Atmosphere lamp control method and device changing along with environment, medium and product

Abstract

The invention discloses an atmosphere lamp control method and device changing along with environment, a medium and a product, and relates to the technical field of Internet of Vehicles. The method comprises the following steps: acquiring physiological state data of a driver and driving state data of a vehicle, and combining the physiological state data and the driving state data to obtain safety state assessment; image data of the environment outside the vehicle and emotional state data of people in the vehicle are obtained, and environment experience evaluation is obtained based on the image data and the emotional state data; based on the safety state assessment, determining a safety priority and a corresponding safety early warning instruction through a preset safety analysis model; performing preset atmosphere matching processing based on the environment experience evaluation, and generating a basic atmosphere adjustment instruction; and according to the safety priority, performing dynamic superposition or replacement of the safety early warning instruction on the basic atmosphere adjustment instruction, and generating an atmosphere lamp control instruction. By implementing the technical scheme provided by the invention, the problem of control logic splitting of the existing atmosphere lamp is effectively solved.

Description

Technical Field

[0001] This application relates to the technical field of vehicle networking, specifically to a method, device, medium, and product for controlling ambient lighting that changes with the environment. Background Technology

[0002] In existing technologies, in-vehicle ambient lighting systems are primarily used to enhance the aesthetic experience and comfort within the cabin. Their control methods are typically isolated and preset; for example, the driver manually selects a static color and brightness via the central control screen, or the vehicle switches to a corresponding fixed color scheme based on preset driving modes (such as Sport mode and Eco mode). Some more advanced solutions may link the ambient lighting to a single vehicle status signal, such as causing the ambient lighting to flash red synchronously when emergency braking or forward collision warning is triggered, serving as a warning. These control logics are implemented based on independent, predefined rule sets, handling comfort adjustments and functional warnings separately.

[0003] However, the aforementioned existing technologies have a significant technical problem in controlling ambient lighting. Because the control logic for creating the atmosphere and the control logic for safety warnings operate independently and separately, the system cannot respond uniformly and intelligently to complex driving scenarios. Specifically, when a scenario requiring a comfortable atmosphere (such as smooth cruising on an urban expressway at night) coexists with a potential safety risk (such as the system detecting early signs of driver fatigue), the existing system lacks an effective arbitration and integration mechanism. The system either continues to maintain the preset comfortable atmosphere while ignoring the potential safety needs, or it is abruptly and completely taken over by the safety warning logic, prematurely disrupting the overall driving experience and failing to achieve a smooth, hierarchical integration between safety warnings and the ambient atmosphere. Summary of the Invention

[0004] To address the aforementioned technical issues, this application provides an ambient lighting control method, device, medium, and product that adapts to environmental changes.

[0005] The first aspect of this application provides a method for controlling ambient lighting that changes with the environment, employing the following technical solution: Acquire driver's physiological state data and vehicle driving state data, and combine the physiological state data and driving state data to obtain a safety status assessment; Acquire image data of the external environment and emotional state data of the occupants inside the vehicle, and obtain an environmental experience assessment based on the image data and the emotional state data; Based on the security status assessment, security priorities and corresponding security warning instructions are determined through a preset security analysis model. Based on the environmental experience assessment, a preset atmosphere matching process is performed to generate basic atmosphere adjustment instructions; Based on the safety priority, the basic atmosphere adjustment command is dynamically superimposed or replaced with the safety warning command to generate an ambient light control command.

[0006] By adopting the above technical solution, the problem of fragmented control logic in existing ambient lighting is effectively solved. First, the system simultaneously acquires the driver's physiological state (e.g., fatigue level) and the vehicle's driving status (e.g., speed, acceleration) to comprehensively assess safety risks. Simultaneously, it combines external environmental images (e.g., road type, lighting conditions) and occupant emotional data to generate an environmental experience assessment. Then, based on the safety status assessment, a safety analysis model determines safety priorities and warning commands, and based on the environmental experience assessment, basic atmosphere adjustment commands are generated. Crucially, the system dynamically overlays (e.g., slightly incorporates warning colors) or replaces (e.g., full-screen red) warning commands on the basic atmosphere commands according to safety priorities, achieving hierarchical integration of safety warnings and environmental atmosphere.

[0007] Optionally, the step of acquiring the driver's physiological state data and the vehicle's driving state data, and combining the physiological state data and driving state data to obtain a safety status assessment includes: The driver's first facial image sequence is acquired, and the eyelid opening and closing frequency and gaze direction vector are extracted from the first facial image sequence to form the physiological state data; The vehicle's steering wheel angular rate, brake pedal pressure value, and vehicle acceleration are acquired to form the vehicle's driving state data. Within a preset time window, the physiological state data and the driving state data are vectorized and combined to obtain a multidimensional state vector sequence, and the trajectory morphology features of the multidimensional state vector sequence are calculated. The safety status assessment is obtained by concatenating the trajectory morphology features with the corresponding physiological state data and driving state data within the preset time window.

[0008] By adopting the above technical solutions, the accuracy and foresight of safety status assessments are improved. Specifically, by analyzing continuous facial image sequences such as the frequency of eyelid opening and closing and the direction of gaze of the driver, early physiological signs of fatigue or distraction can be accurately captured; at the same time, vehicle dynamic data such as steering wheel turning rate and brake pedal pressure objectively reflect the smoothness and urgency of driving operations. The key innovation lies in vectorizing and combining these two types of heterogeneous data within a time window and extracting their trajectory morphological features. This is no longer an isolated analysis of instantaneous data points, but rather the identification of potential risks from the dynamic changing trends of behavioral sequences.

[0009] Optionally, acquiring image data of the external environment and emotional state data of occupants, and obtaining an environmental experience assessment based on the image data and the emotional state data, includes: Acquire images of the exterior scene within a preset range of the vehicle, perform semantic segmentation on the exterior scene images, and extract preset environment category features corresponding to the exterior scene images to form the image data; The vehicle is used to detect and identify all occupants, and voice signals and second facial image sequences of occupants other than the driver are acquired. For each person in the vehicle, facial expression features are extracted from the first facial image sequence or the second facial image sequence, and corresponding acoustic features are extracted from the speech signal. A corresponding emotion vector is generated through a preset multimodal emotion recognition model. All the emotional vectors are aggregated to obtain the emotional state data, which represents the overall emotional state of the people in the vehicle. The image data is integrated with the emotional state data to generate the environmental experience assessment.

[0010] By adopting the above technical solutions, a comprehensive and accurate quantitative assessment of the environmental experience is achieved. Specifically, the system not only identifies the categories of the external environment (such as highways and city night scenes) through semantic segmentation, providing objective scene basis for atmosphere creation, but more importantly, it innovatively adopts multimodal fusion technology to comprehensively analyze the facial expressions and voice features of all occupants (including drivers and passengers), generating accurate emotion vectors, and further aggregating them into data representing the overall emotional state. This multimodal method effectively avoids misjudgments from a single data source (such as relying solely on facial expressions), improving the robustness of emotion recognition. Ultimately, by integrating accurate environmental semantics with the true overall emotional state of the occupants, the generated "environmental experience assessment" not only reflects external objective conditions but also profoundly aligns with the subjective feelings of the occupants.

[0011] Optionally, the step of aggregating all the emotion vectors to obtain the emotion state data includes: Separate the driver's emotional vector corresponding to the driver and the passenger's emotional vector set corresponding to the other occupants from all the emotional vectors; Based on the driver's emotional polarity score of the driver's emotional vector and the average emotional polarity score of the passenger's emotional vector set, it is determined whether an emotional conflict state exists; When the emotional conflict state exists, the driver's emotional vector is used as the emotional state data; When the emotional conflict state does not exist, the driver's emotional vector and the passenger's emotional vector set are weighted and fused with preset weights to obtain a fused emotional vector, and the fused emotional vector is used as the emotional state data.

[0012] By adopting the above technical solution, the problem of conflicting atmosphere requirements in multi-occupant scenarios is effectively solved, ensuring the rationality of environmental experience assessment and prioritizing driving safety. When the system identifies a significant difference in emotional polarity between the driver and passengers (e.g., the driver is tense while the passengers are cheerful), it prioritizes adopting the driver's emotional vector as the overall emotional state. This design ensures that in potentially risky scenarios, atmosphere adjustment prioritizes serving the core needs of stabilizing the driver's state and ensuring driving safety. Conversely, when emotional states are consistent, a weighted fusion is used to generate the overall emotion, taking into account the experience of all passengers.

[0013] Optionally, determining whether an emotional conflict exists based on the driver's emotional polarity score of the driver's emotional vector and the average emotional polarity score of the passenger emotional vector set includes: The driver's emotion vector is multiplied by a preset first polarity weight vector to obtain the driver's emotion polarity score. The average passenger emotion vector is obtained by averaging all passenger emotion vectors in the set of passenger emotion vectors. The average passenger emotion vector is then multiplied by the inner product of the average passenger emotion vector and the preset second polarity weight vector to obtain the average emotion polarity score. When the product of the driver's emotional polarity score and the average emotional polarity score is less than a preset negative conflict threshold, and the absolute values ​​of both the driver's emotional polarity score and the average emotional polarity score are greater than a preset emotional intensity threshold, the emotional conflict state is determined to exist.

[0014] By adopting the above technical solution and introducing a quantitative mathematical model to accurately define the state of emotional conflict, the arbitration decision-making process becomes more objective and reliable, effectively avoiding misjudgments that may be caused by subjective rules. First, by using a preset polarity weight vector, the emotional polarity scores of the driver and passenger groups are calculated separately, transforming the abstract emotional vector into a comparable scalar, laying the foundation for conflict detection. Second, the conflict determination conditions are designed to be highly targeted, requiring not only that the emotional polarities of both parties be opposite (the product being negative), but also that the intensity of each party's emotion exceed a threshold. This ensures that a conflict state is triggered only when both parties hold clear and opposing emotions (such as the driver being intensely anxious while the passenger is highly excited), preventing false conflicts caused by slight emotional fluctuations or noise.

[0015] Optionally, the step of dynamically superimposing or replacing the safety warning instructions on the basic atmosphere adjustment instructions according to the safety priority to generate ambient light control instructions includes: When the security priority is the first preset security level, the warning color of the security warning instruction is superimposed on the color parameter of the basic atmosphere adjustment instruction with a preset transparency to obtain the blended color parameter, and the ambient light control instruction is generated based on the blended color parameter. When the safety priority is the second preset safety level, the brightness parameter of the basic atmosphere adjustment command is reduced by a preset amount to determine the reference brightness parameter. The brightness change waveform parameter and brightness change frequency parameter in the safety warning command are combined with the reference brightness parameter to generate the ambient light control command. When the security priority is the third preset security level, the ambient light control command is generated based on the parameters of the security warning command.

[0016] By adopting the above technical solution, a seamless transition from gentle prompts to emergency intervention in safety warnings and atmosphere is achieved, effectively solving the abrupt, either-or switching problem in traditional systems. The technical effects are as follows: When the safety priority is at the lowest level (Level 1), a color overlay method is used to incorporate warning colors while maintaining the original atmosphere, achieving early warning without compromising comfort; when the intermediate Level 2 safety priority is triggered, the base brightness is reduced and a dynamic waveform is overlaid, significantly improving the warning effect while retaining some atmosphere elements, forming a moderately intense reminder; and when the highest Level 3 safety priority is reached, it completely switches to a warning command, ensuring that the driver's attention is quickly drawn in an emergency.

[0017] Optionally, the method further includes: Obtain the planned driving route information of the vehicle and the map data corresponding to the planned driving route information; Based on the planned driving route information and the map data, preset high-risk driving scenarios on the vehicle's driving route are identified; When the distance between the vehicle and the high-risk driving scenario is detected to be less than the preset trigger distance, the safety priority is combined with the preset risk escalation level to obtain the target safety priority; Based on the target safety priority, the basic atmosphere adjustment command is dynamically superimposed or replaced with the safety warning command to generate the ambient light control command.

[0018] By adopting the above technical solution, a significant upgrade has been achieved in the ambient lighting control system, moving from passive response to proactive early warning. The system can proactively identify pre-defined high-risk scenarios (such as sharp bends, ramps, or construction zones) ahead based on the planned route and map data, and actively increase safety priority based on distance thresholds when the vehicle approaches these areas. This location-based predictive judgment allows the system to initiate early warning fusion before potential risks actually occur, such as gradually increasing the warning hue in the ambient lighting or initiating dynamic flashing before entering a sharp bend, providing the driver with valuable preparation time.

[0019] A second aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any of the foregoing.

[0020] A third aspect of this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described in any of the preceding descriptions.

[0021] A fourth aspect of this application provides a computer program product that, when run on an electronic device, causes the electronic device to perform the method as described in any of the preceding claims.

[0022] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: By constructing a dual analysis system encompassing safety status assessment and environmental experience assessment, and introducing a dynamic fusion mechanism based on safety priorities, the core problem of fragmented ambient lighting control logic in existing technologies is effectively solved. The system comprehensively utilizes multi-source data, including driver physiological state, vehicle driving status, external environmental semantics, and multimodal emotions of occupants, to achieve a holistic perception of the driving scenario. Temporal trajectory analysis enhances the foresight of safety risk assessment; an emotional conflict arbitration mechanism ensures the rationality of decision-making in multi-person scenarios; a tiered response strategy achieves a smooth transition from gentle overlay to emergency switching between warnings and atmosphere; and path prediction upgrades from passive response to proactive early warning. Ultimately, in complex driving scenarios, the system intelligently arbitrates safety and comfort needs, achieving tiered and seamless integration of safety warnings and environmental atmosphere, significantly improving driving safety while maximizing the continuity and comfort of the driving experience. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the system architecture of an embodiment of an ambient lighting control method that adapts to environmental changes, according to this application. Figure 2 This is a flowchart illustrating an ambient lighting control method that adapts to environmental changes, as disclosed in an embodiment of this application. Figure 3 This is another schematic flowchart of an ambient lighting control method that adapts to environmental changes, as disclosed in an embodiment of this application. Figure 4 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application.

[0024] Explanation of reference numerals in the attached figures: 100, System architecture; 101, First terminal device; 102, Second terminal device; 103, Third terminal device; 104, Network; 105, Server; 401, Processor; 402, Communication bus; 403, User interface; 404, Network interface; 405, Memory. Detailed Implementation

[0025] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0026] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0027] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0028] Figure 1 This is a schematic diagram of the system architecture of an embodiment of an ambient lighting control method that adapts to environmental changes, according to this application.

[0029] like Figure 1As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0030] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as model training applications, video recognition applications, web browser applications, social platform software, etc.

[0031] Terminal devices 101, 102, and 103 can be either hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices with displays, including but not limited to smartphones, tablets, e-book readers, MP3 (Moving Picture Experts Group Audio Layer III) players, MP4 (Moving Picture Experts Group Audio Layer IV) players, laptops, and desktop computers, etc. When terminal devices 101, 102, and 103 are software, they can be installed in the aforementioned electronic devices. They can be implemented as multiple software programs or software modules (e.g., multiple software programs or software modules used to provide distributed services) or as a single software program or software module. No specific limitations are imposed here.

[0032] This embodiment discloses an ambient lighting control method that adapts to changes in the environment. Figure 2 This is a flowchart illustrating an ambient lighting control method that adapts to environmental changes, as disclosed in an embodiment of this application. Figure 2 As shown, the method includes the following steps: S201. Obtain the driver's physiological state data and the vehicle's driving state data, and combine the physiological state data and driving state data to obtain a safety status assessment. "Physiological state data" refers to a series of indicators that can quantitatively characterize the driver's current physiological and mental state, such as, but not limited to, eyelid opening and closing frequency and gaze direction vector obtained through in-vehicle cameras, or heart rate and heart rate variability collected through wearable devices. Its purpose is to monitor the driver's fatigue, distraction, or tension in real time. "Vehicle driving status data," on the other hand, refers to data obtained from the vehicle's bus (such as the CAN bus) that objectively reflects the vehicle's dynamic characteristics and the driver's driving behavior, such as steering wheel angle rate, brake pedal pressure value, and vehicle acceleration, used to assess the smoothness and urgency of driving operations. "Safety status assessment" is not simply a list of information, but a comprehensive judgment result obtained by deeply integrating the above two heterogeneous data sources. This assessment can be a multi-dimensional feature vector or a quantitative risk score.

[0033] Optionally, acquiring the driver's physiological state data and the vehicle's driving state data, and combining the physiological state data and driving state data to obtain a safety status assessment includes: acquiring a first facial image sequence of the driver, and extracting eyelid opening and closing frequency and gaze direction vectors from the first facial image sequence to constitute the physiological state data; acquiring the vehicle's steering wheel turning rate, brake pedal pressure value, and vehicle acceleration to constitute the vehicle's driving state data; within a preset time window, vectorizing and combining the physiological state data and the driving state data to obtain a multidimensional state vector sequence, and calculating the trajectory morphology features of the multidimensional state vector sequence; and concatenating the trajectory morphology features with the corresponding physiological state data and driving state data within the preset time window to obtain the safety status assessment.

[0034] Specifically, the process of acquiring the driver's first facial image sequence and extracting physiological state data from it may include, but is not limited to, the following feasible embodiments. In the first embodiment, the system analyzes the facial images acquired by the camera, uses a facial key point detection algorithm (such as the Dlib library) to calculate the aspect ratio of the eyes to quantify the eyelid opening and closing state, and statistically analyzes its variation to obtain the eyelid opening and closing frequency. Simultaneously, it combines the head posture and pupil position calculated by a head posture estimation algorithm (such as PnP) to estimate the gaze direction vector. Finally, the frequency value and the vector are treated as a data pair or structure to constitute the physiological state data in this embodiment. As an alternative, in the second embodiment, the system inputs the acquired image sequence into a pre-trained convolutional neural network. This network can directly regress the quantized value representing the degree of eyelid opening and closing and the gaze direction vector. The system performs temporal analysis on the former to obtain the eyelid opening and closing frequency, and then packages the frequency and the latter (the gaze direction vector) into a unified feature vector as the physiological state data. In a third, more preferred embodiment, the system uses a time-of-flight camera to acquire a three-dimensional point cloud of the face. It accurately determines the eyelid opening and closing state and calculates the frequency by directly measuring the physical distance between the upper and lower eyelid point clouds. Simultaneously, it calculates a high-precision gaze direction vector by fitting a three-dimensional eyeball model. Finally, these two precise parameters are combined to form a high-dimensional and information-rich physiological state data set. Furthermore, the system accesses the vehicle's Controller Area Network (CAN) bus and reads relevant messages in real time at a preset high frequency (e.g., 50Hz or 100Hz). Specifically, the steering wheel angle rate can be obtained by performing real-time differential calculation (i.e., time derivative) on the steering wheel angle sensor signal value broadcast on the CAN bus; the brake pedal pressure value can be directly read from the CAN messages of vehicles equipped with electronic stability control systems or anti-lock braking systems, which typically contain precise values ​​measured by the brake master cylinder pressure sensor; vehicle acceleration, especially longitudinal and lateral acceleration, is usually measured by the vehicle's onboard inertial measurement unit and broadcast via the CAN bus. As an optional implementation, when the vehicle's CAN bus does not directly provide brake pressure values, the system can read the travel signal from the brake pedal displacement sensor and calculate the corresponding brake pedal pressure value using a pre-calibrated "travel-pressure" function model or lookup table. In another optional embodiment, for vehicles that cannot obtain data via the CAN bus or can only acquire partial data, the system can utilize external sensor modules for data acquisition. For example, an independent inertial measurement unit can be installed on the vehicle to acquire triaxial acceleration; or a force sensor externally mounted on the brake pedal can be used to directly measure brake pressure; or an angular velocity sensor mounted on the steering column can be used to directly acquire the steering wheel angular rate. Ultimately, the system combines these three key parameters (steering wheel angular rate v_steer, brake pedal pressure p_brake, and vehicle acceleration a_vehicle) collected at the same timestamp or within a very short time window, for example, to form a three-dimensional vector [v_steer, p_brake, a_vehicle], thus creating driving status data that can reflect the vehicle's current driving intentions and dynamic responses in real time.

[0035] Furthermore, within a preset time window (e.g., 3 seconds or other configurable durations as needed), the system synchronizes or aligns the aforementioned acquired physiological state data with the vehicle driving state data, and performs vectorized combination at each sampling moment to form a multidimensional state vector sequence. In a preferred implementation, to eliminate the influence of different physical dimensions, each data component can be normalized before combination. The combined sequence forms a trajectory in the multidimensional state space. The core of this embodiment lies in calculating the features that characterize the shape of this trajectory. The calculation methods for this feature include, but are not limited to, the following: The first approach is a statistical method, which calculates the mean, variance, kurtosis, and total length of the trajectory sequence across all dimensions (i.e., the sum of Euclidean distances between points in the sequence). The second approach is a geometric method, which calculates the average curvature, degree of twist, or the volume of its minimum boundary hyperrectangle to quantify the smoothness and spatial extent of the trajectory. The third approach is a frequency domain analysis method, which performs a Fourier transform on each dimension of the sequence and extracts its spectral energy, dominant frequency, and other features to reveal the periodic patterns of driver behavior and vehicle state changes. Any one or more of the above-mentioned features calculated together constitute the basis for subsequent judgments of the driver's state.

[0036] Furthermore, since the original physiological and driving state data are sequential data within the time window, a preferred approach to facilitate concatenation with a single trajectory morphology feature vector is to first extract features from these sequential data. For example, this involves calculating statistical measures such as the mean, variance, maximum, and minimum values ​​within the time window, thus converting the sequential data into a fixed-dimensional feature vector. Subsequently, the trajectory morphology feature vector, the feature vector of the physiological state data, and the feature vector of the driving state data are concatenated end-to-end to form a unified, high-dimensional fused feature vector. Finally, this fused feature vector is input into a pre-trained safety status assessment model. This model can be one of several machine learning models. For example, the first approach is a support vector machine, which excels at binary classification (e.g., fatigue / non-fatigue) in small sample situations; the second approach is a random forest or gradient boosting decision tree, these ensemble models are robust and can output multi-class results (e.g., alert, mild fatigue, severe fatigue) or risk probabilities; the third approach is a multilayer perceptron neural network, which can learn complex nonlinear relationships between features. The model outputs the final safety status assessment for this assessment period, such as a specific classification label or a continuous risk score.

[0037] S202. Acquire image data of the external environment and emotional state data of the occupants inside the vehicle, and obtain an environmental experience assessment based on the image data and the emotional state data; "Image data of the external environment" does not refer to the raw pixel information itself, but rather to structured data or labels that can characterize the state of the external world, parsed from the visual information of the vehicle's external scene. Examples include weather conditions (sunny, rainy), road types (highways, city streets), or surrounding landscape features (mountain views, building complexes). "Emotional state data of occupants" refers to a quantitative description of the driver's or passenger's inner psychological feelings, which can be expressed as discrete emotional categories (e.g., happy, calm, anxious) or coordinate values ​​on a continuous emotional dimension (e.g., valence-arousal model). "Environmental experience assessment" is a comprehensive evaluation indicator that aims to quantify the correlation between the external environment and the emotional state of occupants. The assessment result can be a qualitative level (e.g., positive, neutral, negative), a specific classification label (e.g., experiencing pleasure, experiencing depression), or a continuous numerical score, reflecting the harmony or quality of the current interaction between people and the environment.

[0038] Optionally, acquiring image data of the external environment and emotional state data of occupants, and obtaining an environmental experience assessment based on the image data and the emotional state data, includes: acquiring images of the external scene within a preset range of the vehicle, performing semantic segmentation on the external scene images, and extracting preset environmental category features corresponding to the external scene images to constitute the image data; detecting and identifying all occupants in the vehicle, and acquiring voice signals and second facial image sequences of occupants other than the driver; for each occupant, extracting facial expression features from the first facial image sequence or the second facial image sequence, and extracting corresponding acoustic features from the voice signal, generating a corresponding emotion vector through a preset multimodal emotion recognition model; aggregating all the emotion vectors to obtain the emotional state data, which represents the overall emotional state of the occupants; and integrating the image data and the emotional state data to generate the environmental experience assessment.

[0039] Specifically, one or more in-vehicle cameras (e.g., a front-view wide-angle camera, a dashcam, or a surround-view camera system) capture real-time images of the scene in front of or around the vehicle. After capture, the system calls a deep learning semantic segmentation model pre-deployed in the in-vehicle computing unit to process the image. Several technical solutions can be used to achieve this. The first solution is to use the U-Net model based on an encoder-decoder structure, whose skip connection structure helps preserve high-resolution detail information. The second solution is to use the DeepLab series models with a hollow spatial pyramid pooling module, which can capture contextual information at different scales. The core function of this model is to assign a predefined category label to each pixel in the image, thereby generating a pixel-level semantic segmentation map. Subsequently, the system parses this semantic segmentation map, counting the number of pixels belonging to different "preset environment categories" or their proportion of the total pixels. These preset environment categories may include, but are not limited to: sky, buildings, vegetation, roads, water bodies, vehicles, pedestrians, etc. Ultimately, the system combines these statistically derived proportions into a fixed-dimensional feature vector, such as [sky proportion, building proportion, vegetation proportion, ...]. This feature vector can quantitatively represent the basic composition of the current external visual environment, which is image data.

[0040] Furthermore, to comprehensively perceive the status of all occupants inside the vehicle, the system first scans the entire cabin space using one or more wide-angle cameras installed inside the vehicle (e.g., in the center of the roof or near the rearview mirror). By running a pre-set occupant detection algorithm, the system can locate and identify all occupants. One specific implementation is to use a deep learning-based object detection model (e.g., YOLO or SSD), which can output the bounding box position of each person in the image in real time. Another alternative is to use traditional face detection algorithms (e.g., a cascaded classifier based on Haar features) to locate each person's face. After identifying the occupant's location, the system collects mixed speech signals inside the vehicle through a microphone array and uses sound source localization and beamforming technology to direct the main pickup beam towards different speakers based on the determined occupant's location, thereby separating the independent speech signal belonging to each person. Simultaneously, for non-driver occupants identified as passengers, the system continuously extracts image frames from the video stream of the wide-angle cameras based on their located facial regions; these consecutive image frames constitute a second facial image sequence.

[0041] Furthermore, for facial image sequences, the system extracts expression features through a facial feature extraction module. One feasible approach is to directly extract high-dimensional deep expression feature vectors from image frames using a pre-trained convolutional neural network. Alternatively, the system can first locate facial landmarks (such as the corners of the eyes and mouth) using a facial landmark detection algorithm (e.g., methods from the Dlib library), and then calculate the relative distances, angles, and other geometric features between these points to construct expression features. Simultaneously, for the corresponding speech signal, the system extracts its acoustic features through an acoustic feature extraction module. In a preferred embodiment, the system extracts a set of low-order acoustic descriptors, including Mel-frequency cepstral coefficients, fundamental frequency, energy, and zero-crossing rate. In another alternative, the system can convert the speech signal into a spectrogram and input it as a two-dimensional image into another pre-trained convolutional neural network to automatically learn emotion-related time-frequency pattern features. Subsequently, the extracted expression features and acoustic features are simultaneously input into a pre-defined multimodal emotion recognition model, which is responsible for fusing information from both modalities to generate the final emotion vector. The model can be implemented in various ways: One approach is to simply concatenate the facial expression feature vector and the acoustic feature vector into a longer feature vector, which is then fed into a classifier (such as a support vector machine or a multilayer perceptron). Another approach is to use an attention-based fusion network that dynamically learns whether to prioritize visual or auditory information at different times, thus achieving more intelligent feature fusion. The final output emotion vector can be a one-hot encoding representing a discrete emotion category (such as happy, sad, or neutral), or coordinates in a continuous emotion space (such as a valence-arousal two-dimensional space), enabling a refined quantitative description of the emotional state.

[0042] One specific aggregation scheme employs a weighted average method, which involves weighted summation and normalization of all emotion vectors. In this scheme, different weights can be assigned to different individuals. For example, considering driving safety, the driver's emotion vector is given a higher weight (e.g., 0.6), while the remaining weights are evenly distributed among other passengers. The resulting weighted average vector is the emotion state data. Alternatively, the aggregation process can determine the dominant emotion. Specifically, the system can calculate the intensity of each emotion vector (e.g., the Euclidean distance from the vector to the neutral origin in the valence-arousal emotion space) and select the emotion vector with the highest intensity as the emotion state data representing the most prominent emotion. In some more complex embodiments, the emotion state data can also be a collection containing multi-dimensional statistical information. For example, this data may include not only the mean vector of all emotion vectors (representing the average emotion) but also their variance or standard deviation vectors (representing the consistency or dispersion of emotions within the vehicle), thus providing richer and more comprehensive decision-making basis for subsequent vehicle environment adjustment strategies.

[0043] Furthermore, the system first analyzes the image data to extract scene labels, such as city night scene, highway, sunny day, rainy day, or traffic congestion. Then, the rule engine performs an evaluation matching based on the combination of scene labels and emotional state data. For example, the rule can be preset as follows: when the scene label is coastal highway and the emotional state data is positive, a high-score evaluation labeled "pleasant journey" is generated; while when the scene label is traffic congestion and the emotional state data is negative or anxious, a low-score evaluation labeled "stressful commute" is generated. As a more advanced implementation, the integration process can be accomplished using a pre-trained machine learning fusion model (e.g., a multimodal neural network). This model takes feature vectors extracted from the image data (e.g., color saturation, scene complexity, weather parameters) and the emotional state data (e.g., valence and arousal components of the emotion vector) as multiple inputs and directly outputs a quantified environmental experience index (e.g., a score from 0 to 100) or a predefined experience category. The final environmental experience assessment, whether in the form of tags, scores, or more complex data structures, aims to provide a macroscopic and quantitative description of the current scene-emotional interaction state, providing a basis for decision-making in subsequent intelligent interactions or service recommendations.

[0044] Optionally, the step of aggregating all the emotion vectors to obtain the emotional state data includes: separating the driver's emotion vector corresponding to the driver and the passenger emotion vector set corresponding to the other occupants from all the emotion vectors; determining whether there is an emotional conflict state based on the driver's emotion polarity score of the driver's emotion vector and the average emotion polarity score of the passenger emotion vector set; when the emotional conflict state exists, using the driver's emotion vector as the emotional state data; when the emotional conflict state does not exist, performing a weighted fusion of the driver's emotion vector and the passenger emotion vector set with preset weights to obtain a fused emotional vector, and using the fused emotional vector as the emotional state data.

[0045] In a preferred embodiment of this application, to differentiate between the emotions of the driver and passengers, the system first needs to group these vectors after obtaining the emotion vectors of all occupants in the vehicle. This process relies on the identity identifier attached to each vector during the emotion vector generation stage. One implementation is that the system can identify the occupants based on their seating positions captured by the vehicle's cameras. For example, the person in the preset driver's seat (such as the left front seat) can be identified as the driver, while those in other seats can be identified as passengers. Based on this, their corresponding emotion vectors can be assigned to the driver's emotion vector set and the passenger's emotion vector set, respectively. Another more flexible approach is that the system can use facial recognition technology to assign a unique ID to each occupant and query a preset role database based on this ID (for example, the database records the ID of the vehicle owner or a user with driving privileges as the driver), thereby more accurately and dynamically completing the separation and grouping of emotion vectors.

[0046] Furthermore, the system first calculates a quantifiable and comparable emotional polarity score for each emotion vector. One specific calculation scheme is that if the emotion vector is a coordinate point in a valence-arousal two-dimensional emotion space, its valence component can be directly extracted as the emotional polarity score. This score is typically in the range [-1, 1], with positive values ​​representing positive emotions and negative values ​​representing negative emotions. Another feasible scheme is that if the emotion vector represents discrete emotion categories (such as happiness or sadness), the system can map different emotion categories to corresponding numerical scores according to a preset mapping table. After obtaining the score for each vector, the system calculates the arithmetic mean of all scores in the passenger emotion vector set. Subsequently, the system determines the conflict state by comparing the driver's emotional polarity score with the passenger's average emotional polarity score. For example, when the signs of these two scores are opposite, the system can determine that an emotional conflict exists. In a more refined judgment scheme, an emotional conflict is only determined to exist when the absolute difference between the two scores exceeds a preset conflict threshold (e.g., 0.5), to avoid misjudgments caused by minor emotional fluctuations.

[0047] Furthermore, when the system determines an emotional conflict state based on any of the aforementioned methods, the technical solution of this application will prioritize the driver's emotional state and driving safety. In this case, a selective operation will be performed, namely, directly selecting the driver's emotional vector corresponding to the driver and using it as the final output emotional state data, while temporarily ignoring the passenger's emotional state. The core technical idea of ​​this strategy is that when the driver and passenger's emotions are significantly inconsistent (e.g., the driver is tense due to road conditions, while the passenger is entertaining themselves), the vehicle's intelligent interaction system should prioritize responding to the driver's needs, such as helping the driver concentrate or alleviate negative emotions by adjusting voice prompts, changing ambient lighting tints, or lowering entertainment volume, rather than trying to cater to the passenger's positive emotions. Therefore, directly using the driver's emotional vector ensures that subsequent vehicle control and service adjustments always prioritize safety and driving experience.

[0048] Furthermore, if the system determines that there are no significant emotional conflicts within the vehicle, it indicates that the overall emotional atmosphere inside the vehicle is relatively harmonious or consistent. In this case, emotional fusion can more comprehensively and accurately reflect the overall in-vehicle experience. The system will then perform a weighted fusion of the driver's emotional vector and the passenger's emotional vector set. Specifically, the system can first perform a preliminary aggregation of all emotional vectors within the passenger emotional vector set. A simple aggregation method is to calculate their arithmetic mean to obtain an average passenger emotional vector. Subsequently, the system, according to preset weights, performs a weighted summation of the driver's emotional vector and this average passenger emotional vector to obtain the final fused emotional vector. For example, a higher weight (e.g., 0.6) can be assigned to the driver's emotional vector, and the remaining weight (e.g., 0.4) can be assigned to the average passenger emotional vector to reflect the driver's central role in the driving process. The final fused emotional vector will be output as emotional state data. The preset weights can be configured and adjusted according to vehicle type, user preferences, or the current vehicle driving mode (e.g., passenger weights can be appropriately increased when set to family travel mode), thereby flexibly balancing the emotional contributions of different passengers.

[0049] Optionally, determining whether an emotional conflict exists based on the driver's emotional polarity score of the driver's emotional vector and the average emotional polarity score of the passenger emotional vector set includes: performing an inner product operation on the driver's emotional vector and a preset first polarity weight vector to obtain the driver's emotional polarity score; performing a vector arithmetic mean operation on all passenger emotional vectors in the passenger emotional vector set to obtain an average passenger emotional vector, and performing an inner product operation on the average passenger emotional vector and a preset second polarity weight vector to obtain the average emotional polarity score; determining that the emotional conflict exists when the product of the driver's emotional polarity score and the average emotional polarity score is less than a preset negative conflict threshold, and the absolute value of both the driver's emotional polarity score and the absolute value of the average emotional polarity score are greater than a preset emotional intensity threshold.

[0050] Specifically, the system performs an inner product operation on the driver's emotion vector and a preset first polarity weight vector. The dimension of the polarity weight vector is the same as that of the emotion vector; essentially, it is a projection vector used to define the direction of emotion polarity, aiming to extract a scalar value representing the positive or negative tendency of the multidimensional emotion vector. For example, in a common embodiment where the emotion vector is represented by a two-dimensional coordinate system of [valence, arousal]—where valence is the core dimension measuring emotional pleasure, its value directly indicating the positive (e.g., joy, satisfaction) or negative (e.g., sadness, anger) attribute of the emotion; and arousal measures the intensity dimension of the emotion, its value representing the level of physiological activation (e.g., from calm to excitement)—this first polarity weight vector can be configured as [1, 0]. Under this configuration, the result of the inner product operation is equivalent to directly extracting the valence component as the polarity score, because valence itself is the most direct measure of the positive or negative nature of the emotion. As an optional, more refined scheme, this weight vector can also be configured as [1, -0.1]. This means that when assessing emotional polarity, the positive and negative information of valence is the primary factor, while the negative impact of arousal is given slight consideration. The underlying technical idea is that excessively high arousal, even with positive valence, can potentially interfere with driving tasks requiring concentration. Therefore, a slight penalty is applied when calculating the final polarity score. In this way, the system can flexibly customize the weight vector according to different emotional models and application scenarios, thereby accurately calculating the driver's emotional polarity score.

[0051] Similarly, when assessing the overall emotional polarity of the passenger group, the system employs a standardized processing procedure. First, the system aggregates all discrete passenger emotional vectors within the passenger emotional vector set into an average passenger emotional vector that represents the whole. A specific aggregation method is to take the arithmetic mean of each component of all passenger emotional vectors to obtain a composite vector. Then, using the same method as for driver emotional vectors, the system performs an inner product operation between this average passenger emotional vector and a preset second polarity weight vector to obtain the average emotional polarity score. To ensure consistency between the driver and passenger emotional polarity assessment standards, the second polarity weight vector is typically set to be the same as the first polarity weight vector. Of course, in some specially customized embodiments, they can be different to reflect a specific focus on different dimensions of passenger emotional polarity. This step ultimately outputs a scalar value that quantifies the average positive or negative emotional tendency of all passengers.

[0052] Furthermore, after obtaining the quantified emotional polarity scores for both the driver and passenger, the system ultimately determines whether an emotional conflict exists based on a complex logical condition with dual constraints. The first constraint is directional opposition: the system multiplies the driver's emotional polarity score by the average emotional polarity score and checks if the product is less than a preset negative conflict threshold (e.g., -0.05). Since the product can only be negative when the two scores have opposite signs (one positive and one negative), this condition effectively filters out situations where the driver and passenger's emotional states are fundamentally opposed. The second constraint is intensity significance: to avoid misjudgment due to weak (close to neutral) emotions from one or both parties, the system further requires that the absolute value of the driver's emotional polarity score and the absolute value of the average emotional polarity score must both be greater than a preset emotional intensity threshold (e.g., 0.3). This ensures that a meaningful conflict only exists when both parties exhibit sufficiently strong and opposite emotions. Only when both the directional opposition and intensity significance conditions are simultaneously met does the system determine that an emotional conflict exists.

[0053] S203. Based on the security status assessment, determine the security priority and corresponding security warning instructions through a preset security analysis model; Specifically, the core function of this safety analysis model is to establish a mapping from a complex, continuous, or discrete state space to a specific, hierarchical action space. One concrete implementation is that the model can be a multi-dimensional decision matrix or lookup table built based on expert knowledge. In this approach, developers pre-divide various foreseeable safety state assessment results and assign a unique safety priority (e.g., from Level 1: Emergency to Level 3: Alert) and associated safety warning instructions to each division or state combination (e.g., instruction code A01 corresponds to a high-frequency beeping sound and seat vibration, instruction code C05 corresponds to a "Please note the road conditions ahead" icon displayed on the central control screen). When the system is running, it only needs to use the current safety assessment result as an index to quickly and deterministically retrieve the corresponding priority and instruction. As a more flexible and adaptive advanced solution, the safety analysis model can be a pre-trained machine learning model, such as a decision tree, random forest, or a lightweight neural network. This model acquires decision-making capabilities by learning from a massive dataset of driving scenario data (containing various sensor data, driver states, and expert-annotated optimal handling solutions) during the development phase. In actual vehicle operation, the model receives real-time safety status assessment vectors as input and infers and outputs an optimal combination of (priority, instruction) in real time. The advantage of this approach lies in its ability to handle complex nonlinear relationships and its better generalization ability for edge scenarios not explicitly present in the training set, thus achieving a more accurate and intelligent safety response. Regardless of the approach adopted, the ultimate goal is to transform abstract risk assessments into concrete, executable intervention actions that match the risk level, providing a clear basis for subsequent instruction execution.

[0054] In one specific implementation, the security analysis model can be a threshold judgment system. For example, when the security status is assessed as a risk score between 0 and 1, the following can be preset: a score in the range [0.7, 1.0] corresponds to a third preset security level; a score in the range [0.4, 0.7) corresponds to a second preset security level; and a score in the range (0.1, 0.4) corresponds to a first preset security level. S204. Based on the environmental experience assessment, perform preset atmosphere matching processing to generate basic atmosphere adjustment instructions; Specifically, the relatively abstract environmental experience assessment results output from the previous step (e.g., assessment of driving on a mountain road at dusk, with the driver in a stable mood, or experiencing slight fatigue while viewing city lights) are transformed into a set of specific, executable instructions to adjust the in-vehicle environmental equipment and create a matching atmosphere. One concrete implementation is that the in-vehicle system pre-stores multiple atmosphere theme configuration files. Each configuration file corresponds to one or a type of environmental experience assessment result and defines in detail the target states of a series of subsystems under that theme. For example, a "Tranquil Night Flight" theme might define {Ambient Lights: Dark Blue, Brightness 20%; Music: Playing a soft instrumental playlist, Volume 15%; Air Conditioning: 23°C, Gentle Breeze Mode; In-Vehicle Fragrance: Releasing a subtle woody scent}. When the environmental experience assessment results received by the system match the trigger conditions for "Tranquil Night Flight," the configuration file is loaded, and its parameters are parsed into a set of basic atmosphere adjustment instructions, which are then sent to the lighting controller, multimedia host, air conditioning controller, and fragrance system, respectively. As another more dynamic and personalized implementation, this atmosphere matching process can be a system based on a rule engine or weighted algorithm. Instead of relying on a fixed overall theme, this system breaks down environmental experience assessment results into multiple dimensions (such as time, geography, music preference, and mood), and sets independent adjustment logic for each dimension. For example, the evening characteristic in the assessment results will trigger the lighting system to adjust to warmer tones and lower brightness; the mountain road characteristic may trigger the music system to play more rhythmic music to improve driving focus; and the calm mood characteristic will ensure that all adjustments (such as volume and airflow) are gradual rather than abrupt. Ultimately, the adjustment instructions generated by these independent logics are aggregated to form the final basic atmosphere adjustment instruction set.

[0055] S205. Based on the safety priority, dynamically superimpose or replace the safety warning instructions on the basic atmosphere adjustment instructions to generate ambient light control instructions.

[0056] In one specific embodiment, the execution of this step can be conceived as a real-time, rule-based arbitration process. Assume the in-vehicle system is currently in an ambient theme, with the corresponding basic ambient adjustment command being {the ambient lights throughout the vehicle are a soft green, with a brightness of 40%}. If the vehicle's sensors (such as a forward-facing millimeter-wave radar) detect a potential rear-end collision risk, step S203 will output a Level 1 (highest level) safety priority and a corresponding safety warning command, which might be {the ambient lights throughout the vehicle flash red at a frequency of 5 Hz}. Upon receiving the Level 1 priority, the arbitration module will immediately execute replacement logic, completely discarding the current ambient theme command and outputting the high-frequency red flashing command as the sole ambient light control command to the lighting controller, thereby achieving the strongest, non-intrusive visual warning. Consider another scenario: if the system detects a vehicle to the left rear via a blind spot monitoring sensor, the safety analysis module outputs a Level 2 (medium level) safety priority and a corresponding warning command {the ambient lights on the left A-pillar and left door panel turn yellow and flash slowly at a frequency of 1 Hz}. In this case, the arbitration module will execute dynamic overlay logic. It does not cancel the overall green atmosphere, but rather precisely overlays the warning command onto the basic command to generate a composite ambient light control command: {The ambient light on the right side of the vehicle and other non-conflicting areas remains a soft green with a brightness of 40%; at the same time, the ambient light on the left A-pillar and the left door panel switches to yellow, increases the brightness to 80%, and flashes at a frequency of 1 Hz}.

[0057] Optionally, the step of dynamically superimposing or replacing the safety warning instruction with the basic atmosphere adjustment instruction according to the safety priority to generate an ambient light control instruction includes: when the safety priority is a first preset safety level, superimposing the warning hue of the safety warning instruction with a preset transparency onto the color parameters of the basic atmosphere adjustment instruction to obtain a blended color parameter, and generating the ambient light control instruction based on the blended color parameter; when the safety priority is a second preset safety level, reducing the brightness parameter of the basic atmosphere adjustment instruction by a preset amount to determine a reference brightness parameter, combining the brightness change waveform parameter and brightness change frequency parameter in the safety warning instruction with the reference brightness parameter to generate the ambient light control instruction; when the safety priority is a third preset safety level, generating the ambient light control instruction based on the parameters of the safety warning instruction.

[0058] Specifically, when the safety priority is the first preset safety level, such as corresponding to a lower-level warning message (e.g., entering a congested area or low fuel level), the system will execute a gentle color blending strategy. Specifically, the system obtains a preset warning hue (e.g., yellow representing congestion) and a preset low transparency value (e.g., 20%) from the safety warning instruction. Then, the system blends this 20% transparent yellow with the color parameter of the current basic ambient lighting instruction (e.g., blue representing comfort) using the alpha channel. The final blended color parameter will be a slightly yellowish blue, preserving the original comfortable atmosphere while conveying a warning message to the driver through subtle hue changes. Based on this blended color, the system generates and issues ambient lighting control commands, achieving non-intrusive information notification.

[0059] Furthermore, when the safety priority is the second preset safety level, such as corresponding to a medium-level warning requiring the driver's attention (e.g., lane departure), the system will employ a warning strategy based on dynamic brightness changes. First, the system lowers the brightness parameter of the current basic ambient lighting command (e.g., 60%) by a preset margin (e.g., a 20% reduction), resulting in a baseline brightness parameter of 40%. This aims to reduce the brightness of the background light to highlight the impending warning. Next, the system extracts the brightness change waveform parameter (e.g., a sine wave, i.e., a breathing light effect) and the brightness change frequency parameter (e.g., 0.8 Hz) from the safety warning command. Combining these two dynamic parameters with the baseline brightness parameter, the resulting ambient light control command will cause the designated ambient light (e.g., the door panel light on the side of the vehicle deviating from its lane) to maintain its original color while flashing smoothly in a breathing pattern at a frequency of 0.8 Hz, centered at 40% brightness. This dynamic effect is more attractive than static color changes and effectively guides the driver's attention.

[0060] When the safety priority is set to the third preset safety level, which typically corresponds to the most urgent dangerous situation (such as an impending collision or emergency braking), the system will execute the highest-level replacement strategy. In this case, the system will completely ignore any currently executing basic ambient lighting commands. It will directly and completely adopt all parameters from the safety warning command to generate the final ambient lighting control command. For example, the safety warning command might be defined as {light color: high-saturation red, brightness: 100%, waveform: square wave strobe, frequency: 5 Hz}. The final generated command will drive the ambient lights throughout the vehicle or in key areas to execute a high-brightness, high-frequency red strobe. This uncompromising replacement operation ensures that in the most critical moments, the vehicle lighting system can provide the strongest and clearest visual warning signal, allowing the driver's attention to be fully focused on avoiding danger.

[0061] Figure 3This is another schematic flowchart of an ambient lighting control method that adapts to environmental changes, as disclosed in an embodiment of this application. S301. Obtain the planned driving route information of the vehicle and the map data corresponding to the planned driving route information; In one specific embodiment, step S301 begins with the vehicle's central infotainment system or dedicated navigation unit. For example, when the driver inputs a destination (such as the Oriental Pearl Radio & TV Tower) via a touchscreen or voice command and initiates navigation, the in-vehicle navigation module immediately calculates one or more recommended routes. The planned driving route information in this step can be understood as the optimal route ultimately determined and output by the navigation module. Its data form can be a series of ordered geographic coordinates (latitude and longitude sequence) or a series of connected road identifier sequences. The methods for obtaining this route information can include: firstly, the ambient lighting control system acts as a subscriber, listening in real-time to and receiving route data packets broadcast by the navigation module via the vehicle's internal communication bus (such as CAN bus or in-vehicle Ethernet); secondly, the ambient lighting control system actively requests the navigation service through its internal software interface to obtain the current active route data. Simultaneously, to obtain map data corresponding to the planned driving route information, the system initiates a bounded query request based on the acquired route coordinates or road IDs to its internal offline map database or an online map service provider (such as the cloud services of Gaode Maps or Baidu Maps) connected via an in-vehicle communication module (such as a 4G / 5GT-Box). This query does not download a map of the entire area, but rather precisely extracts specific map layers and attribute data within a preset range along the route and its surroundings (e.g., 500 meters on each side of the route). This data may include, but is not limited to: road type (e.g., highways, coastal roads, mountain roads), road geometry (e.g., curvature, slope), and the location and type of key infrastructure or points of interest (e.g., tunnels, bridges, toll stations, famous landmarks, tourist area boundaries), etc. The system then associates and caches the acquired structured route information and map data containing rich semantic tags.

[0062] S302. Based on the planned driving route information and the map data, identify the preset high-risk driving scenarios on the vehicle's driving route; In one specific embodiment, the core of step S302 is an automated scene matching and labeling process that relies on a predefined high-risk driving scenario library. This library stores definition rules for various high-risk scenarios, each rule associating a scenario with a specific set of map data features. After obtaining the complete planned path, the system traverses each segment of the path (e.g., using 100 meters as an analysis unit) and extracts the corresponding map data. Subsequently, the extracted map data is matched against the rules in the scenario library. For example, a "continuous sharp bends mountain road" rule in the scenario library might be defined as: on a mountain road segment, there are three or more consecutive bends with a curvature greater than a preset curvature threshold (e.g., 0.002m⁻¹), and the straight-line distance between any two bends is less than a preset length threshold (e.g., 200 meters). When the system analyzes that the map data of a certain segment fully meets this rule, it labels that segment and its start and end coordinates as a continuous sharp bends mountain road scenario. Similarly, other pre-defined high-risk scenarios, such as long downhill sections (e.g., sections with a gradient of less than -5% and a continuous length exceeding 2 kilometers), tunnel clusters (e.g., passing through more than two tunnels consecutively within 2 kilometers), and accident-prone areas (points directly marked in map data), are also identified through this rule-matching method. An alternative approach is to use a trained machine learning model (e.g., decision trees or support vector machines) to combine various map features of each road segment (such as curvature, gradient, road grade, historical weather data, etc.) into a feature vector, inputting this vector into the model for classification. The model then directly outputs whether the road segment belongs to a high-risk scenario and its specific type.

[0063] S303. When the distance between the vehicle and the high-risk driving scenario is detected to be less than the preset trigger distance, the safety priority is combined with the preset risk enhancement level to obtain the target safety priority. In this embodiment, the system maintains an ambient lighting strategy library, which can be viewed as a mapping table or database. It maps identifiers for various preset high-risk driving scenarios to a set of specific, carefully designed scenario warning ambient lighting instructions. For example, when the system determines that the vehicle is about to enter a scenario identified as a series of sharp mountain bends in 500 meters, it triggers a time or distance checker. This checker continuously calculates the estimated arrival time based on the vehicle's current speed. Ten seconds before the estimated arrival time (i.e., the preset time), or 300 meters from the starting point (i.e., the preset distance), the system queries the strategy library for the continuous sharp mountain bend scenario and invokes its corresponding warning instruction. This instruction might be specifically defined as: switching the ambient light color of the door panel on the turning side (e.g., the left side when turning left) to a dynamically flowing yellow, with the flow direction consistent with the turning direction, while simultaneously reducing the brightness of the dashboard and passenger-side ambient lights to 30%. For example, when the system detects that the tunnel entrance is 1 kilometer ahead, it will call the corresponding instruction 5 seconds before entering the tunnel: smoothly transition the color temperature of all ambient lights in the car to 4000K neutral white light, and linearly increase the brightness from the current value to 70%, so as to adapt to the changes in light inside the tunnel in advance and reduce visual impact.

[0064] S304. Based on the target safety priority, dynamically superimpose or replace the safety warning instruction on the basic atmosphere adjustment instruction to generate the ambient light control instruction.

[0065] Specifically, the system packages the scene warning ambient light command into a data frame conforming to the vehicle lighting controller's communication protocol and sends it out via the vehicle bus (such as CAN or vehicle Ethernet). It's worth noting that while this scene warning ambient light command has a high execution priority, its integration with the vehicle's basic ambient lighting can be flexible. One implementation is that the command completely overrides (replaces) the current regular ambient light command to ensure clear communication of the warning information, such as forcibly switching to a warning color when entering an accident-prone area. Another implementation is that the command is overlaid with or partially modified from the basic ambient lighting; for example, when passing city landmarks, only the roof ambient light projects a specific pattern or color, while the ambient lighting in other areas of the vehicle remains unchanged, achieving a coexistence of scene-based and personalized ambient lighting.

[0066] This embodiment also discloses an electronic device, as shown in the reference. Figure 4 The electronic device may include: at least one processor 401, at least one communication bus 402, user interface 403, network interface 404, and at least one memory 405.

[0067] The communication bus 402 is used to enable communication between these components.

[0068] The user interface 403 may include a display screen and a camera. Optionally, the user interface 403 may also include a standard wired interface and a wireless interface.

[0069] The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0070] The processor 401 may include one or more processing cores. The processor 401 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 405, and by calling data stored in memory 405. Optionally, the processor 401 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 401 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 401.

[0071] The memory 405 may include random access memory (RAM) or read-only memory. Optionally, the memory 405 may include a non-transitory computer-readable storage medium. The memory 405 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 405 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 405 may also be at least one storage device located remotely from the aforementioned processor 401. Figure 4As shown, the memory 405, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for controlling ambient lighting in accordance with environmental changes.

[0072] exist Figure 4 In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 401 can be used to call an application program stored in the memory 405 that controls an ambient light according to the environment. When executed by one or more processors 401, the electronic device executes one or more methods as described in the above embodiments.

[0073] In some embodiments of this application, a computer-readable storage medium is provided, including instructions that, when executed on the electronic device, cause the electronic device to perform an ambient lighting control method that adapts to environmental changes, as described in this application.

[0074] In some embodiments of this application, a computer program product is also provided, which, when run on an electronic device, causes the electronic device to execute an ambient lighting control method that changes with the environment as described in the embodiments of this application.

[0075] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0076] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and 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 shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0077] 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.

[0078] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 405 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 in the various embodiments of this application. The aforementioned memory 405 includes various media capable of storing program code, such as a USB flash drive, portable hard drive, magnetic disk, or optical disk.

[0079] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will be readily apparent to those skilled in the art upon consideration of the disclosure in this specification. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for controlling ambient lighting that changes with the environment, characterized in that, Applied to a server, the method includes: Acquire driver's physiological state data and vehicle driving state data, and combine the physiological state data and driving state data to obtain a safety status assessment; Acquire image data of the external environment and emotional state data of the occupants inside the vehicle, and obtain an environmental experience assessment based on the image data and the emotional state data; Based on the security status assessment, security priorities and corresponding security warning instructions are determined through a preset security analysis model. Based on the environmental experience assessment, a preset atmosphere matching process is performed to generate basic atmosphere adjustment instructions; Based on the safety priority, the basic atmosphere adjustment command is dynamically superimposed or replaced with the safety warning command to generate an ambient light control command.

2. The method according to claim 1, characterized in that, The process of acquiring driver physiological state data and vehicle driving state data, and combining the physiological state data and driving state data to obtain a safety status assessment includes: The driver's first facial image sequence is acquired, and the eyelid opening and closing frequency and gaze direction vector are extracted from the first facial image sequence to form the physiological state data; The vehicle's steering wheel angular rate, brake pedal pressure value, and vehicle acceleration are acquired to form the vehicle's driving state data. Within a preset time window, the physiological state data and the driving state data are vectorized and combined to obtain a multidimensional state vector sequence, and the trajectory morphology features of the multidimensional state vector sequence are calculated. The safety status assessment is obtained by concatenating the trajectory morphology features with the corresponding physiological state data and driving state data within the preset time window.

3. The method according to claim 2, characterized in that, The acquisition of image data of the external environment and emotional state data of the occupants, and the environmental experience assessment based on the image data and the emotional state data, includes: Acquire images of the exterior scene within a preset range of the vehicle, perform semantic segmentation on the exterior scene images, and extract preset environment category features corresponding to the exterior scene images to form the image data; The vehicle is used to detect and identify all occupants, and voice signals and second facial image sequences of occupants other than the driver are acquired. For each person in the vehicle, facial expression features are extracted from the first facial image sequence or the second facial image sequence, and corresponding acoustic features are extracted from the speech signal. A corresponding emotion vector is generated through a preset multimodal emotion recognition model. All the emotional vectors are aggregated to obtain the emotional state data, which represents the overall emotional state of the people in the vehicle. The image data is integrated with the emotional state data to generate the environmental experience assessment.

4. The method according to claim 3, characterized in that, The process of aggregating all the emotion vectors to obtain the emotion state data includes: Separate the driver's emotional vector corresponding to the driver and the passenger's emotional vector set corresponding to the other occupants from all the emotional vectors; Based on the driver's emotional polarity score of the driver's emotional vector and the average emotional polarity score of the passenger's emotional vector set, it is determined whether an emotional conflict state exists; When the emotional conflict state exists, the driver's emotional vector is used as the emotional state data; When the emotional conflict state does not exist, the driver's emotional vector and the passenger's emotional vector set are weighted and fused with preset weights to obtain a fused emotional vector, and the fused emotional vector is used as the emotional state data.

5. The method according to claim 4, characterized in that, The step of determining whether an emotional conflict exists based on the driver's emotional polarity score of the driver's emotional vector and the average emotional polarity score of the passenger emotional vector set includes: The driver's emotion vector is multiplied by a preset first polarity weight vector to obtain the driver's emotion polarity score. The average passenger emotion vector is obtained by averaging all passenger emotion vectors in the set of passenger emotion vectors. The average passenger emotion vector is then multiplied by the inner product of the average passenger emotion vector and the preset second polarity weight vector to obtain the average emotion polarity score. When the product of the driver's emotional polarity score and the average emotional polarity score is less than a preset negative conflict threshold, and the absolute values ​​of both the driver's emotional polarity score and the average emotional polarity score are greater than a preset emotional intensity threshold, the emotional conflict state is determined to exist.

6. The method according to claim 1, characterized in that, The step of dynamically superimposing or replacing the safety warning instructions on the basic atmosphere adjustment instructions according to the safety priority to generate ambient light control instructions includes: When the security priority is the first preset security level, the warning color of the security warning instruction is superimposed on the color parameter of the basic atmosphere adjustment instruction with a preset transparency to obtain the blended color parameter, and the ambient light control instruction is generated based on the blended color parameter. When the safety priority is the second preset safety level, the brightness parameter of the basic atmosphere adjustment command is reduced by a preset amount to determine the reference brightness parameter. The brightness change waveform parameter and brightness change frequency parameter in the safety warning command are combined with the reference brightness parameter to generate the ambient light control command. When the security priority is the third preset security level, the ambient light control command is generated based on the parameters of the security warning command.

7. The method according to claim 1, characterized in that, The method further includes: Obtain the planned driving route information of the vehicle and the map data corresponding to the planned driving route information; Based on the planned driving route information and the map data, preset high-risk driving scenarios on the vehicle's driving route are identified; When the distance between the vehicle and the high-risk driving scenario is detected to be less than the preset trigger distance, the safety priority is combined with the preset risk escalation level to obtain the target safety priority; Based on the target safety priority, the basic atmosphere adjustment command is dynamically superimposed or replaced with the safety warning command to generate the ambient light control command.

8. An electronic device, characterized in that, The device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. The user interface and the network interface are both used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product is run on an electronic device, it causes the electronic device to perform the method as described in any one of claims 1-7.

Images