Multi-dimensional joint decision-making vehicle-mounted odor adjustment method, device, equipment and product

Abstract

Embodiments of the present application provide a kind of multi-dimensional joint decision vehicle-mounted smell adjustment method, device, equipment and product, involve the technical field of vehicle control.The method comprises: obtaining the gas detection signal in the vehicle cabin and driving scene information;Time sequence feature analysis is carried out to the gas detection signal, the smell category and the duration of smell category are determined, and the amplitude analysis is carried out to the gas detection signal, the smell intensity corresponding to the smell category is determined;Smell category, duration, smell intensity and driving scene information are input into decision tree model, and smell adjustment strategy is generated;According to the device control instruction of smell adjustment strategy generation, it is to control the smell adjustment equipment in the vehicle cabin and run.The method is used to reach the purpose of optimizing the smell adjustment effect in vehicle intelligent cabin.

Description

Technical Field

[0001] This application relates to the technical field of vehicle control, and in particular to a method, apparatus, device and product for vehicle odor regulation based on multi-dimensional joint decision-making. Background Technology

[0002] With the development of intelligent cockpit technology, air quality management and odor recognition in intelligent cockpits are gradually becoming key aspects of improving user experience.

[0003] Current technologies typically only determine the presence or absence of odors based on the concentration of volatile organic compounds (VOCs) in the smart cockpit and then take corresponding actions (such as turning on the air conditioning or opening windows) based on the determination. However, this method of odor regulation does not consider factors such as the specific type and concentration of the odor, as well as differences in the vehicle's driving scenario. It cannot provide targeted odor regulation based on the actual situation inside the smart cockpit, resulting in poor odor regulation effects. Summary of the Invention

[0004] This application provides a multi-dimensional joint decision-making method, apparatus, device, and product for in-vehicle odor regulation, in order to optimize the odor regulation effect in the intelligent cockpit of a vehicle.

[0005] In a first aspect, embodiments of this application provide a multi-dimensional joint decision-making method for in-vehicle odor regulation, including:

[0006] Acquire gas detection signals from inside the vehicle cabin and driving scenario information;

[0007] Temporal feature analysis is performed on the gas detection signal to determine the odor category and duration of the odor category, and amplitude analysis is performed on the gas detection signal to determine the odor intensity corresponding to the odor category;

[0008] The odor category, duration, intensity, and driving scenario information are input into the decision tree model to generate an odor regulation strategy.

[0009] Based on the odor control strategy, control commands are generated to control the operation of the odor control equipment in the vehicle cabin.

[0010] In one possible implementation, amplitude analysis is performed on the gas detection signal to determine the odor intensity corresponding to the odor category, specifically including:

[0011] Acquire gas detection signals from multi-channel odor sensors, and determine the baseline response value and saturation response value of any channel odor sensor according to the odor category;

[0012] Based on the baseline response value and the saturation response value, the response amplitude of the corresponding gas detection signal is normalized.

[0013] Based on the odor category, the corresponding odor weight is determined, and the response amplitude after normalization is weighted according to the odor weight to calculate the comprehensive intensity score;

[0014] The odor intensity corresponding to the odor category is determined based on the comprehensive intensity score.

[0015] In one possible implementation, the odor control strategy includes at least the equipment operating intensity and the equipment operating duration, and the method further includes:

[0016] Acquire images of the occupants inside the vehicle cabin;

[0017] By using image recognition, occupant behavior is identified from occupant images. When odor-discomfort behaviors are found in the occupant behavior identification results, the equipment operation intensity and / or equipment operation duration in the odor control strategy are increased. Odor-discomfort behaviors include at least one of the following: frowning expression, disgust expression, coughing, and sneezing.

[0018] In one possible implementation, the method further includes:

[0019] When sleep behavior is detected in the occupant behavior recognition results, the equipment operation intensity in the odor regulation strategy is reduced. Sleep behavior includes at least one of the following: closing eyes, head tilting back, drowsiness, and supine posture.

[0020] In one possible implementation, the method further includes:

[0021] Identify the occupants in the vehicle cabin based on their images;

[0022] Based on the occupant identification results, the odor sensitivity level is determined;

[0023] Update odor regulation strategies based on odor sensitivity levels.

[0024] In one possible implementation, the method further includes:

[0025] Determine the vehicle driving scenario based on the driving scenario information;

[0026] When the vehicle driving scenario is a high-speed driving scenario, a tunnel driving scenario, and / or a rainy driving scenario, a window-opening prohibition policy is generated.

[0027] When the vehicle is being driven at night, determine the maximum window opening width;

[0028] Update the odor control strategy based on the policy of prohibiting window opening or the maximum window opening range.

[0029] In one possible implementation, the method further includes:

[0030] The concentration of carbon dioxide gas in the vehicle cabin is determined based on the gas detection signal.

[0031] If the rate of increase in carbon dioxide concentration exceeds a first preset rate, and the odor category and corresponding odor intensity remain unchanged, maintain the current odor regulation strategy; and / or,

[0032] The concentration of fine particulate matter in the vehicle cabin is determined based on gas detection signals.

[0033] When the rate of increase in the concentration of fine particulate matter exceeds the second preset rate, the current odor control strategy will be adjusted to an air purification strategy.

[0034] In one possible implementation, the method further includes:

[0035] When the duration of an odor category is greater than or equal to a preset duration threshold, the odor category, duration, odor intensity, and driving scenario information are input into the decision tree model; and / or,

[0036] In-vehicle alerts are generated based on an odor control strategy, wherein the in-vehicle alerts include at least one of voice alerts, light alerts, and display alerts.

[0037] Secondly, embodiments of this application provide a multi-dimensional joint decision-making vehicle odor control device, comprising:

[0038] The acquisition module is used to acquire gas detection signals in the vehicle cabin and driving scenario information;

[0039] The determination module is used to perform time-series feature analysis on the gas detection signal, determine the odor category and duration of the odor category, and perform amplitude analysis on the gas detection signal to determine the odor intensity corresponding to the odor category.

[0040] The generation module is used to input odor category, duration, odor intensity, and driving scenario information into the decision tree model to generate odor regulation strategies.

[0041] The generation module is also used to generate equipment control commands based on the odor control strategy to control the operation of the odor control equipment in the vehicle cabin.

[0042] In one possible implementation, the determining module specifically includes:

[0043] The numerical determination submodule is used to acquire the gas detection signals from the multi-channel odor sensor and determine the baseline response value and saturation response value of any channel odor sensor according to the odor category.

[0044] The calculation submodule is used to perform normalization calculation on the response amplitude of the corresponding gas detection signal based on the baseline response value and the saturation response value.

[0045] The scoring submodule is used to determine the corresponding odor weight based on the odor category, and to calculate the comprehensive intensity score by weighting the normalized response amplitude according to the odor weight.

[0046] The intensity determination submodule is used to determine the odor intensity corresponding to the odor category based on the comprehensive intensity score.

[0047] In one possible implementation, the odor control strategy includes at least the device operating intensity and the device operating duration, and the acquisition module is also used to acquire images of occupants inside the vehicle cabin.

[0048] The generation module is also used to perform occupant behavior recognition on occupant images through image recognition. When odor-discomfort behavior is found in the occupant behavior recognition results, the device operation intensity and / or device operation duration in the odor regulation strategy are increased. Odor-discomfort behavior includes at least one of the following: frowning expression, disgust expression, coughing, and sneezing.

[0049] In one possible implementation, the generation module is further configured to: reduce the device operation intensity in the odor regulation strategy when sleep behavior is found in the occupant behavior recognition results, wherein sleep behavior includes at least one of closing eyes, head tilting back, drowsiness, and supine posture.

[0050] In one possible implementation, the device further includes:

[0051] The identity recognition module is used to identify the occupants in the vehicle cabin based on their images.

[0052] The determination module is also used to determine the odor sensitivity level based on the occupant identification results;

[0053] The generation module is also used to update the odor regulation strategy based on the odor sensitivity level.

[0054] In one possible implementation, the determining module is further configured to determine the vehicle driving scenario based on the driving scenario information;

[0055] The generation module is also used to generate a window-opening prohibition strategy when the vehicle driving scenario is a high-speed driving scenario, a tunnel driving scenario, and / or a rainy driving scenario.

[0056] The generation module is also used to determine the maximum window opening width when the vehicle driving scenario is a nighttime driving scenario;

[0057] The generation module is also used to update the odor control strategy based on the prohibited window opening policy or the maximum window opening range.

[0058] In one possible implementation, the determining module is further configured to determine the concentration of carbon dioxide gas in the vehicle cabin based on the gas detection signal;

[0059] The generation module is also used to maintain the current odor regulation strategy when the rate of increase in carbon dioxide gas concentration is greater than a first preset rate, and the odor category and corresponding odor intensity remain unchanged; and / or,

[0060] The determination module is also used to determine the concentration of fine particulate matter in the vehicle cabin based on the gas detection signal;

[0061] The generation module is also used to adjust the current odor regulation strategy to an air purification strategy when the rate of increase in the concentration of fine particulate matter is greater than a second preset rate.

[0062] In one possible implementation, the generation module is further configured to input the odor category, duration, odor intensity, and driving scenario information into the decision tree model when the duration of the odor category is greater than or equal to a preset duration threshold; and / or, the device further includes:

[0063] The voice module is used to generate in-vehicle prompts based on an odor control strategy. The in-vehicle prompts include at least one of voice prompts, light prompts, and display prompts.

[0064] Thirdly, embodiments of this application provide an in-vehicle electronic device, including: a memory and a processor;

[0065] The memory stores the instructions that the computer executes;

[0066] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0067] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0068] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0069] The multi-dimensional joint decision-making method, apparatus, device, and product for in-vehicle odor regulation provided in this application first acquire gas detection signals and driving scenario information within the vehicle cabin. Based on the gas detection signals, it determines the odor category, duration, and intensity, achieving refined odor identification within the smart cabin and obtaining basic data for generating odor regulation strategies. Then, it inputs the odor category, duration, intensity, and driving scenario information into a decision tree model to generate odor regulation strategies. This allows for simultaneous consideration of odor type characteristics, odor change states, and vehicle driving scenario factors during the air management process within the in-vehicle smart cabin. This enables more precise and diversified control of the odor regulation equipment within the vehicle cabin, thereby improving the accuracy of odor regulation decisions, enhancing the scenario adaptability of the regulation strategies, optimizing the odor regulation effect in the vehicle smart cabin, and improving the user experience. Attached Figure Description

[0070] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0071] Figure 1 A flowchart illustrating the multi-dimensional joint decision-making method for in-vehicle odor regulation provided in this application. Figure 1 ;

[0072] Figure 2 A flowchart illustrating the multi-dimensional joint decision-making method for in-vehicle odor regulation provided in this application. Figure 1 ;

[0073] Figure 3 Schematic diagram of the multi-dimensional joint decision-making vehicle odor regulation device provided in this application Figure 1 ;

[0074] Figure 4 Schematic diagram of the multi-dimensional joint decision-making vehicle odor regulation device provided in this application Figure 2 ;

[0075] Figure 5 A schematic diagram of the structure of the vehicle-mounted electronic device provided in this application.

[0076] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0077] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application.

[0078] In-vehicle odor control technology falls under the field of intelligent cabin environmental control, primarily used for monitoring air quality, identifying odors, and coordinating the control of ventilation and purification equipment within the vehicle cabin. Related systems are typically connected to in-vehicle gas sensors, air conditioning units, air purifiers, window actuators, and vehicle status information acquisition modules to automatically adjust the cabin air environment while the vehicle is in motion.

[0079] Existing air management solutions in smart vehicle cabins mostly control air purifiers and air conditioning units to circulate internal and external air or control window opening and closing based on changes in volatile organic compound (VOC) concentration detected by gas sensors. The basic principle is to compare the detected value with a preset threshold; when the result exceeds the threshold, ventilation or purification actions are executed; otherwise, the current state is maintained.

[0080] These solutions primarily rely on a single concentration criterion—the presence or absence of odors in the vehicle's smart cabin—making it difficult to distinguish between different odor sources such as smoke, food, and perfume. They also struggle to differentiate between short-term fluctuations and persistent odors, resulting in a rigid cabin air quality management strategy. Furthermore, in complex driving environments such as highways, tunnels, and traffic jams, directly triggering window opening or ventilation based solely on changes in gas concentration detection values ​​can lead to mismatches between the adjustment actions and the actual driving conditions, posing safety risks or wasting energy.

[0081] To address the aforementioned technical issues, this application provides a multi-dimensional joint decision-making method for in-vehicle odor regulation. By employing a multi-dimensional joint decision-making processing path, the time-varying and amplitude-varying characteristics of the gas detection signal are analyzed separately to distinguish odor type, duration, and intensity. This is then combined with driving scenario information for unified judgment. By linking the gas acquisition module, scenario information acquisition module, decision model, and odor regulation device in the vehicle's intelligent cockpit, the air conditioning action within the intelligent cockpit no longer relies on a single concentration threshold. Instead, it generates control strategies that better reflect the actual situation based on odor attributes and the driving environment. This improves the accuracy and scenario adaptability of odor recognition during in-vehicle air management, ultimately optimizing the odor regulation effect in the vehicle's intelligent cockpit.

[0082] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0083] Figure 1 A flowchart illustrating the multi-dimensional joint decision-making method for in-vehicle odor regulation provided in this application. Figure 1 ,like Figure 1 As shown, the method includes:

[0084] S101: Acquire gas detection signals and driving scenario information within the vehicle cabin.

[0085] For example, the gas detection signal is used to characterize the odor-related detection results in the air inside the vehicle cabin, and serves as the input basis for subsequent identification of odor type, duration and intensity; driving scenario information is used to describe the current driving environment of the vehicle and serves as scenario input in the subsequent decision-making process.

[0086] In one example, the entity executing step S101 can be an odor control module deployed in the vehicle's cockpit controller. This control module is communicatively connected to the in-vehicle gas collection module, the vehicle network bus, and the air conditioning control unit. The control module receives sensor data output from the gas detection device and acquires information about the vehicle's current driving scenario to form input information for subsequent analysis and decision-making.

[0087] For the aforementioned gas detection device, an 8-channel micro-electro-mechanical systems (MEMS) gas sensor array can be used. The position of each sensor in the array and the number of sensors at each position can be set according to the actual situation, such as placing the sensors above the driver's head, above the passenger's head, or at the rear center air vent. Each MEMS gas sensor can use different sensitive materials (such as metal oxides, metal-organic frameworks, polymers, graphene, etc.) as the gas sensing material to cover the main odor molecules such as volatile organic compounds (VOCs), thiols, amines, fatty acids, aldehydes, and ketones, forming an odor fingerprint spectrum.

[0088] In another example, a baseline value determination / initialization should also be performed on the aforementioned gas detection device. This can be done by first purifying the air inside the vehicle, such as by closing the windows and running the air conditioner in recirculation mode for 5 minutes. Then, the baseline response vector R0 = [R0(1), R0(2), ..., R0(8)] of the 8-channel sensor array under standard temperature and humidity conditions (e.g., setting the baseline temperature T0 = 25 degrees Celsius and the baseline humidity H0 = 50%RH) is recorded as the baseline value. After the baseline value is determined, subsequent detected gas signals can be compared with the baseline value to determine the type and / or intensity of the odor.

[0089] In one possible implementation, the method further includes: correcting the detected gas detection signal based on the detected temperature and humidity information, so as to use the corrected gas detection signal for time-series feature analysis.

[0090] For example, during temperature compensation correction, the 8-channel sensor array can synchronously read the current temperature T during each sampling. For each channel of the gas sensor, the temperature compensation factor can be determined based on the temperature sensitivity coefficient of the sensitive material in the gas sensor (e.g., obtained by looking up a table), where the temperature sensitivity coefficient can be pre-determined through factory calibration experiments.

[0091] Subsequently, the gas detection signal detected under the current temperature conditions is corrected to the equivalent detection value at the reference temperature T0. For example, for a gas sensor whose sensitive material is a metal oxide, when the temperature rises, the resistance in the sensor decreases and the sensitivity increases, requiring a compensation factor greater than 1; conversely, when the temperature falls, a compensation factor less than 1 is used.

[0092] Similarly, humidity compensation can be set after temperature compensation. The basic logic of humidity compensation is as follows: in high humidity environments, the competitive adsorption of water molecules leads to a decrease in sensor sensitivity, which needs to be corrected by multiplying by a compensation coefficient greater than 1; conversely, in low humidity environments, it is corrected by multiplying by a compensation coefficient less than 1. The compensation coefficient is determined by the humidity sensitivity coefficient of the gas sensor's own sensitive material, which can be pre-determined through factory calibration experiments.

[0093] Based on the above analysis, step S101 obtains gas detection signals and driving scenario information in the vehicle cabin, so that subsequent processing no longer relies solely on a single gas / odor concentration value to trigger air purification strategies. Instead, it establishes a correspondence between odor state and vehicle operating environment, providing an input basis for distinguishing different odor events and their corresponding control methods.

[0094] S102: Perform time-series feature analysis on the gas detection signal to determine the odor category and duration of the odor category, and perform amplitude analysis on the gas detection signal to determine the odor intensity corresponding to the odor category.

[0095] For example, temporal feature analysis is used to extract the variation patterns in the time dimension of continuously acquired gas detection signals in order to identify odor categories and determine the duration of the odor category in the current detection cycle; amplitude analysis is used to evaluate the odor intensity corresponding to the odor category from the perspective of signal strength.

[0096] In one example, the control module extracts a gas detection signal sequence from the current analysis window, identifies the current odor category based on the temporal variation characteristics of the gas detection signal, and determines the duration based on the time period during which the odor category is continuously identified. The odor category can correspond to one of the following: body odor, smoke, food smell, perfume smell, pet smell, fresh smell, abnormal odor (unidentified odor), or no odor.

[0097] In the corresponding amplitude analysis, the odor intensity corresponding to the odor category can be determined based on the amplitude change of the gas detection signal, and the odor intensity can be divided into five levels: none, weak, medium, strong, and extremely strong, or mapped to a continuous intensity score.

[0098] Specifically, when the gas detection signal maintains a high amplitude and exhibits a continuous response pattern within multiple consecutive analysis windows, the odor intensity corresponding to that odor category can be determined to be high; when the signal fluctuates only within a short period of time, it can be determined to be of medium or low intensity level by combining the duration result, thereby making the intensity determination more consistent with the actual odor event.

[0099] It should be noted that the time series feature analysis and amplitude analysis in step S102 above can be performed in parallel in the same calculation process. That is, after extracting the analysis window, the time features and amplitude features are calculated synchronously, and then a unified description result of the odor event containing odor category, duration and odor intensity is formed. Alternatively, the time series feature analysis can be performed first to determine the candidate odor category, and then the corresponding intensity analysis can be performed.

[0100] In another example, AI models can be used to identify the type of odor collected. For instance, a lightweight model suitable for automotive-grade microcontroller units (MCUs) can be deployed in the vehicle's cockpit controller to perform time-series feature processing on the gas detection signals collected by an 8-channel sensor array after temperature and humidity compensation, resulting in a normalized feature vector.

[0101] Subsequently, the time window length W = 50 sampling points (corresponding to a continuous sampling sequence of approximately 5 seconds, with a sampling frequency of 10Hz) was set, and the sliding step size S = 10 sampling points. Each sample is an 8xW two-dimensional input matrix, where the 8 rows correspond to the feature dimensions of the signal acquired by the 8-channel sensor array, and the W columns correspond to the time step size.

[0102] The aforementioned feature matrix is ​​input into the first convolutional layer for low-level feature extraction. This can be achieved using 16 1D convolutional kernels, each with a size of 3 (kernel_size=3). One-dimensional convolution operations are performed on the eight feature channels in the time dimension. During the convolution operation, the ReLU activation function or other activation functions can be used; the specific convolution process will not be detailed here. This layer outputs 16 feature maps, each capturing the local variation pattern of the gas sensor response within the aforementioned time window, such as rising edges, falling edges, and peak values.

[0103] The output structure described above is then fed into the second convolutional layer for higher-order feature extraction. This can be achieved using 32 1D convolutional kernels (kernel_size=3) to further convolve the 16 feature maps output from the first convolutional layer, ultimately producing 32 feature maps. This convolutional layer is used to extract higher-order temporal feature combinations, such as peak patterns (rising then falling), valley patterns (falling then rising), and complex patterns like continuous rises and continuous falls.

[0104] Then, Global Average Pooling (GAP) is performed on each feature map output by the second convolutional layer. The values ​​of each feature map at all time steps are averaged and compressed into a scalar. That is, the 32 feature maps are compressed into a 32-dimensional feature vector, which achieves the effects of dimensionality reduction and anti-overfitting.

[0105] The aforementioned 32-dimensional feature vector is then input into a fully connected layer, where a linear transformation operation is performed. This layer can contain 8 output neurons (corresponding to 8 odor categories). The original 8-dimensional score vector `logits` output by the fully connected layer is then normalized (e.g., using the Softmax function), converting the original output into a posterior probability distribution for each category, yielding the probabilities, or confidence scores, for each of the 8 odor categories. Finally, the odor category with the highest probability is selected as the final odor classification result.

[0106] It should be noted that when training the above model, offline training can be performed in the cloud using a labeled odor dataset, such as employing the cross-entropy loss function and the Adam optimizer. Furthermore, the trained model can be compressed; for example, quantization techniques can be used to compress the model parameters from 32-bit floating-point numbers to 8-bit integers, significantly reducing model size and computational cost, so that it can be deployed on automotive-grade microcontrollers to meet the real-time performance and resource constraints required for automotive applications.

[0107] Based on the above analysis, step S102 processes the time variation characteristics and amplitude variation characteristics respectively, and then summarizes the results of the two to distinguish between short-term fluctuations and persistent odors, distinguish odor events from different sources, and provide clear category, duration and intensity parameters for subsequent strategy generation, thereby solving the problem that a single concentration threshold scheme is difficult to accurately identify the source and persistence of odors.

[0108] S103: Input the odor category, duration, odor intensity, and driving scenario information into the decision tree model to generate an odor regulation strategy. The odor regulation strategy may also include a voice prompt strategy.

[0109] For example, the decision tree model is used to generate an odor regulation strategy that matches the current vehicle cabin state and driving environment based on multi-dimensional input conditions. For instance, the odor regulation control module in the vehicle cabin controller receives the odor category, duration, and intensity output in step S102, while simultaneously reading the driving scenario information from step S101 and constructing a decision input vector. The decision tree model can be stored in the local memory of the vehicle controller, and its node rules can be pre-established based on historical experimental data, manual annotation experience, and scenario constraints. This model can adapt the odor regulation strategy to the driving scenario information, select the appropriate processing direction based on the odor category, and determine the regulation intensity and duration based on the duration and intensity, thereby outputting a complete odor regulation strategy.

[0110] In one example, the decision tree model can be divided into multiple levels. First, it branches according to the odor category (body odor, smoke odor, food odor, perfume odor, pet odor, fresh odor, abnormal odor, no odor). Different odor types can be mapped to different basic action strategies to determine the device's operating mode. The specific mapping relationship can be set according to the actual situation. For example, the basic strategy corresponding to smoke odor is "purification + fragrance"; the basic strategy corresponding to body odor is "ventilation + purification"; the basic strategy corresponding to food odor is "open window + purification"; and the basic strategy corresponding to perfume odor is "dilution + filtration".

[0111] In the second-level branch, the equipment operating intensity is determined by 5 levels (none, weak, medium, strong, and extremely strong). The higher the level, the greater the operating intensity. For example: weak level corresponds to purification fan speed level 1 and fragrance concentration level 1; medium level corresponds to purification fan speed level 2 and fragrance concentration level 2; strong level corresponds to purification fan speed level 3 and fragrance concentration level 3 plus voice prompt; extremely strong level corresponds to purification fan speed level 4 (the highest level), fragrance off plus emergency voice prompt.

[0112] In the third-level branch, the operating time of the equipment is determined based on the duration of the odor. This can be achieved using a piecewise function. For example, if the duration is 10-30 seconds, the ventilation time can be set to 1 minute. If the duration is longer than 30 seconds, the equipment can be set to run continuously until the odor disappears. At this point, the operating intensity of the equipment can also be adjusted according to the duration to improve the efficiency of air purification.

[0113] In the fourth-level branch, the actions determined by the first three levels are adjusted according to the driving scenario information of the vehicle, such as avoiding the strategy of opening windows for ventilation in high-speed scenarios or rainy and snowy weather scenarios.

[0114] The aforementioned vehicle driving scenario information can also include the vehicle's own operating scenario and the surrounding environmental scenario, such as the Heating, Ventilation and Air Conditioning (HVAC) scenario. For example, in a high-temperature environment, the odor control strategy can prioritize switching to the internal circulation + air conditioning cooling mode. Another example is in a direct sunlight scenario, where the priority is to enhance seat ventilation in conjunction with air conditioning cooling.

[0115] Vehicle driving scenario information can also include traffic and road scenarios. For example, in congested areas, the odor control strategy can prioritize switching to the internal circulation mode. It can also include vehicle parking information. For example, after ensuring that no one is in the car and the doors are locked, deep purification (such as ozone sterilization, ultraviolet disinfection, etc.) can be performed, but ventilation must be completed before the next use of the car.

[0116] In another example, the odor control strategy, as an intermediate result between the decision-making outcome and the execution control, includes at least the following: equipment operating mode, equipment operating intensity, and equipment operating duration. The equipment operating mode can include air conditioning control, ventilation control, air purification, or a combination of control modes; the equipment operating intensity can include fan speed, purifier power level, or the operating level of other regulating devices; the equipment operating duration can be set to a fixed duration or can be set to remain unchanged for several detection cycles while awaiting verification.

[0117] For example, when the decision input shows that the odor category is smoke, the duration is more than 30 seconds, and the odor intensity is high, the decision tree model can output an odor regulation strategy that improves ventilation and enhances purification intensity; when the input shows that the odor category is food, the duration is short, and the odor intensity is medium, the model can output an odor regulation strategy that runs at a medium regulation intensity for a preset duration.

[0118] Based on the above analysis, step S103 unifies the determination of odor category, duration, odor intensity and driving scenario information, so that the generation of adjustment strategy no longer depends on a single threshold trigger, but combines odor attributes and scenario constraints to determine the control direction and control intensity, thereby making the generated strategy consistent with the actual driving environment.

[0119] S104: Generate equipment control commands based on the odor control strategy to control the operation of the odor control equipment in the vehicle cabin.

[0120] For example, device control instructions are used to translate odor control strategies into executable control information for actual actuators, and drive the odor control devices in the vehicle cabin to perform corresponding actions. Odor control devices in the vehicle cabin may include air conditioning, internal and external circulation fans, fragrance devices, air purifiers, windows, and other devices related to cabin airflow control.

[0121] In one example, the control module parses the odor regulation strategy output in step S103 above, mapping the operating mode, operating intensity, and operating duration to device-level control variables. After mapping, the control module sends control messages to the corresponding execution unit via the vehicle communication bus and receives execution confirmation status to determine whether the command has taken effect.

[0122] For strategies requiring continuous control, the control module can also start a timer to maintain the target device state for the duration set by the strategy, and decide whether to restore the default state or continue to execute the next round of adjustment based on the latest detection results after the expiration.

[0123] During the operation of the relevant equipment, the control module can reacquire the gas detection signal in each control cycle and perform a closed-loop comparison between the new detection result and the current execution state. For example, when the detected odor intensity has fallen below the target threshold and remains below it for several cycles, the control module can generate a weakening or stopping command to cause the equipment to exit the enhanced regulation state; when the detection result does not improve or rises again, the control module can adjust the control parameters based on the updated decision results.

[0124] Based on the above analysis, this application provides a multi-dimensional joint decision-making method for in-vehicle odor regulation, including: acquiring gas detection signals and driving scenario information within the vehicle cabin; performing time-series feature analysis on the gas detection signals to determine the odor category and duration of the odor category, and performing amplitude analysis on the gas detection signals to determine the odor intensity corresponding to the odor category; inputting the odor category, duration, odor intensity, and driving scenario information into a decision tree model to generate an odor regulation strategy; and generating device control commands based on the odor regulation strategy to control the operation of odor regulation devices within the vehicle cabin. Through this method, the time variation characteristics and amplitude variation characteristics of the gas detection signals are extracted and input into the decision tree model along with the driving scenario information, constructing a multi-dimensional joint decision-making link for in-vehicle intelligent cabin air management. This enables the system to identify odor categories and odor intensities, and distinguish between short-term fluctuations and persistent odors, so as to output corresponding device linkage control strategies under different vehicle driving scenarios. Therefore, the gas regulation actions in the intelligent cockpit are generated based on odor attributes and driving environment, which can address issues such as false triggering, insufficient response, and scene mismatch, thereby optimizing the odor regulation effect in the vehicle's intelligent cockpit.

[0125] In one possible implementation, amplitude analysis is performed on the gas detection signal to determine the odor intensity corresponding to the odor category. Specifically, this includes: acquiring gas detection signals from a multi-channel odor sensor, and determining the baseline response value and saturation response value of any channel odor sensor according to the odor category; normalizing the response amplitude of the corresponding gas detection signal based on the baseline response value and saturation response value; determining the corresponding odor weight according to the odor category, and weighting the normalized response amplitude according to the odor weight to calculate a comprehensive intensity score; and determining the odor intensity corresponding to the odor category based on the comprehensive intensity score.

[0126] For example, the odor control module in the vehicle cabin controller acquires gas detection signals synchronously collected by multiple odor sensors, and calls the corresponding calibration parameters based on the identified odor category. These calibration parameters include the baseline response value and saturation response value of each channel. The baseline response value and saturation response value can be obtained offline from historical sampling data, or they can be corrected online by combining the current temperature and humidity and sensor drift.

[0127] For the response amplitude of any channel, a normalization operation can be performed by subtracting the baseline response value from the current detection value and then dividing by the difference between the saturation response value and the baseline response value, so that the amplitudes of different channels under different odor categories are mapped to normalized values ​​within a unified range.

[0128] In one example, taking a single-channel gas sensor as an example, after purifying the air inside the vehicle, the baseline response value in this environment is first obtained as the normalized value of 0; then, the saturated response value at a known concentration of a certain type of gas is obtained as the normalized value of 1. Afterwards, when the gas sensor of this channel obtains the gas detection signal of the corresponding odor, it can first perform temperature and humidity correction, and then perform normalization operation on the amplitude of the corrected signal so as to linearly map the response value under the current gas environment to the interval [0, 1].

[0129] Considering the different chemical molecular characteristics of different odor types, their response patterns in an 8-channel MEMS gas sensor array also differ. For example, smoke odor mainly contains sulfides and VOCs, showing strong responses in the thiol and VOC sensitive channels; food odor mainly contains amines and fatty acids, showing strong responses in the amine and fatty acid sensitive channels; perfume odor mainly contains aldehydes, ketones, and esters, showing strong responses in the aldehyde and ketone sensitive channels and specific metal-organic framework channels. Therefore, for each odor type, a set of 8-dimensional weight vectors is pre-defined. These weight vectors can be obtained through a network model. For example, using a supervised learning network model, with the goal of minimizing intensity prediction error, the weight coefficients of each channel are optimized through a backpropagation algorithm using various odor samples of known concentrations.

[0130] Subsequently, based on the odor types identified in the previous steps, the odor weights of each channel are determined, and then the response amplitudes after normalization are weighted to obtain a comprehensive intensity score.

[0131] After obtaining the overall intensity score, the odor intensity of the corresponding odor category can be obtained by interval mapping. For example, the overall intensity score is mapped to 5 levels. When the overall intensity score ∈ [0, 0.1], it is mapped to no level, meaning the gas sensor response is within the baseline fluctuation range and there is no effective odor signal; when the overall intensity score ∈ [0.1, 0.3], it is mapped to a weak level, meaning a faint odor is detected, but the concentration is low and may not have a significant impact on the human body; when the overall intensity score ∈ [0.3, 0.5], it is mapped to a medium level, meaning a moderate intensity odor is detected, and the concentration reaches the level that the human body can perceive. Air conditioning strategies can be taken in a timely manner according to the gas type, duration, personal needs of the occupants, or other factors; when the overall intensity score ∈ [0.5, 0.7], it is mapped to a strong level, meaning a relatively strong odor is detected, which may cause discomfort, and corresponding air conditioning strategies should be taken; when the overall intensity score ∈ [0.7, 1], it is mapped to an extremely strong level, meaning a very strong odor is detected, and corresponding air conditioning strategies should be taken immediately, and the equipment operating intensity and / or equipment operating time should be appropriately increased.

[0132] It should be noted that the above-mentioned mapping interval threshold settings can be calibrated and adjusted according to actual application scenarios and user preferences.

[0133] In another example, the aforementioned odor weights can also be given by a weight table corresponding to the category, and multiplied by the normalized amplitude of each channel to form a comprehensive intensity score, which is then mapped to different discrete intensity levels or continuous intensity values.

[0134] By incorporating the channel baseline, saturation boundary, and weights of the gas sensor into the amplitude evaluation process, given that the odor category has been determined, the odor intensity output is matched with the specific odor type. This allows the amplitudes of different sensing channels to be aggregated and calculated on the same scale, and combined with the odor weights to form a comprehensive intensity score. This helps ensure the accuracy of odor intensity calculation, enabling the output results to accurately reflect the actual strength differences of the target odor in the cabin, and to adapt to the differences in response of different channels, thereby improving the stability and consistency of odor intensity determination.

[0135] Figure 2 A flowchart illustrating the multi-dimensional joint decision-making method for in-vehicle odor regulation provided in this application. Figure 2 ,like Figure 2 As shown, in this embodiment... Figure 1 Based on the embodiments, a detailed description of the multi-dimensional joint decision-making method for in-vehicle odor regulation is provided. This method includes:

[0136] S201: Acquire gas detection signals and driving scenario information within the vehicle's cabin.

[0137] S202: Perform time-series feature analysis on the gas detection signal to determine the odor category and duration of the odor category, and perform amplitude analysis on the gas detection signal to determine the odor intensity corresponding to the odor category.

[0138] S203: Input the odor category, duration, odor intensity, and driving scenario information into the decision tree model to generate an odor regulation strategy; wherein, the odor regulation strategy includes at least the equipment operating intensity and the equipment operating duration.

[0139] S204: Acquire images of the occupants inside the vehicle's cabin.

[0140] S205: By using image recognition, occupant behavior recognition is performed on occupant images. When odor-discomfort behavior is found in the occupant behavior recognition results, the equipment operation intensity and / or equipment operation duration in the odor control strategy are increased. Odor-discomfort behavior includes at least one of the following: frowning expression, disgust expression, coughing, and sneezing.

[0141] S206: Generate equipment control commands based on the odor control strategy to control the operation of the odor control equipment in the vehicle cabin.

[0142] For example, an image acquisition device can be installed in the smart cockpit to acquire images of occupants. The image acquisition device may include one or more in-vehicle cameras, which may be located near the center console, A-pillar, roof, or rearview mirror to cover the areas where the driver and passengers commonly sit.

[0143] The acquired occupant images are first processed by target detection to locate the occupant's head or upper body area, and then input into a pre-trained behavior recognition model for identification. This behavior recognition model can be a convolutional neural network, a temporal network, or a combination of both. After obtaining the behavior category output by the model, it is compared with a set of odor-discomfort behaviors. When any of the following is matched—frowning, disgust, coughing, or sneezing—the control module increases the original equipment operating intensity and operating time, which can be manifested by increasing the fan speed, increasing the purification air volume, or extending the continuous operating time.

[0144] In one example, by setting up image acquisition devices to obtain occupant images, richer and more granular decision factors are provided for the aforementioned odor regulation strategy, thereby optimizing the strategy. For instance, when occupant images are detected and facial expressions indicating disgust or discomfort, such as frowning, pursing lips, or covering the nose, it is determined that the occupants are dissatisfied with the current in-vehicle odor. In this case, regardless of the current odor type or intensity level, air purification should be initiated, increasing the purification intensity by one level and increasing airflow, based on the current odor regulation strategy.

[0145] Additionally, when the strategy includes activating the fragrance system, the fragrance concentration can be reduced by 50% or immediately switched to another fragrance type. For example, if a rear passenger frowns after the fragrance system is detected to be activated and a certain type of perfume is present in the vehicle, the control module will automatically reduce the fragrance concentration involved in the odor adjustment strategy by one level.

[0146] For example, when image recognition detects that an occupant is coughing or sneezing repeatedly, it indicates that the occupant is sensitive to the current odor or air quality. In this case, the purification level should be increased to the highest setting, the fragrance release should be turned off, and the user should be gently asked via voice whether they need to activate the negative ion purification function.

[0147] It should be noted that if the above-mentioned facial expressions of the occupants are not detected during the above recognition process, or if the occupants are detected to be smiling or happy, it indicates that the occupants are accepting or satisfied with the current odor regulation strategy. This can be recorded and used as positive feedback to strengthen the decision-making model.

[0148] By incorporating occupant physiological feedback into the odor control within the intelligent cockpit, the system no longer relies solely on gas detection results. Instead, it incorporates real-time occupant responses to odors, allowing for compensatory adjustments to operating intensity and duration. This enables the odor control strategy to adjust promptly when occupants experience significant discomfort, enhancing equipment output and ensuring that cabin odor treatment aligns with occupant comfort.

[0149] In one possible implementation, the method further includes: when sleep behavior is found in the occupant behavior recognition results, reducing the device operation intensity in the odor regulation strategy, wherein sleep behavior includes at least one of closing eyes, head tilting back, drowsiness, and supine posture.

[0150] For example, when the above image recognition results show sleep behavior, the control module can lower the device operating intensity parameter corresponding to the odor adjustment strategy. The device operating intensity may include at least one of the following: fan speed, purifier power, air volume, or odor release.

[0151] In one example, if the front passenger and rear passengers are detected to be asleep or resting with their eyes closed and heads tilted back, even if the odor control strategy output by the decision tree model includes a window-opening instruction, the windows can be opened or closed according to the set minimum ventilation volume to avoid wind noise disturbance. Simultaneously, the corresponding in-vehicle voice prompt can be muted, and the driver can be notified only through silence and flashing lights that in-vehicle air conditioning is in progress.

[0152] In another example, when reducing the operating intensity of the device in the odor control strategy, the current operating intensity can be switched from a higher level to a lower level, or proportionally reduced to 50% to 80% of the original intensity, and the adjusted parameters can be written into the execution instructions of the odor control device.

[0153] In this way, the odor regulation strategy can dynamically converge the execution intensity based on the occupant's sleep state, and the control parameter adjustment can be directly triggered by the image recognition results. This allows the odor regulation to maintain the basic air handling capacity while reducing the interference to resting occupants, thus making the in-vehicle odor regulation process more in line with the occupant's state in the vehicle cabin.

[0154] In one possible implementation, the method further includes: identifying the occupants in the vehicle cabin based on occupant images; determining an odor sensitivity level based on the occupant identification results; and updating the odor regulation strategy based on the odor sensitivity level.

[0155] For example, an in-vehicle camera installed in the vehicle can capture images of the vehicle cabin, detect, extract and compare facial regions in the images, and combine them with pre-stored occupant face templates or historical occupant files to output occupant identification results.

[0156] In one example, occupant identification can be categorized as solo travelers, groups, children, elderly, and pregnant women. Simultaneously, the presence of pets in the cabin can also be identified. Then, through mapping, the occupant identification results are categorized into different sensitivity levels, such as extremely sensitive, sensitive, normal, and insensitive. Based on these sensitivity levels, parameters in the odor control strategy, such as fragrance concentration, fragrance type, window opening width, pop-up volume, pop-up timing, and voice frequency, are dynamically adjusted.

[0157] In the above process, the recognition of occupant images is not only used for identity recognition, but also for the recognition of the number and distribution of people, so as to provide personalized adjustment strategies. For example, when a child or elderly passenger is detected on the left side of the rear seat, the window opening is automatically reduced to avoid discomfort for such passengers due to the window being open.

[0158] In scenarios with multiple occupants, each occupant may have different sensitivities to odors and different personal preferences. In such cases, a conservative strategy should be adopted. For example, if occupants have inconsistent sensitivities to odors, the strategy can be adjusted based on the highest sensitivity level among them. Or, if different occupants have different physiological reactions to the same odor, such as if an occupant exhibits negative emotions (e.g., frowning or covering their nose), the strategy should be adjusted based on the corresponding occupant's identification result.

[0159] It should be noted that different weights can be assigned to occupants in different seats in order to generate an overall adjustment strategy, with the driver's weight being the highest.

[0160] In another example, the identification results can also be compared with the occupant identities (such as occupant IDs) stored in the vehicle to identify whether a specific occupant (such as an allergic occupant, a pregnant woman, or an occupant with specific needs) is currently in the cabin.

[0161] When the identification result corresponds to a registered passenger, it can be directly associated with the passenger's preset odor sensitivity level; when the identification result corresponds to an unregistered passenger, the corresponding level can be determined based on facial feature similarity, passenger's historical feedback records, or statistical mapping relationship with similar passengers.

[0162] Once the identification results are obtained, a fifth branch can be added based on the fourth branch mentioned above. The parameters in the odor regulation strategy can be adjusted in a final personalized manner according to the occupant identification results and the corresponding odor sensitivity level.

[0163] Odor sensitivity levels can be set according to actual conditions, such as extremely sensitive, sensitive, normal, and insensitive levels. Different levels correspond to different combinations of equipment operating parameters. For example, pregnant women, infants, and people with allergies can be set to the extremely sensitive level. For these groups, the adjustment strategy should include prohibiting opening windows, not using fragrances, or using only non-irritating, mild plant fragrances (such as lavender or chamomile). At the same time, the voice prompt volume should be at the minimum or low volume. Alternatively, the elderly and children can be set to the sensitive level. For these occupants, the window ventilation strategy should be prioritized over the fragrance strategy.

[0164] The vehicle-mounted camera can be an automotive-grade wide-angle camera, which can maintain a relatively stable imaging effect under low light conditions. At the same time, it can be combined with a seat pressure sensor to assist in occupant identification (such as by weight).

[0165] By adopting the above method, the odor regulation strategy in this application embodiment no longer performs uniform control based solely on gas detection values, but can be updated differently based on the occupant's identity, so that the equipment operating parameters match the occupant's tolerance to odors, thereby improving the targeting of odor regulation and the comfort of the vehicle cabin, and enhancing the control adaptability in different occupant co-riding scenarios.

[0166] It should be noted that the above image recognition process is performed in real time only within the vehicle-side trusted execution environment, and only extracts and outputs expression tags (such as frowning, sleeping, smiling, etc.) and posture tags (such as lying on one's back, lying on one's side, etc.). It does not store any original image data, transmit it to the cloud, perform identity verification, or associate it with any personal identity information, ensuring full compliance with relevant privacy requirements.

[0167] In one possible implementation, the method further includes: determining the vehicle driving scenario based on driving scenario information; generating a window-opening prohibition strategy when the vehicle driving scenario is a high-speed driving scenario, a tunnel driving scenario, and / or a rainy driving scenario; determining the maximum window opening range when the vehicle driving scenario is a night driving scenario; and updating the odor adjustment strategy based on the window-opening prohibition strategy or the maximum window opening range.

[0168] For example, the vehicle driving scenario can be jointly determined based on the road type output by the navigation module, the tunnel identification output by the positioning module, the rainfall status output by the rain sensor, the ambient illuminance output by the light sensor, and the vehicle speed signal.

[0169] When the scenario is determined to be a high-speed driving scenario (e.g., vehicle speed exceeding 80 km / h), a tunnel driving scenario, and / or a rainy driving scenario, the control module generates a window-opening prohibition strategy and blocks the window opening command from the window actuator. The corresponding ventilation mode in the odor control strategy is updated to only allow air conditioning recirculation, air purifier operation, or air conditioning linkage without implementing window opening ventilation. For example, opening windows at high speeds will generate significant wind noise, affecting driver attention and passenger comfort, and posing safety hazards such as flying stones. Therefore, the air conditioning strategy is replaced with: recirculation + high-power purification mode, relying on the air conditioning system for air renewal. Similarly, because the air in tunnels contains a high concentration of vehicle exhaust fumes, opening windows will actually introduce more harmful gases. Therefore, the air conditioning strategy is replaced with: recirculation + highest-level purification, and the original strategy is automatically restored at the tunnel exit.

[0170] When the scenario is determined to be nighttime driving, the control module determines the maximum window opening range based on the current vehicle speed, the current window position, and nighttime lighting conditions. It then limits the executable window opening range to within this range, which corresponds to a preset percentage of the window's opening distance, thus controlling the window opening. For example, due to poor visibility at night, opening the window too wide poses a safety hazard. Therefore, the air conditioning strategy can be replaced with: small window opening + internal air circulation + low-level air purification.

[0171] During operation, the odor control strategy first establishes basic control parameters based on odor type, duration, and intensity. Then, it modifies window-related control items based on the vehicle's driving scenario. If a policy prohibiting window opening exists, the window opening control item is directly deleted or frozen. If a maximum window opening range exists, the output limit of the window opening control item is restricted to that range. Through this method, the strategy output remains consistent with the current driving environment, and the control module generates device control commands accordingly, driving the windows, air conditioning, and purification equipment to operate collaboratively.

[0172] With this approach, the odor control strategy can automatically match window opening constraints in different scenarios such as high speed, tunnels, rainy days, and nighttime, reducing false triggering when windows are not suitable for opening, and ensuring that the vehicle odor control process meets both environmental adaptability and driving safety requirements.

[0173] In one possible implementation, the method further includes: determining the carbon dioxide gas concentration in the vehicle cabin based on a gas detection signal; maintaining the current odor control strategy when the rate of increase of the carbon dioxide gas concentration is greater than a first preset rate and the odor category and corresponding odor intensity remain unchanged; and / or determining the concentration of fine particulate matter in the vehicle cabin based on a gas detection signal; and adjusting the current odor control strategy to an air purification strategy when the rate of increase of the fine particulate matter concentration is greater than a second preset rate.

[0174] For example, upon receiving a gas detection signal, the carbon dioxide concentration and fine particulate matter concentration values ​​at the corresponding time are extracted according to a preset sampling period. Differential calculations are then performed on two or more consecutive sampling points to obtain the concentration rise rate. When the carbon dioxide concentration rise rate exceeds a first preset rate, and the odor category and its corresponding odor intensity obtained from the preceding odor identification results remain unchanged, it can be determined that the ventilation conditions in the cabin have changed, for example, due to a switch from external circulation to internal circulation leading to carbon dioxide concentration accumulation, rather than a new odor event. The currently issued odor adjustment command should remain unchanged, and the original ventilation, odor removal, or air supply control should continue to be executed to avoid false triggering.

[0175] When the rate of increase of fine particulate matter concentration (such as PM2.5 concentration) exceeds the second preset rate, it is determined that there is an external pollution source (such as passing through a construction section or a congested section). The control module switches the current odor adjustment strategy to an air purification strategy, prioritizes the air purification process over the odor recognition process, and outputs corresponding equipment control commands to drive the air conditioner's internal and external circulation, the purifier or filter device to operate in coordination.

[0176] It should be noted that the first and second preset rates mentioned above can be calibrated according to the vehicle model, vehicle cabin volume and sensor sensitivity. In practical applications, the threshold can also be adaptively updated based on historical driving data. This application does not limit this.

[0177] It should be noted that for the 8-dimensional feature matrix composed of signals collected by the 8-channel sensor array, carbon dioxide concentration and fine particulate matter concentration can be added, and corresponding normalization and sampling processing can be performed to form a 10-dimensional feature matrix, which is then input into the first convolutional layer for low-level feature extraction. The specific process will not be elaborated here.

[0178] In this scheme, when the carbon dioxide concentration rises rapidly but the odor type and intensity remain unchanged, it indicates that the change in the vehicle cabin air is mainly due to a change in the environmental ventilation status, rather than a new odor event. Therefore, the current odor control strategy is maintained. When the fine particulate matter concentration rises rapidly, the current strategy is switched to an air purification strategy to match the control output with the change in particulate pollution in the vehicle cabin.

[0179] By adopting the above method, the control module can distinguish and judge the air status of the vehicle cabin by combining the concentration of carbon dioxide gas and fine particulate matter, so that the odor adjustment strategy remains stable when the odor does not change, and switches to purification control in a timely manner when particulate pollution worsens, thereby improving the scenario adaptability and control accuracy of vehicle cabin air management.

[0180] In one possible implementation, the method further includes: when the duration of an odor category is greater than or equal to a preset duration threshold, inputting the odor category, duration, odor intensity, and driving scenario information into a decision tree model; and / or generating in-vehicle prompt information according to an odor adjustment strategy, wherein the in-vehicle prompt information includes at least one of voice prompt information, light prompt information, and display prompt information.

[0181] For example, after obtaining the type identification result of a certain odor, the control module continuously counts the effective existence time of the odor category and compares the timing result with a preset duration threshold in order to filter out the influence of instantaneous interference signals (such as the moment of perfume spraying) and avoid false triggering.

[0182] When the duration reaches or exceeds the threshold, the odor category, the corresponding duration, the odor intensity obtained from amplitude analysis, and the current driving scenario information are combined to form a feature vector and input into the decision tree model. The preset duration threshold can be configured separately for each odor category, or it can be uniformly configured as a single duration parameter (e.g., 10s). In practical applications, this threshold can also be adjusted based on vehicle model configuration, vehicle cabin volume, or calibration data; this application does not impose any limitations on this.

[0183] In one example, a sliding window counter can be maintained to record the number of consecutive sampling frames for the current odor type. The sampling interval for each frame can be set to 100ms (corresponding to a sampling frequency of 10Hz). When the number of consecutive frames N is greater than or equal to 100 (i.e., the duration is greater than or equal to 10 seconds), the subsequent decision-making process can be triggered.

[0184] In this process, odor intensity can also be introduced, that is, the subsequent decision-making process is triggered only when the odor intensity is higher than a certain intensity threshold (such as intensity level higher than no level) and the duration is greater than 10 seconds.

[0185] If the scent type changes, the intensity drops to zero, or there is a continuous interruption in frames, the counter is reset. This mechanism effectively filters out momentary interference signals such as the instant perfume is sprayed or the brief odor when pedestrians pass by, preventing false triggering.

[0186] In another example, to ensure that occupants are promptly aware of the operational status of the intelligent cockpit, corresponding in-vehicle prompts should be generated when generating the odor control strategy. These prompts can be voice prompts, light prompts, display prompts, or other types of prompts. Voice prompts are those broadcast via the in-vehicle multimedia equipment, and the volume can be adjusted according to the occupants' needs. Light prompts primarily indicate the illumination of corresponding indicator lights on the vehicle control panel, or ambient lighting. Display prompts are those displayed on in-vehicle display devices (such as in-vehicle terminal displays or head-up displays), and their display duration, font size, and font color can be set according to actual conditions.

[0187] By using the above process and setting a threshold duration as the trigger condition for the decision tree model, interference from instantaneous odor signals can be reduced. The model only performs correlation calculations between odor category, intensity, and driving scenario in continuous odor situations, outputting an odor control strategy that matches the current vehicle cabin state. This transforms odor control from a single-step decision to a dynamic decision based on continuous features, improving the targeting and scenario adaptability of strategy updates.

[0188] Figure 3 A schematic diagram of the structure of the multi-dimensional joint decision-making vehicle odor regulation device provided in this application is shown below. Figure 3 As shown, the multi-dimensional joint decision-making vehicle odor control device 30 provided in this embodiment includes:

[0189] The acquisition module 301 is used to acquire gas detection signals in the vehicle cabin and driving scenario information;

[0190] The determination module 302 is used to perform time-series feature analysis on the gas detection signal to determine the odor category and duration of the odor category, and to perform amplitude analysis on the gas detection signal to determine the odor intensity corresponding to the odor category.

[0191] The generation module 303 is used to input odor category, duration, odor intensity and driving scenario information into the decision tree model to generate odor regulation strategy;

[0192] The generation module 303 is also used to generate equipment control commands based on the odor control strategy to control the operation of the odor control equipment in the vehicle cabin.

[0193] The multi-dimensional joint decision-making vehicle odor control device 30 provided in this embodiment can execute the method provided in the above-described method embodiment. By acquiring cabin gas detection signals and driving scenario information simultaneously through the acquisition module 301, odor changes can be associated with the vehicle's current driving environment. The determination module 302 performs time-series feature analysis and amplitude analysis respectively, thereby separating and identifying odor category, duration, and intensity, enabling the distinction between short-term fluctuations and persistent odors, as well as odors from different sources. The generation module 303 inputs the multi-dimensional parameters into the decision tree model, and then generates a more suitable odor control strategy based on scenarios such as highways, tunnels, and traffic congestion, thus avoiding false triggering or insufficient response caused by relying solely on a single concentration threshold. Furthermore, the generation module 303 outputs device control commands based on the strategy to control the operation of air conditioning, purification, ventilation, or window opening equipment, thereby improving the identification accuracy, adjustment targeting, and scenario adaptability of cabin air management, while taking into account both passenger comfort and environmental safety during driving.

[0194] Figure 4 Schematic diagram of the multi-dimensional joint decision-making vehicle odor regulation device provided in this application Figure 2 ,like Figure 4 As shown, in this embodiment... Figure 3 Based on the embodiments, a detailed description of the multi-dimensional joint decision-making vehicle odor regulation device 40 is provided. This multi-dimensional joint decision-making vehicle odor regulation device 40 includes:

[0195] The acquisition module 401 is used to acquire gas detection signals in the vehicle cabin and driving scenario information;

[0196] The determination module 402 is used to perform time-series feature analysis on the gas detection signal, determine the odor category and duration of the odor category, and perform amplitude analysis on the gas detection signal to determine the odor intensity corresponding to the odor category.

[0197] The generation module 403 is used to input odor category, duration, odor intensity and driving scenario information into the decision tree model to generate odor regulation strategy;

[0198] The generation module 403 is also used to generate equipment control commands based on the odor control strategy to control the operation of the odor control equipment in the vehicle cabin.

[0199] In one possible implementation, the determining module 402 specifically includes:

[0200] The numerical determination submodule 4021 is used to acquire the gas detection signal of the multi-channel odor sensor and determine the baseline response value and saturation response value of any channel odor sensor according to the odor category.

[0201] The calculation submodule 4022 is used to perform normalization calculation on the response amplitude of the corresponding gas detection signal based on the baseline response value and the saturation response value.

[0202] The scoring submodule 4023 is used to determine the corresponding odor weight according to the odor category, and to calculate the comprehensive intensity score by weighting the normalized response amplitude according to the odor weight.

[0203] The intensity determination submodule 4024 is used to determine the odor intensity corresponding to the odor category based on the comprehensive intensity score.

[0204] In one possible implementation, the odor control strategy includes at least the device operating intensity and the device operating duration, and the acquisition module 401 is also used to acquire images of occupants in the vehicle cabin.

[0205] The generation module 403 is also used to perform occupant behavior recognition on the occupant image through image recognition. When there is odor discomfort behavior in the occupant behavior recognition result, the device operation intensity and / or device operation duration in the odor regulation strategy are increased. The odor discomfort behavior includes at least one of frowning expression, disgust expression, coughing, and sneezing.

[0206] In one possible implementation, the generation module 403 is further configured to: reduce the equipment operation intensity in the odor regulation strategy when sleep behavior is found in the occupant behavior recognition results, wherein sleep behavior includes at least one of closing eyes, head tilting back, drowsiness, and supine posture.

[0207] In one possible implementation, the device 40 further includes:

[0208] The identity recognition module 404 is used to identify the identity of the occupants in the vehicle cabin based on the occupant image;

[0209] The determining module 402 is also used to determine the odor sensitivity level based on the occupant identification result;

[0210] The generation module 403 is also used to update the odor regulation strategy based on the odor sensitivity level.

[0211] In one possible implementation, the determining module 402 is further configured to determine the vehicle driving scenario based on the driving scenario information;

[0212] The generation module 403 is also used to generate a window-opening prohibition strategy when the vehicle driving scenario is a high-speed driving scenario, a tunnel driving scenario, and / or a rainy driving scenario.

[0213] The generation module 403 is also used to determine the maximum window opening range when the vehicle driving scenario is a night driving scenario;

[0214] The generation module 403 is also used to update the odor control strategy based on the prohibited window opening strategy or the maximum window opening range.

[0215] In one possible implementation, the determining module 402 is further configured to determine the carbon dioxide gas concentration in the vehicle cabin based on the gas detection signal.

[0216] The generation module 403 is further configured to maintain the current odor control strategy when the rate of increase in carbon dioxide gas concentration is greater than a first preset rate, and the odor category and corresponding odor intensity remain unchanged; and / or,

[0217] The determination module 402 is also used to determine the concentration of fine particulate matter in the vehicle cabin based on the gas detection signal;

[0218] The generation module 403 is also used to adjust the current odor regulation strategy to an air purification strategy when the rate of increase of the concentration of fine particulate matter is greater than a second preset rate.

[0219] In one possible implementation, the generation module 403 is further configured to input the odor category, duration, odor intensity, and driving scenario information into the decision tree model when the duration of the odor category is greater than or equal to a preset duration threshold; and / or, the device 40 further includes:

[0220] The voice module 405 is used to generate in-vehicle prompt information according to the odor adjustment strategy, wherein the in-vehicle prompt information includes at least one of voice prompt information, light prompt information, and display prompt information.

[0221] The multi-dimensional joint decision-making vehicle odor regulation device 40 provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0222] Figure 5 This is a structural schematic diagram of the vehicle-mounted electronic device provided in this application. Figure 5 As shown, the vehicle-mounted electronic device 50 provided in this embodiment includes at least one processor 501 and a memory 502. Optionally, the vehicle-mounted electronic device 50 further includes a communication component 503. The processor 501, memory 502, and communication component 503 are connected via a bus 504.

[0223] In the specific implementation process, at least one processor 501 executes computer execution instructions stored in memory 502, causing at least one processor 501 to execute the above-mentioned multi-dimensional joint decision-making vehicle odor regulation method.

[0224] The aforementioned vehicle-mounted electronic device 50 can be integrated into the intelligent cockpit controller. The memory 502 stores the program instructions corresponding to gas detection, scene recognition, decision-making models, and execution control. After the processor 501 calls the instructions, it can jointly analyze the time and amplitude change characteristics of the gas signal in the cockpit and generate adjustment strategies in combination with vehicle speed, road environment, and ventilation conditions. This enables the same hardware platform to have the ability to identify odors, judge scenes, and link control, thereby improving the accuracy, comfort, and scene adaptability of cockpit air management.

[0225] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0226] The memory may include random access memory (RAM) and non-volatile memory (NVM), such as at least one disk storage device.

[0227] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0228] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described multi-dimensional joint decision-making method for in-vehicle odor regulation.

[0229] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-mentioned multi-dimensional joint decision-making method for vehicle odor regulation.

[0230] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0231] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0232] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0233] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0234] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0235] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0236] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0237] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A multi-dimensional joint decision-making method for in-vehicle odor regulation, characterized in that, include: Acquire gas detection signals from inside the vehicle cabin and driving scenario information; The gas detection signal is subjected to time-series feature analysis to determine the odor category and the duration of the odor category, and the gas detection signal is subjected to amplitude analysis to determine the odor intensity corresponding to the odor category. The odor category, duration, intensity, and driving scenario information are input into a decision tree model to generate an odor regulation strategy. Based on the odor control strategy, device control commands are generated to control the operation of the odor control device in the vehicle cabin.

2. The method according to claim 1, characterized in that, The step of performing amplitude analysis on the gas detection signal to determine the odor intensity corresponding to the odor category specifically includes: Acquire the gas detection signals from the multi-channel odor sensor, and determine the baseline response value and saturation response value of any channel odor sensor according to the odor category; Based on the baseline response value and the saturation response value, the response amplitude of the corresponding gas detection signal is normalized. Based on the odor category, the corresponding odor weight is determined, and the response amplitude after normalization is weighted according to the odor weight to calculate the comprehensive intensity score; Based on the comprehensive intensity score, the odor intensity corresponding to the odor category is determined.

3. The method according to claim 1, characterized in that, The odor control strategy includes at least the equipment operating intensity and the equipment operating duration, and the method further includes: Acquire images of the occupants inside the vehicle cabin; The occupant image is used to identify occupant behavior. When odor-discomfort behavior is found in the occupant behavior identification results, the equipment operation intensity and / or equipment operation time in the odor regulation strategy are increased. The odor-discomfort behavior includes at least one of the following: frowning expression, disgust expression, coughing, and sneezing.

4. The method according to claim 3, characterized in that, The method further includes: When sleep behavior is detected in the occupant behavior recognition results, the device operation intensity in the odor regulation strategy is reduced, wherein the sleep behavior includes at least one of the following: closing eyes, head tilting back, drowsiness, and supine posture.

5. The method according to claim 3, characterized in that, The method further includes: Based on the occupant images, the identities of the occupants in the vehicle cabin are identified; Based on the occupant identification results, the odor sensitivity level is determined; The odor regulation strategy is updated based on the odor sensitivity level.

6. The method according to any one of claims 1-5, characterized in that, The method further includes: Based on the driving scenario information, determine the vehicle driving scenario; When the vehicle driving scenario is a high-speed driving scenario, a tunnel driving scenario, and / or a rainy driving scenario, a window-opening prohibition strategy is generated. When the vehicle driving scenario is a nighttime driving scenario, determine the maximum window opening range; The odor control strategy is updated based on the prohibited window opening strategy or the maximum window opening range.

7. The method according to any one of claims 1-5, characterized in that, The method further includes: The concentration of carbon dioxide gas in the vehicle cabin is determined based on the gas detection signal. When the rate of increase in carbon dioxide gas concentration is greater than a first preset rate, and the odor category and corresponding odor intensity remain unchanged, the current odor regulation strategy is maintained; and / or, The concentration of fine particulate matter in the vehicle cabin is determined based on the gas detection signal. When the rate of increase in the concentration of fine particulate matter exceeds a second preset rate, the current odor control strategy is adjusted to an air purification strategy.

8. The method according to any one of claims 1-5, characterized in that, The method further includes: When the duration of the odor category is greater than or equal to a preset duration threshold, the odor category, the duration, the odor intensity, and the driving scenario information are input into the decision tree model; and / or, The vehicle-mounted prompt information is generated according to the odor adjustment strategy, wherein the vehicle-mounted prompt information includes at least one of voice prompt information, light prompt information, and display prompt information.

9. A multi-dimensional joint decision-making vehicle odor control device, characterized in that, include: The acquisition module is used to acquire gas detection signals in the vehicle cabin and driving scenario information; The determination module is used to perform time-series feature analysis on the gas detection signal to determine the odor category and the duration of the odor category, and to perform amplitude analysis on the gas detection signal to determine the odor intensity corresponding to the odor category; The generation module is used to input the odor category, the duration, the odor intensity, and the driving scenario information into the decision tree model to generate an odor regulation strategy; The generation module is also used to generate device control commands according to the odor regulation strategy to control the operation of the odor regulation device in the vehicle cabin.

10. A vehicle-mounted electronic device / computer-readable storage medium / computer program product, characterized in that, The vehicle-mounted electronic device includes: a memory and a processor; The memory stores computer-executed instructions; The processor executes the computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-8; and / or, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-8; and / or, The computer program product includes a computer program that, when executed by a processor, implements the method described in any one of claims 1-8.

Images