Method, system, and media for sharing driving decisions of autonomous vehicles

Abstract

The application relates to the technical field of automatic driving and discloses a driving decision sharing method, system and medium of an automatic driving vehicle, wherein the driving decision sharing method of the automatic driving vehicle comprises the following steps: collecting facial expressions, sound features and physiological signal data of passengers through multiple sensors to construct a multi-channel emotion perception system; extracting cultural background features of the passengers to construct an emotion recognition model sensitive to cultural dimensions; analyzing user historical driving data and emotional feedback to extract unique driving habits and preference features of the users; combining emotional evaluation values of the cultural layer and the individual layer to construct a cultural-individual double-layer emotion model; generating a personalized comfort degree score based on a current emotional state and individual preferences, and selecting the best driving decision in a multi-objective optimization mode; the application improves user acceptance and commercialization potential of an automatic driving system, and solves the technical problem that the automatic driving system lacks emotion state and driving preference perception ability in the prior art.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and more specifically, to methods, systems, and media for sharing driving decisions in autonomous vehicles. Background Technology

[0002] With the rapid development of artificial intelligence and sensor technology, autonomous driving technology has moved from the laboratory to the forefront of commercial applications. Currently, autonomous driving systems mainly focus on core technology modules such as perception, localization, planning, and control, which ensure that vehicles can drive safely and efficiently in complex environments. However, existing autonomous driving systems still have significant shortcomings in human-machine interaction and user experience.

[0003] Traditional autonomous driving decision-making systems are primarily optimized for safety and efficiency, employing preset driving parameters and decision rules, lacking consideration for passenger emotional states and personal preferences. This "vehicle-centric" design philosophy prevents the system from adjusting driving behavior based on real-time passenger feedback, resulting in a poor passenger experience and even human-machine decision-making conflicts. For example, the system might choose the shortest route or the fastest speed, but these choices could cause passenger discomfort or anxiety.

[0004] In the field of affective computing, although some technologies have attempted to apply emotion recognition to driving scenarios, these technologies typically employ single-modal emotion recognition methods, such as relying solely on facial expressions or voice features, making it difficult to comprehensively and accurately capture the complex emotional states of passengers. Furthermore, existing emotion recognition systems generally ignore the impact of cultural differences on emotional expression and driving preferences, failing to meet the demands of a globalized market.

[0005] Furthermore, existing technologies lack effective mechanisms to deeply integrate passenger emotional states with driving decision-making systems. Most systems only make simple adjustments when they detect extreme emotional states (such as fear or anger), failing to achieve fine-grained emotional responses and a personalized driving experience. This fragmented design makes it difficult for autonomous driving systems to gain user trust and acceptance.

[0006] Based on the above background, this invention proposes a driving decision-sharing method and system for autonomous vehicles, aiming to establish a human-machine collaborative decision-sharing mode through multi-channel emotion perception, culturally sensitive emotion recognition, and personalized driving preference learning, thereby improving the user experience and acceptance of autonomous driving systems. Summary of the Invention

[0007] This invention provides a method, system, and medium for sharing driving decisions in autonomous vehicles, addressing the technical problems in related technologies such as the lack of perception of passengers' emotional states and driving preferences, the inability to adjust driving behavior based on passengers' real-time emotional feedback, resulting in poor driving experience and human-machine decision-making conflicts.

[0008] This invention provides a method for sharing driving decisions in autonomous vehicles, comprising the following steps:

[0009] A multi-channel emotion perception system is constructed by collecting passengers' facial expressions, voice features, and physiological signal data through multiple sensors.

[0010] Extract passengers' cultural background characteristics and construct a culturally sensitive emotion recognition model;

[0011] Analyze users' historical driving data and emotional feedback to extract users' unique driving habits and preferences;

[0012] By combining the emotional assessment values ​​of the cultural and individual levels, a two-level emotional model of culture and individual is constructed. The two-level emotional model of culture and individual dynamically adjusts the weights of the cultural and individual levels in the final emotional assessment through an adaptive weight adjustment mechanism.

[0013] Based on the current emotional state and individual preferences, a personalized comfort score is generated, and a multi-objective optimization approach is used to select the best driving decision that satisfies both safety and functionality while maximizing passenger comfort.

[0014] In one specific implementation, the steps of constructing the multi-channel emotion perception system include:

[0015] Collect facial expression data, voice feature data, and physiological signal data;

[0016] The collected multimodal emotion data were preprocessed to extract facial action unit features, Mel frequency cepstral coefficient features, and physiological signal statistical features;

[0017] By using an attention-weighted fusion mechanism, emotional features from different modalities are fused to generate a unified emotional state vector.

[0018] In one specific implementation, the step of constructing a culture-sensitive emotion recognition model includes:

[0019] Construct a multicultural driving preference database that includes driving preference data from users with different cultural backgrounds;

[0020] Cultural feature vectors are extracted based on passengers' geographical location, language preferences, and personal profiles.

[0021] Construct a culturally sensitive weight adjustment function to map the cultural feature vectors to weight vectors for each feature dimension in the emotion recognition model;

[0022] The cultural sensitivity weight vector is combined with the sentiment feature vector to calculate the cultural sensitivity sentiment assessment value.

[0023] In one specific implementation, the step of extracting the user's unique driving habits and preferences includes:

[0024] Collect data on users' emotional responses, proactive intervention behaviors, and explicit feedback in different driving scenarios;

[0025] Extract acceleration preference, turning radius preference, following distance preference, lane change preference, speed stability preference, route selection preference, and other driving behavior characteristics from the collected historical interaction data;

[0026] Based on the extracted driving behavior features, a personalized driving preference model is constructed.

[0027] Based on the user's personalized driving preference vector and current driving decision, calculate the individual-level emotional assessment value.

[0028] In one specific implementation, the adaptive weight adjustment mechanism of the cultural-individual dual-level emotion model dynamically calculates the weights of the cultural layer and the individual layer based on the amount of historical interaction data of the user. For new users or users with less data, the cultural layer has a higher weight; for long-term users or users with abundant data, the individual layer has a higher weight.

[0029] In one specific implementation, the step of generating a personalized comfort score includes:

[0030] Based on comprehensive emotional assessment values ​​and current emotional state, the impact of specific driving decisions on passenger comfort is quantitatively evaluated.

[0031] Different weights are assigned to different dimensions of comfort assessment to form a comprehensive comfort score.

[0032] In one specific implementation, the multi-objective optimization method selects the optimal driving decision using the following formula:

[0033] The driving decision that maximizes the weighted sum of safety, efficiency, and comfort scores is selected, with the weights dynamically adjusted based on current road conditions, traffic environment, and passenger emotional state.

[0034] In one specific implementation, the driving decision sharing method for autonomous vehicles further includes the following steps:

[0035] Continuously monitor passengers' emotional responses;

[0036] A closed-loop feedback mechanism is formed based on the monitoring results;

[0037] Continuously optimize the driving decision-making process.

[0038] In one specific implementation, the driving decision sharing system for autonomous vehicles is used to execute a driving decision sharing method for autonomous vehicles, including:

[0039] A multi-channel emotion perception module is used to collect passengers' facial expressions, voice features, and physiological signal data through multiple sensors;

[0040] A culturally sensitive emotion recognition module is used to extract passengers' cultural background features and build corresponding emotion recognition models;

[0041] The individual driving style feature extraction module is used to analyze users' historical driving data and emotional feedback to extract users' unique driving habits and preferences.

[0042] The cultural-individual dual-level emotion model module is used to combine the emotion assessment values ​​of the cultural level and the individual level, and dynamically adjust the weights through an adaptive weight adjustment mechanism.

[0043] The personalized comfort rating and decision optimization module generates a comfort rating based on the current emotional state and individual preferences, and uses a multi-objective optimization approach to select the best driving decision.

[0044] In one embodiment, a computer-readable storage medium is provided for storing computer-readable instructions that, when read by a computer, enable the operation of a driving decision-sharing system for an autonomous vehicle.

[0045] The beneficial effects of this invention are as follows:

[0046] The multi-channel emotion perception system constructed in this invention can simultaneously collect and analyze passengers' facial expressions, voice features, and physiological signals, which greatly improves the accuracy and robustness of emotion recognition and enables the system to fully understand passengers' complex emotional states.

[0047] The culturally sensitive emotion recognition model introduced in this invention effectively solves the technical challenge of cross-cultural emotion understanding, enabling the system to adapt to the emotional expression patterns and driving preferences of passengers from different cultural backgrounds, and significantly enhancing the system's global adaptability.

[0048] The cultural individual dual-level emotion model proposed in this invention innovatively combines group cultural characteristics and individual preference characteristics. Through an adaptive weight adjustment mechanism, it achieves a smooth transition from "culture-based coarse-grained adaptation" to "individual-based refined customization", which greatly improves the service quality of the system for both new and old users.

[0049] The personalized comfort scoring mechanism and multi-objective optimization decision-making method designed in this invention incorporate passengers' emotional experience into decision-making considerations while ensuring safety and efficiency. This achieves human-machine sharing in driving decisions and effectively enhances passengers' trust and satisfaction with the autonomous driving system.

[0050] The technical solution of this invention continuously learns from passengers' emotional feedback and optimizes the driving decision-making process, forming a closed-loop feedback mechanism. This enables the system to continuously improve service quality as usage time increases, laying a solid foundation for the large-scale commercial application of autonomous driving technology. Attached Figure Description

[0051] Figure 1 This is a flowchart of the driving decision sharing method for autonomous vehicles according to the present invention;

[0052] Figure 2 This is a bar chart comparing the accuracy of multimodal sentiment data fusion according to the present invention;

[0053] Figure 3 This is a radar chart showing the impact of the cultural dimension of this invention on driving style preferences;

[0054] Figure 4 This is a line graph showing the sentiment prediction accuracy of the cultural individual dual-level model of this invention and the traditional model;

[0055] Figure 5 This is a line graph showing the improvement in user satisfaction caused by the multi-objective optimization decision-making method of this invention. Detailed Implementation

[0056] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0057] At least one embodiment of the present invention discloses a method for sharing driving decisions for autonomous vehicles, such as... Figure 1 As shown, it includes the following steps:

[0058] Step 1: Collect passengers' facial expressions, voice features, and physiological signal data through multiple sensors to construct a multi-channel emotion perception system;

[0059] Specifically, it includes the following sub-steps:

[0060] Step 1.1: Collect multimodal sentiment data;

[0061] The system collects multimodal emotion data, including facial expressions, vocal features, and physiological signals.

[0062] Specifically:

[0063] Facial expression data: The system collects key facial information of passengers through in-vehicle high-definition cameras, including the position and movement trajectory of facial features such as eyebrows, eyes, and corners of the mouth;

[0064] Voice feature data: Passenger voice data is collected through a microphone array, and acoustic features such as pitch, volume, speech rate, and voice tremor are extracted;

[0065] Physiological signal data: Physiological signal data such as changes in passenger posture, body temperature and heart rate are collected through pressure sensors, temperature sensors and optional heart rate monitoring devices built into the seat.

[0066] The output is the original multimodal sentiment dataset D. raw It contains sentiment data from different modalities.

[0067] Step 1.2: Preprocess multimodal sentiment data;

[0068] The system processes the collected raw multimodal sentiment data D raw Preprocessing is performed, including noise reduction, feature extraction, and standardization.

[0069] For facial expression data, facial action unit (FAU) features are extracted;

[0070] For sound feature data, Mel-Frequency Cepstral Coefficients (MFCC) and fundamental frequency profile features are extracted;

[0071] For physiological signal data, statistical features and frequency domain features are extracted.

[0072] The preprocessing process can be represented as:

[0073] D processed =f preprocess (D raw );

[0074] Among them, f preprocess This represents a preprocessing function used to perform operations such as noise reduction, feature extraction, and standardization on the original data; D raw This represents raw, multimodal emotion data collected from multiple sensors, including facial expressions, vocal features, and physiological signal data; D processed This indicates that the preprocessed multimodal sentiment data has undergone noise reduction, feature extraction, and standardization, and can be directly used for subsequent sentiment recognition and analysis.

[0075] Step 1.3: Integrate multimodal emotional features to construct a real-time emotional state vector;

[0076] Furthermore, the system utilizes a multimodal feature fusion algorithm to fuse emotional features from different modalities, generating a unified emotional state vector. This implementation employs an attention-weighted fusion mechanism to adaptively adjust the weights of different modal features to adapt to changes in the reliability of each modality under varying circumstances.

[0077] In its implementation, the attention-weighted fusion mechanism dynamically adjusts the reliability of emotion recognition under different driving scenarios. For example, in low-light nighttime driving scenarios, facial expression recognition may be limited. In this case, the system automatically reduces the weight of facial features while increasing the weight of vocal features and physiological signals. Conversely, in noisy environments, the system reduces the weight of vocal features and increases the weight of other modalities. This adaptive adjustment ensures that the system maintains high accuracy in emotion recognition even when the quality of some sensor data deteriorates.

[0078] The formula for calculating the emotion state vector is as follows:

[0079]

[0080] Among them, e t This represents the emotional state vector at the current time t, containing various dimensions of the passenger's emotional characteristics; m modal This indicates the number of sentiment modalities, i.e., the total number of sentiment data types collected by the system; This represents the weight of the i-th emotional modality, used to adjust the importance of different modalities in the fusion process; This represents the feature extraction function for the i-th sentiment modality, used to extract features from the preprocessed data D. processed Extract the emotional features of this modality; D processed This refers to preprocessed multimodal sentiment data, including data that has undergone noise reduction, standardization, and other processing.

[0081] Weight Dynamically calculated using the following attention mechanism:

[0082]

[0083] in, The weight of the i-th sentiment modality is used to determine the importance of that modality in multimodal fusion; exp(q i ) represents the exponential function value of the attention score for the i-th modality; The sum of the exponential function values ​​of all modal attention scores is used for normalization to ensure that the sum of all weights is 1; q i The attention score for the i-th modality is obtained through a neural network; m modalexp represents the number of emotional modalities, such as the number of different modalities like facial expressions, vocal features, and physiological signals; exp represents the natural exponential function.

[0084]

[0085] Where, q i V represents the attention score for the i-th modality; V is the parameter vector of the attention mechanism, used to map intermediate features to scalar attention scores; v T W represents the transpose of vector v, used to perform vector dot product operations; modal It is a weight matrix used for linear transformation of modal features; b represents the feature vector of the i-th mode extracted from the preprocessed data; modal is the bias vector used to adjust the baseline value of the linear transformation; tanh is the hyperbolic tangent activation function, which compresses the feature values ​​to the range [-1,1], enhancing the model's nonlinear expressive power. These parameters are automatically learned during the model training process to maximize the accuracy of emotion recognition.

[0086] In some implementations, the multimodal feature fusion process can employ a more complex hierarchical fusion architecture. For example, the system can first perform internal fusion on similar perceptual data (such as facial expression data from multiple cameras), and then fuse features from different modalities at a higher level. Optionally, the system can also employ a multimodal fusion method based on graph neural networks, representing each modal feature as a graph node and emotional relationships as edges, thereby better capturing the interrelationships and dependencies between modalities.

[0087] Therefore, the output is a real-time sentiment state vector e. t This vector contains information about the passenger's current emotional state, such as the dimensions of pleasure, arousal, and sense of control.

[0088] Step 2: Extract the cultural background features of passengers and construct a culturally sensitive emotion recognition model;

[0089] Specifically, it includes the following sub-steps:

[0090] Step 2.1: Construct a multicultural driving preference database;

[0091] The system collects driving preference data from users with different cultural backgrounds to build a multicultural driving preference database. This database contains user preference data on various driving behaviors (such as acceleration, turning radius, following distance, etc.) from different cultural backgrounds.

[0092] Each record in the database can be represented as:

[0093] R i =(ci ,P i E i );

[0094] Among them, R i Let c represent the i-th record in the multicultural driving preference database, used to store the relationship between driving preferences and emotional responses within a specific cultural context; i This represents a cultural background feature vector, containing multiple dimensions describing cultural characteristics, used to quantify a user's cultural background features; P i This represents a set of driving behavior parameters, including acceleration, deceleration, turning radius, following distance, lane change frequency, and other driving behavior-related parameters, used as quantitative indicators to describe a specific driving style; E i This represents the corresponding set of emotional responses, including values ​​for emotional dimensions such as pleasure, arousal, sense of control, and comfort, used to record the user's emotional experience data under specific driving behaviors.

[0095] The output is a multicultural driving preference database (DB). culture .

[0096] Step 2.2, extract cultural feature vectors;

[0097] The system extracts cultural feature vectors based on passengers' geographical location, language preferences, and personal information. The cultural feature vector *c* contains multiple dimensions describing cultural differences, such as:

[0098]

[0099] Where c represents the cultural feature vector, used to describe the cultural background characteristics of passengers; c1, c2, ... They represent the 1st, 2nd, and kth respectively. cult One cultural dimension component; k cult It refers to the quantity of cultural dimensions.

[0100] These dimensional values ​​can be obtained from a pre-set cultural dimension database based on the passenger's country / region information, or they can be dynamically adjusted through interactive learning.

[0101] In some implementations, in addition to the Hofstadter cultural dimension, the system can integrate other cultural dimension models, such as Schwarz's value theory or Glober's cultural dimension, to obtain a more comprehensive representation of cultural characteristics. Optionally, the system can construct a hybrid cultural feature vector, extracting complementary features from multiple cultural dimension theories.

[0102] Therefore, the output is the cultural feature vector c of the passenger.

[0103] Step 2.3: Construct a culturally sensitive weight adjustment function;

[0104] In addition, the system constructs a culturally sensitive weight adjustment function to adjust the parameters of the emotion recognition model based on cultural feature vectors.

[0105] This function maps the cultural feature vector c to the weight vector w(c) of each feature dimension in the emotion recognition model:

[0106] w(c)=f culture (c);

[0107] Where w(c) represents the weight vector generated based on the cultural feature vector, used to adjust the importance of each feature dimension in the emotion recognition model; c represents the cultural feature vector, containing multiple dimension values ​​describing the passenger's cultural background; f culture It is a culturally sensitive mapping function responsible for converting cultural feature vectors into weight parameters for emotion recognition models. It can be implemented as a neural network or other machine learning model, with the aim of capturing the differences in emotion expression and understanding in different cultural contexts.

[0108] In the specific implementation, f culture It can be represented as:

[0109] f culture (c) = softmax(W) c ·c+b c );

[0110] Among them, W c It is a learnable parameter matrix used to map the cultural feature vector c to the weight space, capturing the degree of influence of different cultural dimensions on emotion recognition; b c is a learnable bias vector used to adjust the baseline value of the mapping, providing flexibility to the model; c is a cultural feature vector containing multiple dimensions describing the user's cultural background; the softmax function is a normalization function that converts the output into a probability distribution form, ensuring that the sum of all weights is 1, which facilitates the reasonable allocation of the importance of each feature in the emotion recognition process.

[0111] In practical applications, this culture-sensitive weighting function is specifically optimized for users in different regions. For example, for passengers from East Asian cultural backgrounds, the system assigns higher weights to smoothness and predictability in driving behavior based on the higher uncertainty avoidance tendency in that culture; while for passengers from North American cultural backgrounds, the system may place greater emphasis on personal control and reaction speed characteristics based on the higher individualism tendency in that culture. In practical applications, the model can automatically load the corresponding initial cultural parameters from the existing Hofstadter cultural dimension database, and then continuously refine and adjust them based on passenger feedback to form a more personalized set of culture-adaptive parameters.

[0112] Step 2.4, calculate the emotional assessment score for cultural sensitivity;

[0113] Finally, the system combines the culturally sensitive weight vector w(c) with the emotional feature vector to calculate the culturally sensitive emotional assessment value.

[0114] The formula for calculating the cultural sensitivity affective score is:

[0115]

[0116] Among them, E c (s, c) represents the evaluation value of emotional stimulus S under cultural context C, i.e., the comprehensive score of emotional response calculated based on a specific cultural context; s represents the emotional stimulus, referring to a specific driving behavior or decision; c represents the cultural feature vector, containing multiple dimensions describing the passenger's cultural background; n feat The number of dimensions representing emotional features, i.e., the total number of features used to describe emotional responses; w i (c) represents the weight of the i-th feature dimension under cultural background C, reflecting the difference in the degree of importance attached to each emotional feature by different cultural backgrounds; This represents the value of the emotional stimulus S on the i-th culturally relevant feature dimension, i.e., the original score of a specific driving behavior on each emotional dimension; This means that all emotional characteristic dimensions are weighted and summed to synthesize the emotional responses of each dimension into a single overall evaluation value.

[0117] In summary, the output is the culturally sensitive sentiment assessment value E. c (s, c), this assessment value reflects the emotional response of passengers to specific driving behaviors within a specific cultural context.

[0118] Step 3: Analyze users' historical driving data and emotional feedback to extract users' unique driving habits and preferences;

[0119] Specifically, it includes the following sub-steps:

[0120] Step 3.1: Collect user historical interaction data;

[0121] During the user's use of the autonomous vehicle, the system collects the user's historical interaction data, including:

[0122] User emotional response data in different driving scenarios;

[0123] User intervention in the autonomous driving system, such as taking over control and adjusting settings;

[0124] Explicit user feedback (such as ratings, reviews, etc.).

[0125] These data constitute the user history interaction dataset Hu It records users' emotional reactions and preferences under different driving decisions.

[0126] Step 3.2, extract driving behavior features;

[0127] The system extracts driving behavior features from the collected historical interaction data to form a user's driving behavior feature set.

[0128] These features include, but are not limited to:

[0129] Acceleration preference: A user's response to different levels of acceleration;

[0130] Turning radius preference: User response to different turning radii;

[0131] Following distance preference: User's response to different following distances;

[0132] Lane change preference: User's response to different lane change methods;

[0133] Speed ​​stability preference: The user's sensitivity to speed fluctuations;

[0134] Route selection preference: The user's response to different route selection strategies.

[0135] Driving behavior feature extraction can be represented as:

[0136] F d =f extract (H u );

[0137] Among them, F d This represents a set of driving behavior features, containing user preference feature vectors across multiple dimensions such as acceleration, turning radius, and following distance; f extract The feature extraction function is an algorithm or mapping relationship that transforms user historical interaction data into structured driving behavior features; H u This represents the user's historical interaction dataset, containing historical data such as the user's emotional reactions, proactive intervention behaviors, and explicit feedback in different driving scenarios. This formula describes the transformation process from raw user interaction data to structured driving behavior characteristics, and is a fundamental step in building personalized driving models.

[0138] Step 3.3: Construct a personalized driving preference model;

[0139] In addition, the system is based on the extracted driving behavior features F d We will build a personalized driving preference model that can predict users' emotional reactions and comfort ratings to different driving decisions.

[0140] This implementation uses a collaborative filtering algorithm based on tensor decomposition, which organizes users, driving behavior, and emotional responses into third-order tensors, and learns the implicit user preference features through tensor decomposition.

[0141] The personalized driving preference model can be represented as:

[0142] p = f predference (F d );

[0143] Where p represents the user's personalized driving preference vector, containing the user's degree of preference for different driving behavior characteristics; f preference The preference model function, F, is a mathematical function that maps driving behavior features to preference vectors. d This represents the driving behavior feature set, containing various driving behavior features extracted from the user's historical interaction data, such as acceleration preferences, turning radius preferences, and following distance preferences. This formula describes how a specific preference model function can be used to transform the user's driving behavior features into a quantified, personalized driving preference vector for subsequent driving decision optimization.

[0144] In the specific implementation, f preferance It can be represented as:

[0145] f preferance (F d ) = W p ·F d +b p ;

[0146] Among them, W p F represents the learnable parameter matrix used to transform driving behavior features into a preference vector, reflecting the degree of influence of different driving behavior features on user preferences; d This represents a user's driving behavior feature set, including the user's preference features across various driving dimensions; b p This represents a learnable bias vector used to adjust the model's baseline values, ensuring that reasonable preference predictions are generated even when the feature values ​​are zero; f preference (F d ) represents the output of the preference model function, i.e., the user's personalized driving preference vector.

[0147] In some implementations, the personalized driving preference model may employ nonlinear representation methods to better capture the complexity and diversity of user preferences. For example, the system can use deep neural networks to construct the personalized driving preference model, including multiple fully connected layers and nonlinear activation functions, to learn more complex user preference features. Optionally, the system may also employ an attention mechanism to focus on differences in user preferences for specific driving scenarios (such as congested roads, highways, mountain roads, etc.).

[0148] In other implementations, personalized driving preference models can incorporate reinforcement learning methods to continuously adjust and optimize model parameters based on real-time user feedback. For example, the system can use positive emotional responses as reward signals and negative emotional responses as penalty signals, optimizing driving decision-making strategies through reinforcement learning algorithms to gradually adapt to the user's personalized preferences.

[0149] Step 3.4: Calculate the individual-level emotional assessment score;

[0150] The system calculates individual-level emotional assessment values ​​based on the user's personalized driving preference vector p and the current driving decision d.

[0151] The formula for calculating individual-level affective assessment scores is as follows:

[0152] E p (a, p) = f individual (a, p);

[0153] Among them, E p (a, p) represents the emotional evaluation value of driving behavior a under individual preference p, that is, the emotional response score of a specific driving behavior based on the user's personal preference; a represents the current driving behavior, including multi-dimensional features such as acceleration, turning radius, and following distance; p represents the user's personalized driving preference vector, reflecting the strength of the user's personal preference in each driving dimension; f individual It is an individual-level sentiment assessment function used to match driving behavior characteristics with user preferences and output the user's sentiment assessment value for that driving behavior.

[0154] In the specific implementation, f individual It can be represented as:

[0155]

[0156] Where m represents the number of dimensions of driving behavior features, such as the total number of features like acceleration, turning radius, and following distance; p i This represents the user's preference value for the i-th driving behavior feature, reflecting the user's level of importance or intensity of preference for that feature; g i (a) represents the value of driving behavior a in the i-th feature dimension, that is, the specific manifestation of the current driving behavior in this feature dimension; f individual (a, p) represents the individual-level sentiment evaluation function. It calculates the user's overall sentiment evaluation of their current driving behavior by weighted summing of preference values ​​and actual driving behavior values ​​across each feature dimension. This formula achieves a quantifiable match between user-personalized preferences and actual driving behavior.

[0157] In summary, the output is the individual-level sentiment assessment value E. p (a, p), this evaluation value reflects a specific user's personalized emotional response to a specific driving behavior.

[0158] Step 4: Combine the emotional assessment values ​​of the cultural layer and the individual layer to construct a two-level emotional model of culture and individual. The two-level emotional model of culture and individual dynamically adjusts the weights of the cultural layer and the individual layer in the final emotional assessment through an adaptive weight adjustment mechanism.

[0159] Specifically, it includes the following sub-steps:

[0160] Step 4.1, design a two-level model structure;

[0161] The system designs a two-tiered emotional model of cultural individuals, including a cultural layer, an individual layer, and a fusion layer.

[0162] The cultural layer processes culturally relevant emotional assessments, the individual layer processes emotional assessments related to individual preferences, and the fusion layer integrates the assessment results from both layers.

[0163] The overall structure of the model can be represented as follows:

[0164] E combined (a, c, p) = f fusion (E c (a, c), E p (a, p));

[0165] Among them, E combined (a, c, p) represents the emotional assessment value of driving behavior a by combining cultural background c and individual preference p, that is, the final emotional response prediction value obtained after considering both cultural factors and personal preferences; f fusion This represents the fusion function, used to integrate the emotional assessment results from the cultural and individual levels to generate a final comprehensive assessment value; E c (a, c) represent the emotional assessment values ​​of the cultural level, reflecting the general emotional response to driving behavior within a specific cultural context; E p (a, p) represents the individual-level emotional assessment value, reflecting the personalized emotional response to driving behavior based on personal historical preferences.

[0166] Step 4.2: Implement adaptive layer weight adjustment;

[0167] The system implements an adaptive layer weight adjustment mechanism, which dynamically adjusts the weights of the cultural layer and the individual layer in the final evaluation based on the passenger's usage history and current context.

[0168] For new users or users with limited data, the cultural layer has a higher weight;

[0169] For long-term users or users with abundant data, the individual level has a higher weight.

[0170] In practice, the calculation formula for the cultural individual's two-level emotional model is as follows:

[0171] E combined (a, c, p) = α cult ·E c (a, c) + β indiv ·E p (a, p);

[0172] Among them, E combined (a, c, p) represents the total emotional assessment value after integrating cultural background and individual preferences; a represents the current driving behavior; c represents the cultural background vector; p represents the user's personalized driving preference vector; E c (a, c) represent the emotional assessment values ​​of the cultural level, reflecting the evaluation of driving behavior within a specific cultural context; E p (a, p) represents the individual-level affective assessment value, reflecting the evaluation of driving behavior based on personal preferences; α cult β indiv Let α represent the weight coefficients of the cultural layer and the individual layer, respectively, satisfying α cult +β indiv =1, ensuring that the sum of the weights of the two levels is 1.

[0173] The adaptive weights are calculated as follows:

[0174]

[0175] Where, α cult The weighting coefficient of the cultural layer in sentiment assessment decreases as user data increases; β indiv This represents the weighting coefficient of the individual level in sentiment assessment, which increases with the increase of user data; n data This indicates the amount of historical user interaction data, including the total amount of driving records and emotional feedback data from the user's past use of the system; γ scale It is a scaling parameter that controls the rate at which the weights change as the amount of data increases; a larger γ... scale This value will enable the system to shift from cultural-level assessment to individual-level assessment more quickly.

[0176] In practical applications, the cultural-individual dual-level sentiment model dynamically balances the influence of the cultural and individual levels based on the user's system usage time and data accumulation. For example, for new users just starting out with an autonomous driving system, the system primarily relies on the cultural level's assessment (higher alpha). cultThe system uses a universal sentiment assessment standard based on the user's country / region and language preferences, applying the corresponding cultural background. As users continue to use the system and accumulate more personal data, the weight of the individual layer will gradually increase (higher beta). indiv The system makes predictions based more on the user's historical emotional responses and preference characteristics.

[0177] This two-tiered structure is particularly suitable for cross-cultural travel scenarios. For example, when international business travelers use autonomous driving services in different countries, the system can retain the user's personal preferences while appropriately incorporating local cultural characteristics, providing a driving experience that is both familiar and adapted to the local environment. For instance, when users accustomed to driving on German highways use autonomous driving services in Asian cities, the system will retain the user's preference for higher speeds while appropriately increasing the emphasis on smoothness in congested traffic conditions.

[0178] Step 4.3: Generate a personalized comfort score;

[0179] In addition, the system is based on the comprehensive sentiment assessment value E(a,c,p) and the current sentiment state e. t It generates personalized comfort scores, quantitatively assessing the impact of specific driving decisions on passenger comfort.

[0180] The formula for calculating personalized comfort scores is:

[0181]

[0182] Where C(d) represents the comfort score of driving decision d, i.e., the quantitative assessment value of the comfort level of a specific driving decision; n represents the number of assessment dimensions, including multiple assessment indicators such as smoothness, speed perception, and cornering comfort; α i f represents the weight of the i-th dimension, reflecting its importance in the overall comfort assessment; i (e t p, d) represent the current emotional state e t Given individual preferences p, the scoring function of driving decision d on the i-th dimension; e t represents the passenger's current emotional state, such as calm, tense, or happy; p represents the passenger's individual preference vector, which includes the intensity of preference for each driving feature; d represents the driving decision to be evaluated, which includes multiple parameters such as speed, acceleration, and route selection.

[0183] Therefore, the output is a personalized comfort score C(d).

[0184] Step 5: Based on the current emotional state and individual preferences, generate a personalized comfort score, and use a multi-objective optimization approach to select the best driving decision that satisfies both safety and functionality while maximizing passenger comfort.

[0185] Specifically, it includes the following sub-steps:

[0186] Step 5.1: Generate a candidate driving decision set;

[0187] Based on current road conditions, traffic rules, and the target route, the system generates a set of candidate driving decision sets D = {d1, d2, ..., d...} that meet safety and functionality requirements. k}. Where D represents the set of all possible candidate driving decisions; d1, d2, d... k These represent the 1st, 2nd, and kth candidate driving decisions, respectively; k represents the total number of candidate driving decisions.

[0188] Step 5.2: Assess the comfort level of the candidate decisions;

[0189] The system processes each decision d in the candidate driving decision set. i A personalized comfort model was used to calculate the comfort score C(d). i ):

[0190]

[0191] Among them, C(d) i ) represents the comfort score of the i-th candidate driving decision, i.e., the quantitative evaluation value of the comfort level of that specific driving decision; n represents the number of evaluation dimensions, including multiple evaluation indicators such as smoothness, speed perception, and cornering comfort; α j f represents the weight coefficient of the j-th dimension, reflecting the importance of that dimension in the overall comfort assessment; j (e t p, d i () indicates the current emotional state e t Under individual preferences p, driving decision d i The scoring function on the j-th dimension; e t d represents the passenger's current emotional state, such as calm, tense, or happy; p represents the passenger's individual preference vector, containing the intensity of preference for each driving feature; i This represents the i-th driving decision to be evaluated, which includes multiple parameters such as speed, acceleration, and route selection.

[0192] Therefore, the output is a set of comfort scores for candidate decisions.

[0193] Step 5.3, Multi-objective optimization decision selection;

[0194] The system performs multi-objective optimization based on safety, functionality, and comfort scores to select the best driving decision.

[0195] This multi-objective optimization problem can be expressed as:

[0196]

[0197] Where, d * The optimal driving decision is the driving decision that the system ultimately chooses to execute; d i Represents the i-th candidate decision in the candidate driving decision set D; argmax represents the parameter value that maximizes the objective function; Safety(d i ) represents decision d i The security score quantifies the security level of the decision; Efficiency(d) i ) represents decision d i The efficiency score quantifies the driving efficiency of this decision, such as time to reach the destination, fuel / electricity consumption, etc.; C(d) i ) represents decision d i The comfort score is the personalized comfort assessment value calculated in the aforementioned steps; λ1, λ2, and λ3 represent the weight coefficients of safety, efficiency, and comfort, respectively, satisfying λ1+λ2+λ3=1, ensuring that the sum of the three weights is 1.

[0198] In practical applications, multi-objective optimization decision-making dynamically adjusts weight parameters based on different scenarios and passenger states. For example, in congested urban road environments, the system might increase the comfort score C(d). i The system assigns a weight λ3 to the safety score, providing a smoother and more comfortable driving experience; however, on highways or in emergency situations, the system will correspondingly increase the safety score. i The weight λ1 is used to ensure driving safety.

[0199] This optimized decision-making method is particularly effective when road conditions change in real time. For example, when a vehicle enters a city from a highway, the system automatically adjusts the weighting based on the change in road type, allowing the driving style to smoothly transition with the environment. When passenger anxiety is detected, such as in unfamiliar areas or in adverse weather conditions, the system increases the comfort weighting to provide a driving experience that better alleviates passenger anxiety.

[0200] Through this dynamic weight adjustment, the system can find the optimal balance between the environment and the emotional state of passengers while ensuring safety. It will neither sacrifice comfort for efficiency nor significantly extend the journey time due to excessive focus on comfort.

[0201] Step 5.4, update driving control parameters;

[0202] The system will select the best driving decision d *These parameters are converted into specific driving control parameters, such as accelerator pedal position, brake pedal position, and steering angle, and then sent to the vehicle control system for execution.

[0203] At the same time, the system continuously monitors passengers' emotional responses, forming a closed-loop feedback loop to continuously optimize the decision-making process.

[0204] In summary, the output results are the updated driving control parameters and execution results.

[0205] Application example of this implementation method:

[0206] Application scenario: Cross-cultural shared mobility services;

[0207] The autonomous driving decision-making sharing method of this invention was deployed on a shared mobility service platform in a major international metropolis.

[0208] The platform serves passengers from diverse cultural backgrounds every day, including local residents and international business travelers.

[0209] Autonomous vehicle fleets need to adapt to the diverse driving preferences and emotional needs of different passengers in order to provide a personalized riding experience.

[0210] Before system deployment, the platform faced the following challenges:

[0211] International travelers have a low acceptance of autonomous vehicles and often give negative reviews because the driving style does not meet their expectations.

[0212] The high rate of human intervention in autonomous driving mode affects service efficiency and safety.

[0213] The same vehicle configuration can result in significant differences in performance among users from different cultural backgrounds.

[0214] Individual users, even within the same cultural context, show significant differences in their preferences for driving styles.

[0215] Therefore, the platform decided to introduce the driving decision sharing method of the present invention to achieve an emotion-driven personalized driving experience.

[0216] Example of a multi-channel emotion perception system implementation:

[0217] Ride-sharing platforms have installed the following sensors on their autonomous vehicle fleets:

[0218] In-vehicle high-definition camera: installed above the central control screen and in the rearview mirror, capturing high-definition facial images at 120 frames per second;

[0219] Microphone array: Composed of 4 directional microphones, installed in the roof lining, providing 360° sound collection capability;

[0220] Seat sensors: Pressure and temperature sensors are embedded in the seat back and seat cushion, with a sampling frequency of 10Hz;

[0221] Optional biosensors: For higher-end models, heart rate monitoring sensors are installed in the steering wheel and armrests.

[0222] In a practical application, the system served a passenger of East Asian background who was using the service for the first time. After the vehicle started, the multi-channel emotion perception system began to operate:

[0223] Data collection phase: The system simultaneously collects passengers' facial expressions (micro-expressions such as raised eyebrows and upturned corners of the mouth), voice data (speaking tone, volume, and speaking speed), and physiological signals (posture changes, hand pressure, and heart rate).

[0224] Preprocessing stage: The system performs noise reduction and feature extraction on the collected raw data.

[0225] For facial images, extract 68 facial key points and their movement trajectories;

[0226] Extract the fundamental frequency profile and volume variation curve from the speech data;

[0227] For physiological signals, pressure distribution maps and heart rate variability indices are extracted.

[0228] Fusion Phase: The system integrates data from various modalities through an attention-weighted fusion mechanism. When the vehicle makes a rapid turn, the camera captures a slightly tense facial expression from the passenger, while the physiological sensors simultaneously detect a slight increase in heart rate and a gripping of the seat armrests. The fusion algorithm combines these features, assigning a weight of 0.3 to facial expressions, 0.2 to vocal features, and 0.5 to physiological signals, generating an emotional state vector of "mild anxiety."

[0229] Through this process, the emotional state vector generated by the system in real time includes dimensions such as pleasure (-0.2, indicating slight discomfort), arousal (0.4, indicating moderate alertness), and sense of control (-0.3, indicating slight unease about vehicle control).

[0230] Example of implementing a two-level emotional model for cultural individuals:

[0231] For passengers with the aforementioned East Asian background, the working process of the cultural individual two-level emotion model is as follows:

[0232] Cultural Feature Extraction: The system identifies passengers’ cultural background as East Asia from their booking information and loads the corresponding cultural feature vectors from the database, including dimensions such as uncertainty aversion index (0.85, indicating high) and individualism index (0.35, indicating low).

[0233] Cultural Sentiment Assessment: Based on the Hofstadter cultural dimension model, the system identifies the typical preference for this cultural background as "smooth and predictable driving style," especially in operations that may cause uncertainty, such as turning and lane changing. The culturally sensitive weight adjustment function adjusts the weights of smoothness and predictability to 0.4 and 0.3, respectively, while lowering the weight of reaction speed to 0.2.

[0234] Individual-level emotional assessment: Since this was the first time the system was used, there was no historical data available for the passenger; the system could only learn from real-time feedback collected during the current trip. After one acceleration maneuver, the system detected a slight increase in the passenger's pleasure level, inferring that the passenger responded well to moderate acceleration (1.2 m / s²). 2 It has a high acceptance rate.

[0235] Two-tiered adaptive weighting: As a new user, the system assigns a higher initial weight to the cultural level (α = 0.8) and a lower weight to the individual level (β = 0.2). As the journey progresses, the system continuously accumulates passenger emotional response data and dynamically adjusts this ratio.

[0236] Comprehensive Emotional Assessment: The system combines the assessment results from both layers to generate a comprehensive emotional assessment value for a specific driving decision. For example, for a turning maneuver, the comprehensive assessment value is: E(turning maneuver) = 0.8 × (-0.5) + 0.2 × (-0.2) = -0.44, indicating that the passenger may experience significant discomfort with the current turning maneuver.

[0237] Personalized comfort rating: The system further incorporates the current emotional state to generate a comfort rating for anticipated driving decisions. For example, for an upcoming lane change, the system calculates a comfort rating of 0.3 (ranging from -1 to 1), predicting that passengers will have a slight level of acceptance.

[0238] Example of multi-objective optimization decision selection implementation:

[0239] Based on passengers' emotional state and comfort ratings, the system needs to decide how to proceed when approaching a busy intersection. The system generates three candidate driving decisions:

[0240] Decision A: Quickly pass through the intersection with an acceleration of 2.0 m / s². 2 To pass through in the shortest possible time.

[0241] Decision B: Pass through at a moderate speed, with an acceleration of 1.2 m / s². 2 , and remain stable.

[0242] Decision C: Slow down and proceed with caution; acceleration is 0.8 m / s². 2 This allows for more reaction time.

[0243] The system performs multi-objective evaluation for each decision:

[0244] Safety rating:

[0245] Decision A: 0.75 (Slightly lower safety at busy intersections at higher speeds);

[0246] Decision B: 0.85 (Medium speed provides good safety);

[0247] Decision C: 0.95 (low speed provides the highest safety);

[0248] Efficiency rating:

[0249] Decision A: 0.90 (Passed with high efficiency);

[0250] Decision B: 0.75 (moderate throughput);

[0251] Decision C: 0.60 (low efficiency);

[0252] Comfort rating (based on passenger sentiment model):

[0253] Decision A: 0.40 (Passengers are not very comfortable with rapid acceleration);

[0254] Decision B: 0.85 (Passengers have a high tolerance for moderate acceleration);

[0255] Decision C: 0.70 (too slow may cause anxiety);

[0256] The system dynamically adjusts the weights based on the current situation: taking into account the passenger's slight anxiety and the busy traffic environment, the system sets the safety weight λ1 = 0.4, the efficiency weight λ2 = 0.2, and the comfort weight λ3 = 0.4.

[0257] Final weighted score:

[0258] Decision A: 0.4 × 0.75 + 0.2 × 0.90 + 0.4 × 0.40 = 0.63;

[0259] Decision B: 0.4 × 0.85 + 0.2 × 0.75 + 0.4 × 0.85 = 0.83;

[0260] Decision C: 0.4 × 0.95 + 0.2 × 0.60 + 0.4 × 0.70 = 0.78;

[0261] The system selects decision B (passing at a moderate speed) with the highest overall score and converts it into specific control parameters: accelerator pedal position 30%, steering wheel angle remains unchanged, and target speed is adjusted to 40km / h.

[0262] like Figures 2 to 5The figures show a comparison of the accuracy of multimodal sentiment data fusion; the impact of cultural dimensions on driving style preferences; the sentiment prediction accuracy of the cultural individual dual-level model and the traditional model; and the improvement of user satisfaction through multi-objective optimization decision-making.

[0263] Overall, this application establishes a human-machine collaborative decision-making sharing model by deeply integrating affective computing technology with autonomous driving decision-making systems. This effectively enhances the user acceptance and commercial potential of autonomous driving systems and is a key technological support for the development of autonomous driving technology to a higher level.

[0264] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A method for sharing driving decisions in autonomous vehicles, characterized in that, Includes the following steps: A multi-channel emotion perception system is constructed by collecting passengers' facial expressions, voice features, and physiological signal data through multiple sensors. Extract passengers' cultural background characteristics and construct a culturally sensitive emotion recognition model; Analyze users' historical driving data and emotional feedback to extract users' unique driving habits and preferences; By combining the emotional assessment values ​​of the cultural and individual levels, a two-level emotional model of culture and individual is constructed. The two-level emotional model of culture and individual dynamically adjusts the weights of the cultural and individual levels in the final emotional assessment through an adaptive weight adjustment mechanism. The formula for calculating the emotional assessment value of the cultural level is: ； in, Indicating cultural background The following emotional stimulation The assessment value is the overall score of emotional response calculated based on a specific cultural context; Indicates emotional stimulation, referring to specific driving behaviors or decisions; This represents a cultural feature vector, containing multiple dimensions describing the cultural background of passengers. The number of dimensions representing emotional features, that is, the total number of features used to describe emotional responses; Indicates the first Each characteristic dimension in cultural background The weighting of these weights reflects the differences in the degree of importance attached to various emotional characteristics by different cultural backgrounds; Indicates emotional stimulation In the The values ​​on each culturally relevant feature dimension, that is, the original scores of a specific driving behavior on each emotional dimension; This means that all emotional feature dimensions are weighted and summed to combine the emotional responses of each dimension into a single overall evaluation value; The formula for calculating the individual level's emotional assessment value is: ； in, Indicating individual preferences Below driving behavior The emotional assessment value, which is the score of the emotional response to a specific driving behavior based on the user's personal preferences; It represents the current driving behavior, including multi-dimensional features such as acceleration, turning radius, and following distance; This represents a user's personalized driving preference vector, reflecting the strength of the user's personal preferences across various driving dimensions. It is an individual-level sentiment assessment function used to match driving behavior characteristics with user preferences and output the user's sentiment assessment value for the driving behavior. ； in, The number of dimensions representing driving behavior characteristics, such as the total number of features like acceleration, turning radius, and following distance; Indicates the user's position in the first month. The preference value for each driving behavior feature reflects the degree of importance or intensity of the user's preference for that feature; Indicates driving behavior In the The value on each feature dimension, that is, the specific manifestation of the current driving behavior on that feature dimension; The emotional evaluation function at the individual level calculates the user's overall emotional evaluation value of the current driving behavior by weighted summing of the preference values ​​of each feature dimension and the actual driving behavior value. Based on the current emotional state and individual preferences, a personalized comfort score is generated, and a multi-objective optimization approach is used to select the best driving decision that satisfies both safety and functionality while maximizing passenger comfort.

2. The driving decision sharing method for autonomous vehicles according to claim 1, characterized in that, The steps for constructing a multi-channel emotion perception system include: Collect facial expression data, voice feature data, and physiological signal data; The collected multimodal emotion data were preprocessed to extract facial action unit features, Mel frequency cepstral coefficient features, and physiological signal statistical features; By using an attention-weighted fusion mechanism, emotional features from different modalities are fused to generate a unified emotional state vector.

3. The driving decision sharing method for autonomous vehicles according to claim 1, characterized in that, The steps for constructing a culture-sensitive emotion recognition model include: Construct a multicultural driving preference database that includes driving preference data from users with different cultural backgrounds; Cultural feature vectors are extracted based on passengers' geographical location, language preferences, and personal profiles. Construct a culturally sensitive weight adjustment function to map the cultural feature vectors to weight vectors for each feature dimension in the emotion recognition model; The cultural sensitivity weight vector is combined with the sentiment feature vector to calculate the cultural sensitivity sentiment assessment value.

4. The driving decision sharing method for autonomous vehicles according to claim 1, characterized in that, The steps for extracting users' unique driving habits and preferences include: Collect data on users' emotional responses, proactive intervention behaviors, and explicit feedback in different driving scenarios; Extract acceleration preference, turning radius preference, following distance preference, lane change preference, speed stability preference, route selection preference, and other driving behavior characteristics from the collected historical interaction data; Based on the extracted driving behavior features, a personalized driving preference model is constructed. Based on the user's personalized driving preference vector and current driving decision, calculate the individual-level emotional assessment value.

5. The driving decision sharing method for autonomous vehicles according to claim 1, characterized in that, The adaptive weight adjustment mechanism of the cultural-individual dual-level emotion model dynamically calculates the weights of the cultural layer and the individual layer based on the amount of historical user interaction data. For new users or users with less data, the cultural layer has a higher weight; for long-term users or users with abundant data, the individual layer has a higher weight.

6. The driving decision sharing method for autonomous vehicles according to claim 1, characterized in that, The steps for generating a personalized comfort score include: Based on comprehensive emotional assessment values ​​and current emotional state, the impact of specific driving decisions on passenger comfort is quantitatively evaluated. Different weights are assigned to different dimensions of comfort assessment to form a comprehensive comfort score.

7. The driving decision sharing method for autonomous vehicles according to claim 1, characterized in that, The multi-objective optimization method selects the best driving decision using the following formula: The driving decision that maximizes the weighted sum of safety, efficiency, and comfort scores is selected, with the weights dynamically adjusted based on current road conditions, traffic environment, and passenger emotional state.

8. The driving decision sharing method for autonomous vehicles according to claim 1, characterized in that, It also includes the following steps: Continuously monitor passengers' emotional responses; A closed-loop feedback mechanism is formed based on the monitoring results; Continuously optimize the driving decision-making process.

9. A driving decision sharing system for autonomous vehicles, used to execute the driving decision sharing method for autonomous vehicles according to any one of claims 1-8, characterized in that, include: A multi-channel emotion perception module is used to collect passengers' facial expressions, voice features, and physiological signal data through multiple sensors; A culturally sensitive emotion recognition module is used to extract passengers' cultural background features and build corresponding emotion recognition models; The individual driving style feature extraction module is used to analyze users' historical driving data and emotional feedback to extract users' unique driving habits and preferences. The cultural-individual dual-level emotion model module is used to combine the emotion assessment values ​​of the cultural level and the individual level, and dynamically adjust the weights through an adaptive weight adjustment mechanism. The personalized comfort rating and decision optimization module generates a comfort rating based on the current emotional state and individual preferences, and uses a multi-objective optimization approach to select the best driving decision.

10. A computer-readable storage medium, characterized in that, It is used to store computer-readable instructions that, when read by a computer, enable the driving decision sharing system of the autonomous vehicle as described in claim 9 to run.

Images