Based on the fused multimodal data, the comfort evaluation information of the seat is evaluated through the comfort evaluation model. The muscle fatigue evaluation information and driving posture evaluation information of the driver and passengers are evaluated separately through the large model. At the same time, based on the target pressure data in the multimodal data, the pressure distribution evaluation information of the seat is obtained. The driving and riding scenario of the car is identified, the current driving condition data is obtained, and the historical adjustment data of the seat is acquired.

[0181] Step S504, Intelligent Decision Making and Parameter Optimization: Based on pressure distribution evaluation information, comfort evaluation information, muscle fatigue evaluation information, driving posture evaluation information, current driving condition data of the vehicle and historical adjustment data of the seat, the parameters to be adjusted for the seat components are dynamically optimized.

[0182] Step S505, Execution and Feedback Optimization: The adjustment commands are executed by each sub-unit of the execution layer, and the adjustment effect of the seat components is monitored in real time to form a closed-loop control.

[0183] In practical applications, multimodal perception and intelligent adjustment have improved the subjective comfort rating of drivers and passengers. Specifically, this manifests as improved pressure distribution uniformity, reduced muscle fatigue, and decreased lower back discomfort after long drives. User research indicates that after using these vehicle control methods in a car, complaints related to ride comfort decreased, and user satisfaction increased.

[0184] In this embodiment, the following beneficial effects can be achieved: by acquiring multimodal data of the car seat and the driver and passengers, pressure distribution evaluation information, comfort evaluation information, muscle fatigue evaluation information, and driving posture evaluation information of the driver and passengers are obtained based on the multimodal data, realizing the fusion perception and evaluation from four dimensions of "pressure-comfort-physiology-posture"; furthermore, by combining the first evaluation information, the second evaluation information, and the target seat comfort value, the adjustment parameters of the seat components are obtained, and the adjustment command of the seat is generated accordingly to control the adjustment of the seat components, realizing the intelligent and precise adaptive adjustment of the seat support, posture and other parameters, effectively improving the comfort of the seat, while optimizing the driving posture of the driver and passengers, alleviating the muscle fatigue of the driver and passengers, and improving driving safety.

[0185] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0186] Based on the same inventive concept, this application also provides a vehicle control device for implementing the vehicle control method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more vehicle control device embodiments provided below can be found in the limitations of the vehicle control method described above, and will not be repeated here.

[0187] In one exemplary embodiment, such as Figure 6 As shown, a vehicle control device 600 is provided, comprising:

[0188] The data acquisition module 601 is used to acquire multimodal data of the car seats and the occupants sitting in the seats.

[0189] The information evaluation module 602 is used to obtain first evaluation information of the seat and second evaluation information of the driver and passengers based on multimodal data; the first evaluation information includes at least pressure distribution evaluation information and comfort evaluation information; the second evaluation information includes at least muscle fatigue evaluation information and driving posture evaluation information.

[0190] The comfort evaluation module 603 is used to obtain the current seat comfort value of the seat based on the first evaluation information and the second evaluation information.

[0191] The parameter optimization module 604 is used to adjust the current parameters of the seat components associated with the first evaluation information and the second evaluation information based on the current seat comfort value and the target seat comfort value, so as to obtain the parameters of the seat components to be adjusted.

[0192] The seat adjustment module 605 is used to generate seat adjustment instructions based on the parameters to be adjusted; the adjustment instructions are used to instruct the seat components to be adjusted based on the parameters to be adjusted.

[0193] In one embodiment, the vehicle control device 600 further includes a fatigue warning module, which is used to input multimodal data, historical multimodal data of the seat and the driver / passenger into a pre-trained human state prediction model to obtain a predicted second evaluation information of the driver / passenger in a future first time period; based on the predicted second evaluation information, perform fatigue identification processing on the driver / passenger to obtain the predicted fatigue state of the driver / passenger in the future first time period; if the predicted fatigue state is detected to meet preset fatigue driving conditions, then obtain a first safety parameter of the seat based on the predicted second evaluation information; generate a first safety adjustment command for the seat based on the first safety parameter; the first safety adjustment command is used to instruct the seat vibration element adjustment to be completed before the arrival of the future first time period based on the first safety parameter.

[0194] In one embodiment, the vehicle control device 600 further includes a driving warning module, which is used to input multimodal data, historical multimodal data of the seat and the occupants into a pre-trained seat state prediction model to obtain a first evaluation information of the seat in a future second time period; based on the first evaluation information, to perform driving state recognition processing on the occupants to obtain a predicted driving state of the occupants in the future second time period; if the predicted driving state is detected to meet preset aggressive driving conditions, to obtain a second safety parameter of the seat based on the first evaluation information; to generate a second safety adjustment command for the seat based on the second safety parameter; the second safety adjustment command is used to instruct the seat vibration element adjustment to be completed before the arrival of the future second time period based on the second safety parameter.

[0195] In one embodiment, the comfort evaluation module 603 is further configured to determine, based on the current driving condition data of the vehicle and the historical adjustment data of the seat, a first weight corresponding to the pressure distribution evaluation information, a second weight corresponding to the comfort evaluation information, a third weight corresponding to the muscle fatigue evaluation information, and a fourth weight corresponding to the driving posture evaluation information; and to perform fusion processing on the pressure distribution evaluation information, comfort evaluation information, muscle fatigue evaluation information, and driving posture evaluation information according to the first weight, second weight, third weight, and fourth weight to obtain the current seat comfort value of the seat.

[0196] In one embodiment, the parameter optimization module 604 is further configured to determine, based on the first weight, second weight, third weight, fourth weight, and target seat comfort value, target pressure distribution evaluation information corresponding to pressure distribution evaluation information, target comfort evaluation information corresponding to comfort evaluation information, target muscle fatigue evaluation information corresponding to muscle fatigue evaluation information, and target driving posture evaluation information corresponding to driving posture evaluation information; based on the target pressure distribution evaluation information, perform parameter optimization processing on the current parameters of the first seat component associated with the pressure distribution evaluation information to obtain the parameters to be adjusted for the first seat component; based on the target comfort evaluation information, perform parameter optimization processing on the current parameters of the second seat component associated with the comfort evaluation information to obtain the parameters to be adjusted for the second seat component; based on the target muscle fatigue evaluation information, perform parameter optimization processing on the current parameters of the third seat component associated with the muscle fatigue evaluation information to obtain the parameters to be adjusted for the third seat component; and based on the target driving posture evaluation information, perform parameter optimization processing on the current parameters of the fourth seat component associated with the driving posture evaluation information to obtain the parameters to be adjusted for the fourth seat component.

[0197] In one embodiment, the information evaluation module 602 is further configured to input multimodal data into a pre-trained comfort evaluation model to obtain seat comfort evaluation information; the comfort evaluation model is trained through supervised learning and reinforcement learning; based on the target pressure data in the multimodal data, seat pressure distribution evaluation information is obtained; multimodal data is input into a pre-trained muscle fatigue evaluation model to obtain driver and passenger muscle fatigue evaluation information; and multimodal data is input into a pre-trained driving posture evaluation model to obtain driver and passenger driving posture evaluation information.

[0198] In one embodiment, the data acquisition module 601 is further configured to acquire raw data of the seat and the occupants on the seat in terms of vision, touch, hearing and physiology; preprocess the various raw data to obtain various preprocessed data of the seat and the occupants; perform spatiotemporal synchronization processing on the various preprocessed data to obtain various synchronized data of the seat and the occupants; and input the various synchronized data into a multimodal fusion model to obtain multimodal data of the seat and the occupants.

[0199] In one embodiment, the raw data includes at least video data, posture data, pressure data, audio data, and physiological data. The vehicle control device 600 also includes a data acquisition module for acquiring video data of the driver and passengers sitting in the seats, performing human posture detection processing on the video data to obtain visual posture data of the driver and passengers; obtaining tactile pressure data based on captured information about changes in resistance values ​​of conductive materials associated with the seat caused by physical deformation of the seat; obtaining auditory audio data based on captured voice signals and ambient sound signals from the vehicle's interior environment; and obtaining physiological data based on captured physiological signals of the driver and passengers.

[0200] The various modules in the aforementioned vehicle control device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the vehicle's processor in hardware form or independent of it, or stored in the vehicle's memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0201] In one exemplary embodiment, such as Figure 7 As shown, an automobile is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0202] Those skilled in the art will understand that Figure 7 The structure shown is a block diagram of a partial structure related to the present application and does not constitute a limitation on the automobile to which the present application is applied. A specific automobile may include more or fewer parts than shown in the figure, or combine certain parts, or have different part arrangements.

[0203] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0204] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0205] 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, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0206] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program mentioned can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0207] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0208] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A vehicle control method, characterized in that, The method includes: Acquire multimodal data of the car seats and the occupants in the seats; Based on the multimodal data, a first evaluation information of the seat and a second evaluation information of the driver and passenger are obtained; the first evaluation information includes at least pressure distribution evaluation information and comfort evaluation information; the second evaluation information includes at least muscle fatigue evaluation information and driving posture evaluation information. Based on the first evaluation information and the second evaluation information, the current seat comfort value of the seat is obtained; Based on the current seat comfort value and the target seat comfort value, the current parameters of the seat components associated with the first evaluation information and the second evaluation information are adjusted to obtain the parameters of the seat components to be adjusted. An adjustment command for the seat is generated based on the parameter to be adjusted; the adjustment command is used to instruct the seat components to be adjusted based on the parameter to be adjusted.

2. The method according to claim 1, characterized in that, After acquiring the multimodal data of the car seats and the occupants in the seats, the process also includes: The multimodal data, the seat, and the historical multimodal data of the driver and passengers are input into a pre-trained human state prediction model to obtain the predicted second evaluation information of the driver and passengers in the first time period in the future. Based on the predicted second evaluation information, fatigue identification processing is performed on the driver and passenger to obtain the predicted fatigue state of the driver and passenger in the future first time period. If the predicted fatigue state is detected to meet the preset fatigue driving conditions, then the first safety parameter of the seat is obtained based on the predicted second evaluation information. Based on the first safety parameter, a first safety adjustment command for the seat is generated; the first safety adjustment command is used to instruct the adjustment of the vibration element of the seat to be completed before the arrival of the first future time period based on the first safety parameter.

3. The method according to claim 1, characterized in that, After acquiring the multimodal data of the car seats and the occupants in the seats, the process also includes: The multimodal data, the seat, and the historical multimodal data of the driver and passengers are input into a pre-trained seat state prediction model to obtain the first evaluation information of the seat in the second future time period. Based on the predicted first evaluation information, the driver and passenger are subjected to driving state identification processing to obtain the predicted driving state of the driver and passenger in the future second time period. If the predicted driving state is detected to meet the preset aggressive driving conditions, then the second safety parameter of the seat is obtained based on the predicted first evaluation information; Based on the second safety parameter, a second safety adjustment command for the seat is generated; the second safety adjustment command is used to instruct the adjustment of the vibration element of the seat to be completed before the arrival of the future second time period based on the second safety parameter.

4. The method according to claim 1, characterized in that, The step of obtaining the current seat comfort value of the seat based on the first evaluation information and the second evaluation information includes: Based on the current driving condition data of the vehicle and the historical adjustment data of the seat, the first weight corresponding to the pressure distribution evaluation information, the second weight corresponding to the comfort evaluation information, the third weight corresponding to the muscle fatigue evaluation information, and the fourth weight corresponding to the driving posture evaluation information are determined. Based on the first weight, the second weight, the third weight, and the fourth weight, the pressure distribution evaluation information, the comfort evaluation information, the muscle fatigue evaluation information, and the driving posture evaluation information are fused to obtain the current seat comfort value of the seat.

5. The method according to claim 4, characterized in that, The step of adjusting the current parameters of the seat components associated with the first evaluation information and the second evaluation information based on the current seat comfort value and the target seat comfort value to obtain the parameters of the seat components to be adjusted includes: Based on the first weight, the second weight, the third weight, the fourth weight, and the target seat comfort value, the target pressure distribution evaluation information corresponding to the pressure distribution evaluation information, the target comfort evaluation information corresponding to the comfort evaluation information, the target muscle fatigue evaluation information corresponding to the muscle fatigue evaluation information, and the target driving posture evaluation information corresponding to the driving posture evaluation information are determined. Based on the target pressure distribution evaluation information, the current parameters of the first seat component associated with the pressure distribution evaluation information are optimized to obtain the parameters to be adjusted for the first seat component. Based on the target comfort evaluation information, the current parameters of the second seat component associated with the comfort evaluation information are optimized to obtain the parameters to be adjusted for the second seat component. Based on the target muscle fatigue evaluation information, the current parameters of the third seat component associated with the muscle fatigue evaluation information are optimized to obtain the parameters to be adjusted for the third seat component. Based on the target driving posture evaluation information, the current parameters of the fourth seat component associated with the driving posture evaluation information are optimized to obtain the parameters to be adjusted for the fourth seat component.

6. The method according to claim 1, characterized in that, The process of obtaining first evaluation information for the seat and second evaluation information for the driver and passenger based on the multimodal data includes: The multimodal data is input into a pre-trained comfort evaluation model to obtain the comfort evaluation information of the seat; the comfort evaluation model is trained through supervised learning and reinforcement learning. Based on the target pressure data in the multimodal data, the pressure distribution evaluation information of the seat is obtained; The multimodal data is input into a pre-trained muscle fatigue evaluation model to obtain the muscle fatigue evaluation information of the driver and passengers. The multimodal data is input into a pre-trained driving posture evaluation model to obtain the driving posture evaluation information of the driver and passengers.

7. The method according to claim 1, characterized in that, The acquisition of multimodal data of the car seats and the occupants in the seats includes: Acquire raw data on the seat and the occupants seated therein in terms of vision, touch, hearing, and physiology; The various original data are preprocessed to obtain various preprocessed data of the seat and the driver / passenger; The various preprocessed data are spatiotemporally synchronized to obtain various synchronized data of the seat and the driver / passenger. The various synchronized data are input into a multimodal fusion model to obtain multimodal data of the seat and the driver / passenger.

8. The method according to claim 7, characterized in that, The raw data includes at least video data, posture data, stress data, audio data, and physiological data; The acquisition of raw data on the seat and the occupants seated therein, including visual, tactile, auditory, and physiological data, includes: Acquire video data of the driver and passenger sitting in the seat, perform human posture detection processing on the video data, and obtain the posture data of the driver and passenger in the visual aspect; The tactile pressure data is obtained based on the captured information about the change in resistance of the conductive material associated with the seat caused by the physical deformation of the seat. The audio data related to hearing is obtained based on the captured voice signals and ambient sound signals of the car's interior environment. The physiological data related to the physiological aspects are obtained based on the captured physiological signals of the drivers and passengers.

9. A vehicle control device, characterized in that, The device includes: The data acquisition module is used to acquire multimodal data of the car seats and the occupants sitting in the seats; The information evaluation module is used to obtain first evaluation information of the seat and second evaluation information of the driver and passenger based on the multimodal data; the first evaluation information includes at least pressure distribution evaluation information and comfort evaluation information; the second evaluation information includes at least muscle fatigue evaluation information and driving posture evaluation information. The comfort evaluation module is used to obtain the current seat comfort value of the seat based on the first evaluation information and the second evaluation information; The parameter optimization module is used to adjust the current parameters of the seat components associated with the first evaluation information and the second evaluation information based on the current seat comfort value and the target seat comfort value of the seat, so as to obtain the parameters of the seat components to be adjusted. A seat adjustment module is used to generate an adjustment command for the seat based on the parameter to be adjusted; the adjustment command is used to instruct the seat components to be adjusted based on the parameter to be adjusted.

10. A vehicle comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

Images